diff --git a/backend/__init__.py b/backend/__init__.py
index f22815b..9d48db4 100644
--- a/backend/__init__.py
+++ b/backend/__init__.py
@@ -1,18 +1 @@
 from __future__ import annotations
-
-import numpy as np
-
-
-def _apply_numpy_compat_aliases() -> None:
-    """Restore removed NumPy scalar aliases still used by some dependencies."""
-    aliases = {
-        "complex": complex,
-        "float": float,
-        "int": int,
-    }
-    for name, value in aliases.items():
-        if not hasattr(np, name):
-            setattr(np, name, value)
-
-
-_apply_numpy_compat_aliases()
diff --git a/backend/data_types.py b/backend/data_types.py
index 2754a4f..38a3ddb 100644
--- a/backend/data_types.py
+++ b/backend/data_types.py
@@ -67,6 +67,43 @@ class LineData:
         return self.data[item]
 
 
+@dataclass
+class MeshModel:
+    vertices: np.ndarray
+    faces: np.ndarray
+    colors: np.ndarray | None = None
+
+    def __post_init__(self) -> None:
+        self.vertices = np.asarray(self.vertices, dtype=np.float32).reshape(-1, 3)
+        self.faces = np.asarray(self.faces, dtype=np.int32).reshape(-1, 3)
+        if self.colors is not None:
+            self.colors = np.asarray(self.colors, dtype=np.uint8).reshape(-1, 3)
+            if len(self.colors) != len(self.vertices):
+                raise ValueError("MeshModel.colors must have one RGB triplet per vertex.")
+
+
+class ImageData(np.ndarray):
+    def __new__(cls, data: Any, metadata: dict[str, Any] | None = None):
+        obj = np.asarray(data).view(cls)
+        obj.metadata = deepcopy(metadata) if isinstance(metadata, dict) else {}
+        return obj
+
+    def __array_finalize__(self, obj):
+        self.metadata = deepcopy(getattr(obj, "metadata", {})) if obj is not None else {}
+
+    def copy_with_metadata(self, *, data: Any | None = None, metadata: dict[str, Any] | None = None) -> "ImageData":
+        base = np.asarray(self if data is None else data)
+        merged = deepcopy(self.metadata)
+        if isinstance(metadata, dict):
+            merged.update(deepcopy(metadata))
+        return ImageData(base, metadata=merged)
+
+
+def image_metadata(image: Any) -> dict[str, Any]:
+    metadata = getattr(image, "metadata", None)
+    return deepcopy(metadata) if isinstance(metadata, dict) else {}
+
+
 def _normalize_hex_color(color: Any, default: str = "#000000") -> str:
     if isinstance(color, str):
         text = color.strip()
@@ -638,10 +675,28 @@ def _sanitize_markup_shapes(shapes: Any) -> list[dict[str, Any]]:
     return parsed
 
 
-def _apply_annotation_overlay(
+def _annotation_context_from_field(field: DataField, colormap: Any) -> dict[str, Any]:
+    legend_min, legend_mid, legend_max = _display_value_range(field)
+    return {
+        "xreal": float(field.xreal),
+        "si_unit_xy": str(field.si_unit_xy),
+        "legend_min": float(legend_min),
+        "legend_mid": float(legend_mid),
+        "legend_max": float(legend_max),
+        "legend_unit": str(field.si_unit_z),
+        "colormap": normalize_colormap_spec(colormap, fallback=field.colormap),
+    }
+
+
+def _annotation_context_from_image(image: Any) -> dict[str, Any] | None:
+    metadata = image_metadata(image)
+    context = metadata.get("annotation_context")
+    return deepcopy(context) if isinstance(context, dict) else None
+
+
+def _apply_annotation_overlay_from_context(
     image: np.ndarray,
-    field: DataField,
-    colormap: Any,
+    context: dict[str, Any],
     spec: dict[str, Any],
 ) -> np.ndarray:
     from PIL import Image, ImageDraw
@@ -657,8 +712,7 @@ def _apply_annotation_overlay(
         current = np.repeat(current[:, :, np.newaxis], 3, axis=2)
 
     height, current_width = current.shape[:2]
-    field_width = max(1, int(field.xres))
-    legend_width = max(72, int(round(field_width * 0.18))) if show_color_map else 0
+    legend_width = max(72, int(round(current_width * 0.18))) if show_color_map else 0
     canvas_width = current_width + legend_width
     canvas = np.full((height, canvas_width, 3), 255, dtype=np.uint8)
     canvas[:, :current_width] = current
@@ -667,20 +721,40 @@ def _apply_annotation_overlay(
     draw = ImageDraw.Draw(pil_image)
     base_font_px = max(6, int(round(text_size)))
 
-    if show_scale_bar and field_width > 0 and np.isfinite(field.xreal) and field.xreal > 0:
-        target_real = field.xreal / 5.0
+    xreal_raw = context.get("xreal")
+    xreal = float(xreal_raw) if xreal_raw is not None else 0.0
+    si_unit_xy = str(context.get("si_unit_xy", "") or "")
+    legend_unit = str(context.get("legend_unit", "") or "")
+    legend_min_raw = context.get("legend_min")
+    legend_mid_raw = context.get("legend_mid")
+    legend_max_raw = context.get("legend_max")
+    legend_min = float(legend_min_raw) if legend_min_raw is not None else 0.0
+    legend_mid = float(legend_mid_raw) if legend_mid_raw is not None else 0.0
+    legend_max = float(legend_max_raw) if legend_max_raw is not None else 0.0
+    colormap = normalize_colormap_spec(context.get("colormap", "gray"), fallback="gray")
+    has_scale_bar = np.isfinite(xreal) and xreal > 0 and bool(si_unit_xy)
+    has_color_legend = (
+        np.isfinite(legend_min)
+        and np.isfinite(legend_mid)
+        and np.isfinite(legend_max)
+        and bool(legend_unit)
+    )
+
+    if show_scale_bar and has_scale_bar and current_width > 0:
+        target_real = xreal / 5.0
         bar_real = _nice_length(target_real)
-        if bar_real > 0 and np.isfinite(field.dx) and field.dx > 0:
-            bar_px = max(1, int(round(bar_real / field.dx)))
-            margin_x = max(8, field_width // 24)
+        if bar_real > 0:
+            px_per_real = current_width / xreal
+            bar_px = max(1, int(round(bar_real * px_per_real)))
+            margin_x = max(8, current_width // 24)
             margin_y = max(8, height // 24)
             bar_height = max(3, int(round(height * 0.012)))
-            bar_px = min(bar_px, max(1, field_width - 2 * margin_x))
+            bar_px = min(bar_px, max(1, current_width - 2 * margin_x))
             x0 = margin_x
             x1 = x0 + bar_px
             y1 = height - margin_y
             y0 = y1 - bar_height
-            text = _format_with_unit(bar_real, field.si_unit_xy)
+            text = _format_with_unit(bar_real, si_unit_xy)
             text_image = _render_overlay_text(text, base_font_px, (255, 255, 255), font_spec=font_spec)
             text_w, text_h = text_image.size
             label_pad = 2
@@ -692,7 +766,7 @@ def _apply_annotation_overlay(
             draw.rectangle((x0, y0, x1, y1), fill=(255, 255, 255))
             pil_image.paste(text_image, (x0, bg_top + label_pad), text_image)
 
-    if show_color_map and legend_width > 0:
+    if show_color_map and has_color_legend and legend_width > 0:
         panel_x0 = current_width
         draw.rectangle((panel_x0, 0, canvas_width, height), fill=(245, 245, 245))
         grad_x0 = panel_x0 + max(8, legend_width // 7)
@@ -706,7 +780,6 @@ def _apply_annotation_overlay(
         pil_image.paste(Image.fromarray(gradient_rgb), (grad_x0, grad_y0))
         draw.rectangle((grad_x0, grad_y0, grad_x0 + grad_w, grad_y1), outline=(40, 40, 40), width=1)
 
-        legend_min, legend_mid, legend_max = _display_value_range(field)
         labels = [
             (legend_max, grad_y0),
             (legend_mid, grad_y0 + grad_h // 2),
@@ -715,7 +788,7 @@ def _apply_annotation_overlay(
         text_x = grad_x0 + grad_w + 8
         for value, y_center in labels:
             text_image = _render_overlay_text(
-                _format_with_unit(value, field.si_unit_z),
+                _format_with_unit(value, legend_unit),
                 base_font_px,
                 (20, 20, 20),
                 font_spec=font_spec,
@@ -727,7 +800,20 @@ def _apply_annotation_overlay(
     return np.asarray(pil_image, dtype=np.uint8)
 
 
-def _apply_markup_overlay(image: np.ndarray, field: DataField, spec: dict[str, Any]) -> np.ndarray:
+def _apply_annotation_overlay(
+    image: np.ndarray,
+    field: DataField,
+    colormap: Any,
+    spec: dict[str, Any],
+) -> np.ndarray:
+    return _apply_annotation_overlay_from_context(
+        image,
+        _annotation_context_from_field(field, colormap),
+        spec,
+    )
+
+
+def _apply_markup_overlay(image: np.ndarray, field: DataField | None, spec: dict[str, Any]) -> np.ndarray:
     from PIL import Image, ImageDraw
 
     current = np.asarray(image, dtype=np.uint8)
@@ -736,8 +822,8 @@ def _apply_markup_overlay(image: np.ndarray, field: DataField, spec: dict[str, A
 
     pil_image = Image.fromarray(current.copy())
     draw = ImageDraw.Draw(pil_image)
-    field_width = max(1, int(field.xres))
-    field_height = max(1, int(field.yres))
+    field_width = max(1, int(field.xres)) if isinstance(field, DataField) else max(1, current.shape[1])
+    field_height = max(1, int(field.yres)) if isinstance(field, DataField) else max(1, current.shape[0])
 
     for shape in _sanitize_markup_shapes(spec.get("shapes")):
         x1 = float(shape["x1"]) * field_width
diff --git a/backend/execution.py b/backend/execution.py
index fcc331b..ccb4596 100644
--- a/backend/execution.py
+++ b/backend/execution.py
@@ -221,6 +221,7 @@ class ExecutionEngine:
         from backend.nodes.preview_image import PreviewImage
         from backend.nodes.print_table import PrintTable
         from backend.nodes.view_3d import View3D
+        from backend.nodes.annotations import Annotations
         from backend.nodes.value_display import ValueDisplay
         from backend.nodes.markup import Markup
         from backend.nodes.cross_section import CrossSection
@@ -234,6 +235,7 @@ class ExecutionEngine:
         from backend.nodes.mask_invert import MaskInvert
         from backend.nodes.mask_combine import MaskCombine
         from backend.nodes.draw_mask import DrawMask
+        from backend.nodes.save import Save
         from backend.nodes.save_image import SaveImage
         from backend.nodes.image import Image
         from backend.nodes.image_demo import ImageDemo
@@ -245,6 +247,7 @@ class ExecutionEngine:
         MaskCombine._broadcast_fn = on_preview
         DrawMask._broadcast_overlay_fn = on_overlay
         View3D._broadcast_mesh_fn = on_mesh
+        Annotations._broadcast_warning_fn = on_warning
         PrintTable._broadcast_table_fn = on_table
         ValueDisplay._broadcast_value_fn = on_value
         Stats._broadcast_value_fn = on_value
@@ -256,6 +259,7 @@ class ExecutionEngine:
         Markup._broadcast_overlay_fn = on_overlay
         Image._broadcast_warning_fn = on_warning
         ImageDemo._broadcast_warning_fn = on_warning
+        Save._broadcast_warning_fn = on_warning
         SaveImage._broadcast_warning_fn = on_warning
 
     def _set_node_id_on_display(self, cls: type, node_id: str) -> None:
@@ -263,6 +267,7 @@ class ExecutionEngine:
         from backend.nodes.preview_image import PreviewImage
         from backend.nodes.print_table import PrintTable
         from backend.nodes.view_3d import View3D
+        from backend.nodes.annotations import Annotations
         from backend.nodes.value_display import ValueDisplay
         from backend.nodes.markup import Markup
         from backend.nodes.cross_section import CrossSection
@@ -278,10 +283,11 @@ class ExecutionEngine:
         from backend.nodes.draw_mask import DrawMask
         from backend.nodes.image import Image
         from backend.nodes.image_demo import ImageDemo
+        from backend.nodes.save import Save
         from backend.nodes.save_image import SaveImage
-        if cls in (PreviewImage, PrintTable, View3D, ValueDisplay, Stats, Histogram, CrossSection, Cursors, CropResizeField, RotateField, Markup,
+        if cls in (PreviewImage, PrintTable, View3D, Annotations, ValueDisplay, Stats, Histogram, CrossSection, Cursors, CropResizeField, RotateField, Markup,
                    ThresholdMask, MaskMorphology, MaskInvert, MaskCombine, DrawMask,
-                   Image, ImageDemo, SaveImage):
+                   Image, ImageDemo, Save, SaveImage):
             cls._current_node_id = node_id
 
     def _auto_preview(
@@ -331,6 +337,16 @@ class ExecutionEngine:
                 on_preview(node_id, encode_preview(arr))
                 return
 
+            if type_name == "ANNOTATION_SOURCE" and on_preview:
+                if isinstance(value, DataField):
+                    arr = render_datafield_preview(value, value.colormap)
+                    on_preview(node_id, encode_preview(arr))
+                    return
+                if isinstance(value, np.ndarray):
+                    arr = image_to_uint8(value)
+                    on_preview(node_id, encode_preview(arr))
+                    return
+
             if type_name == "LINE" and isinstance(value, (np.ndarray, LineData)) and on_preview:
                 preview = self._render_line_preview(cls, slot, result)
                 if preview:
diff --git a/backend/node_menu.py b/backend/node_menu.py
index 2bc2de8..f8175b6 100644
--- a/backend/node_menu.py
+++ b/backend/node_menu.py
@@ -25,6 +25,7 @@ MENU_LAYOUT: dict[str, list[str]] = {
     ],
     "Output": [
         "PreviewImage",
+        "Save",
         "SaveImage",
         "View3D",
         "PrintTable",
diff --git a/backend/nodes/__init__.py b/backend/nodes/__init__.py
index 1b88dcf..234cf27 100644
--- a/backend/nodes/__init__.py
+++ b/backend/nodes/__init__.py
@@ -8,6 +8,7 @@ from backend.nodes import (
     coordinate_pair,
     number,
     range_slider,
+    save,
     save_image,
     # Filters
     gaussian_filter,
diff --git a/backend/nodes/annotations.py b/backend/nodes/annotations.py
index 4d615c4..8e74c3e 100644
--- a/backend/nodes/annotations.py
+++ b/backend/nodes/annotations.py
@@ -1,16 +1,28 @@
 from __future__ import annotations
 import numpy as np
 from backend.node_registry import register_node
-from backend.data_types import COLORMAPS, DataField, normalize_font_spec, resolve_colormap_input
+from backend.data_types import (
+    COLORMAPS,
+    DataField,
+    ImageData,
+    _apply_annotation_overlay_from_context,
+    _annotation_context_from_image,
+    image_to_uint8,
+    normalize_font_spec,
+    resolve_colormap_input,
+)
 
 
 @register_node(display_name="Annotations")
 class Annotations:
+    _broadcast_warning_fn = None
+    _current_node_id: str = ""
+
     @classmethod
     def INPUT_TYPES(cls):
         return {
             "required": {
-                "field": ("DATA_FIELD",),
+                "input": ("ANNOTATION_SOURCE", {"label": "Input"}),
                 "colormap": (["auto"] + list(COLORMAPS), {"hide_when_input_connected": "colormap_map"}),
                 "show_scale_bar": ("BOOLEAN", {"default": True}),
                 "show_color_map": ("BOOLEAN", {"default": True}),
@@ -27,18 +39,18 @@ class Annotations:
             },
         }
 
-    RETURN_TYPES = ("DATA_FIELD",)
-    RETURN_NAMES = ("annotated",)
+    RETURN_TYPES = ("ANNOTATION_SOURCE",)
+    RETURN_NAMES = ("Output",)
     FUNCTION = "render"
 
     DESCRIPTION = (
-        "Attach optional publication-style annotations to a DATA_FIELD without flattening the raw data. "
-        "The preview shows a scale bar and/or side colour legend, while downstream field operations keep the underlying AFM values."
+        "Attach optional publication-style annotations to a DATA_FIELD without flattening the raw data, "
+        "or annotate an IMAGE that carries viewport metadata from View3D."
     )
 
     def render(
         self,
-        field: DataField,
+        input,
         colormap: str,
         show_scale_bar: bool,
         show_color_map: bool,
@@ -46,24 +58,69 @@ class Annotations:
         colormap_map=None,
         font=None,
     ) -> tuple:
+        annotation_spec = {
+            "kind": "annotation",
+            "show_scale_bar": bool(show_scale_bar),
+            "show_color_map": bool(show_color_map),
+            "text_size": float(np.clip(text_size, 6.0, 96.0)) if np.isfinite(text_size) else 14.0,
+            "font": normalize_font_spec(font),
+        }
+
+        if isinstance(input, DataField):
+            resolved_colormap = resolve_colormap_input(
+                colormap,
+                colormap_input=colormap_map,
+                inherited=input.colormap,
+                default="gray",
+            )
+            out = input.replace(
+                colormap=resolved_colormap,
+                overlays=[
+                    *input.overlays,
+                    annotation_spec,
+                ],
+            )
+            return (out,)
+
+        context = _annotation_context_from_image(input)
+        if context is None:
+            self._send_warning(
+                "Annotations image input has no scale metadata, so scale bar and color-map legend cannot be added."
+            )
+            return (ImageData(image_to_uint8(input)),)
+
         resolved_colormap = resolve_colormap_input(
             colormap,
             colormap_input=colormap_map,
-            inherited=field.colormap,
+            inherited=context.get("colormap"),
             default="gray",
         )
-        text_size = float(np.clip(text_size, 6.0, 96.0)) if np.isfinite(text_size) else 14.0
-        out = field.replace(
-            colormap=resolved_colormap,
-            overlays=[
-                *field.overlays,
-                {
-                    "kind": "annotation",
-                    "show_scale_bar": bool(show_scale_bar),
-                    "show_color_map": bool(show_color_map),
-                    "text_size": text_size,
-                    "font": normalize_font_spec(font),
-                },
-            ],
+        context["colormap"] = resolved_colormap
+        missing_features = []
+        xreal = context.get("xreal")
+        if bool(show_scale_bar) and not (isinstance(xreal, (int, float)) and np.isfinite(float(xreal)) and float(xreal) > 0 and str(context.get("si_unit_xy", "")).strip()):
+            missing_features.append("scale bar")
+        if bool(show_color_map):
+            legend_values = (context.get("legend_min"), context.get("legend_mid"), context.get("legend_max"))
+            has_legend_values = all(
+                isinstance(value, (int, float)) and np.isfinite(float(value))
+                for value in legend_values
+            )
+            if not (has_legend_values and str(context.get("legend_unit", "")).strip()):
+                missing_features.append("color-map legend")
+        if missing_features:
+            self._send_warning(
+                f"Annotations image input is missing metadata for: {', '.join(missing_features)}."
+            )
+        annotated = _apply_annotation_overlay_from_context(
+            image_to_uint8(input),
+            context,
+            annotation_spec,
         )
-        return (out,)
+        return (ImageData(annotated, metadata={"annotation_context": context}),)
+
+    def _send_warning(self, message: str):
+        fn = Annotations._broadcast_warning_fn
+        nid = Annotations._current_node_id
+        if fn and nid:
+            fn(nid, message)
diff --git a/backend/nodes/image.py b/backend/nodes/image.py
index a1d7b95..4cf8e18 100644
--- a/backend/nodes/image.py
+++ b/backend/nodes/image.py
@@ -1,4 +1,5 @@
 from __future__ import annotations
+from functools import lru_cache
 import numpy as np
 from pathlib import Path
 
@@ -48,17 +49,21 @@ class Image:
 
         ext = path_obj.suffix.lower()
         resolved_colormap = resolve_colormap_input(colormap, colormap_input=colormap_map, default="viridis")
+        stat = path_obj.stat()
+        cached_fields = Image._load_fields_cached(
+            str(path_obj.resolve()),
+            int(stat.st_mtime_ns),
+            int(stat.st_size),
+        )
+        fields = tuple(field.copy() for field in cached_fields)
 
-        if ext in _SPM_EXTENSIONS:
-            fields = self._load_spm_all(path_obj, ext)
-            for f in fields:
-                f.colormap = resolved_colormap
-            return tuple(fields)
+        for field in fields:
+            field.colormap = resolved_colormap
 
-        field = self._load_image_or_array(path_obj, ext)
-        field.colormap = resolved_colormap
-        self._send_warning("Uncalibrated data — no physical dimensions.")
-        return (field,)
+        if ext not in _SPM_EXTENSIONS:
+            self._send_warning("Uncalibrated data — no physical dimensions.")
+
+        return fields
 
     def _send_warning(self, message: str):
         fn = Image._broadcast_warning_fn
@@ -66,17 +71,28 @@ class Image:
         if fn and nid:
             fn(nid, message)
 
-    def _load_spm_all(self, path: Path, ext: str) -> list[DataField]:
+    @staticmethod
+    @lru_cache(maxsize=32)
+    def _load_fields_cached(path_str: str, mtime_ns: int, size_bytes: int) -> tuple[DataField, ...]:
+        path = Path(path_str)
+        ext = path.suffix.lower()
+        if ext in _SPM_EXTENSIONS:
+            return tuple(Image._load_spm_all(path, ext))
+        return (Image._load_image_or_array(path, ext),)
+
+    @staticmethod
+    def _load_spm_all(path: Path, ext: str) -> list[DataField]:
         if ext == ".gwy":
-            return self._load_gwy_all(path)
+            return Image._load_gwy_all(path)
         elif ext == ".sxm":
-            return self._load_sxm_all(path)
+            return Image._load_sxm_all(path)
         elif ext == ".ibw":
-            return self._load_ibw_all(path)
+            return Image._load_ibw_all(path)
         else:
             raise ValueError(f"Unsupported SPM format: {ext}")
 
-    def _load_gwy_all(self, path: Path) -> list[DataField]:
+    @staticmethod
+    def _load_gwy_all(path: Path) -> list[DataField]:
         try:
             import gwyfile
         except ImportError:
@@ -101,7 +117,8 @@ class Image:
             ))
         return fields
 
-    def _load_sxm_all(self, path: Path) -> list[DataField]:
+    @staticmethod
+    def _load_sxm_all(path: Path) -> list[DataField]:
         try:
             import nanonispy as nap
         except ImportError:
@@ -130,7 +147,8 @@ class Image:
             ))
         return fields
 
-    def _load_ibw_all(self, path: Path) -> list[DataField]:
+    @staticmethod
+    def _load_ibw_all(path: Path) -> list[DataField]:
         try:
             from igor.binarywave import load as load_ibw
         except ImportError:
@@ -193,7 +211,8 @@ class Image:
 
         return fields
 
-    def _load_image_or_array(self, path: Path, ext: str) -> DataField:
+    @staticmethod
+    def _load_image_or_array(path: Path, ext: str) -> DataField:
         if ext == ".npy":
             arr = np.load(str(path)).astype(np.float64)
         elif ext == ".npz":
diff --git a/backend/nodes/markup.py b/backend/nodes/markup.py
index c9e6785..3190946 100644
--- a/backend/nodes/markup.py
+++ b/backend/nodes/markup.py
@@ -1,6 +1,14 @@
 from __future__ import annotations
 from backend.node_registry import register_node
-from backend.data_types import DataField, datafield_to_uint8, encode_preview
+from backend.data_types import (
+    DataField,
+    ImageData,
+    _apply_markup_overlay,
+    encode_preview,
+    image_metadata,
+    image_to_uint8,
+    render_datafield_preview,
+)
 from backend.nodes.helpers import _parse_markup_shapes, _normalize_markup_color
 
 
@@ -12,7 +20,7 @@ class Markup:
     def INPUT_TYPES(cls):
         return {
             "required": {
-                "field": ("DATA_FIELD",),
+                "input": ("ANNOTATION_SOURCE", {"label": "Input"}),
                 "shape": (["line", "rectangle", "circle", "arrow"], {"default": "line"}),
                 "stroke_color": ("STRING", {"default": "#ffd54f", "color_picker": True}),
                 "stroke_width": ("INT", {"default": 3, "min": 1, "max": 64, "step": 1}),
@@ -21,13 +29,13 @@ class Markup:
             }
         }
 
-    RETURN_TYPES = ("DATA_FIELD",)
-    RETURN_NAMES = ("annotated",)
+    RETURN_TYPES = ("ANNOTATION_SOURCE",)
+    RETURN_NAMES = ("Output",)
     FUNCTION = "process"
 
     DESCRIPTION = (
-        "Draw simple vector markup over a DATA_FIELD without flattening the underlying data. "
-        "Choose a shape mode, colour, and stroke width, then drag directly on the preview to place lines, rectangles, circles, or arrows."
+        "Draw simple vector markup over a DATA_FIELD without flattening the underlying data, "
+        "or rasterize markup directly onto an IMAGE."
     )
 
     _broadcast_overlay_fn = None
@@ -35,22 +43,32 @@ class Markup:
 
     def process(
         self,
-        field: DataField,
+        input,
         shape: str,
         stroke_color: str,
         stroke_width: int,
         markup_shapes: str,
     ) -> tuple:
         shapes = _parse_markup_shapes(markup_shapes)
-        out = field.replace(
-            overlays=[
-                *field.overlays,
-                {
-                    "kind": "markup",
-                    "shapes": shapes,
-                },
-            ],
-        )
+        markup_spec = {
+            "kind": "markup",
+            "shapes": shapes,
+        }
+
+        if isinstance(input, DataField):
+            out = input.replace(
+                overlays=[
+                    *input.overlays,
+                    markup_spec,
+                ],
+            )
+            preview_base = render_datafield_preview(input, input.colormap)
+        else:
+            preview_base = image_to_uint8(input)
+            out = ImageData(
+                _apply_markup_overlay(preview_base, None, markup_spec),
+                metadata=image_metadata(input),
+            )
 
         if Markup._broadcast_overlay_fn is not None:
             Markup._broadcast_overlay_fn(
@@ -58,7 +76,7 @@ class Markup:
                 {
                     "kind": "markup",
                     "section_title": "Markup",
-                    "image": encode_preview(datafield_to_uint8(field, field.colormap)),
+                    "image": encode_preview(preview_base),
                     "shape": str(shape),
                     "stroke_color": _normalize_markup_color(stroke_color),
                     "stroke_width": max(1, int(stroke_width)),
diff --git a/backend/nodes/save.py b/backend/nodes/save.py
new file mode 100644
index 0000000..95ef18b
--- /dev/null
+++ b/backend/nodes/save.py
@@ -0,0 +1,260 @@
+from __future__ import annotations
+
+import csv
+import json
+from pathlib import Path
+
+import numpy as np
+
+from backend.node_registry import register_node
+from backend.data_types import DataField, LineData, MeshModel, datafield_to_uint8, image_to_uint8
+
+
+@register_node(display_name="Save")
+class Save:
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "filename": ("STRING", {
+                    "default": "",
+                    "placeholder": "filename",
+                    "placement": "top",
+                }),
+                "directory_path": ("FOLDER_PICKER", {
+                    "default": "",
+                    "label": "directory",
+                    "placement": "top",
+                    "hide_when_input_connected": "directory",
+                    "top_socket_input": "directory",
+                }),
+                "value": ("SAVE_VALUE", {"label": "value"}),
+                "format": ("STRING", {
+                    "default": "TIFF",
+                    "choices_by_source_type": {
+                        "DATA_FIELD": ["TIFF", "PNG", "NPZ"],
+                        "IMAGE": ["PNG", "TIFF", "NPZ"],
+                        "LINE": ["CSV", "NPZ", "JSON"],
+                        "MEASURE_TABLE": ["CSV", "JSON"],
+                        "RECORD_TABLE": ["CSV", "JSON"],
+                        "FLOAT": ["TXT", "JSON"],
+                        "MESH_MODEL": ["OBJ", "STL"],
+                    },
+                    "source_type_input": "value",
+                }),
+            },
+            "optional": {
+                "directory": ("DIRECTORY", {"label": "directory"}),
+            },
+        }
+
+    RETURN_TYPES = ()
+    FUNCTION = "save"
+
+    OUTPUT_NODE = True
+    MANUAL_TRIGGER = True
+    DESCRIPTION = (
+        "Save a single graph value to disk. Supports fields, images, lines, tables, scalars, and 3D meshes."
+    )
+
+    _broadcast_warning_fn = None
+    _current_node_id = None
+
+    def save(
+        self,
+        filename: str,
+        directory_path: str,
+        format: str,
+        value,
+        directory: str | None = None,
+    ):
+        path = self._resolve_save_path(filename, format, directory, directory_path)
+
+        if isinstance(value, MeshModel):
+            self._save_mesh(path, value, format)
+        elif isinstance(value, DataField):
+            self._save_datafield(path, value, format)
+        elif isinstance(value, np.ndarray):
+            if value.ndim == 1:
+                self._save_line(path, LineData(data=value), format)
+            else:
+                self._save_image_or_array(path, value, format)
+        elif isinstance(value, LineData):
+            self._save_line(path, value, format)
+        elif isinstance(value, list):
+            self._save_table(path, value, format)
+        elif isinstance(value, (int, float, np.floating, np.integer)):
+            self._save_scalar(path, float(value), format)
+        else:
+            raise ValueError(f"Save does not support input type: {type(value).__name__}")
+
+        self._send_warning(f"Saved to {path.name}")
+        return ()
+
+    def _resolve_save_path(
+        self,
+        filename: str,
+        format_name: str,
+        directory: str | None,
+        directory_path: str = "",
+    ) -> Path:
+        ext_map = {
+            "PNG": ".png",
+            "TIFF": ".tiff",
+            "NPZ": ".npz",
+            "CSV": ".csv",
+            "JSON": ".json",
+            "OBJ": ".obj",
+            "STL": ".stl",
+            "TXT": ".txt",
+        }
+        ext = ext_map[format_name]
+
+        raw_filename = str(filename).strip() if filename is not None else ""
+        raw_directory = str(directory).strip() if directory is not None else ""
+        if not raw_directory:
+            raw_directory = str(directory_path).strip() if directory_path is not None else ""
+
+        if not raw_filename:
+            raise ValueError("No output filename selected — enter a file name.")
+
+        if raw_directory:
+            dir_path = Path(raw_directory).expanduser()
+            if dir_path.exists() and not dir_path.is_dir():
+                raise ValueError("Directory input expects a folder path, not a file path.")
+            if not dir_path.exists():
+                if dir_path.suffix:
+                    raise ValueError("Directory input expects a folder path, not a file path.")
+                dir_path.mkdir(parents=True, exist_ok=True)
+            path = dir_path / Path(raw_filename).name
+        else:
+            path = Path(raw_filename).expanduser()
+            path.parent.mkdir(parents=True, exist_ok=True)
+
+        if path.suffix.lower() != ext:
+            path = path.with_suffix(ext)
+        return path
+
+    def _save_datafield(self, path: Path, field: DataField, format_name: str):
+        if format_name == "TIFF":
+            import tifffile
+            tifffile.imwrite(str(path), np.asarray(field.data, dtype=np.float32))
+            return
+        if format_name == "NPZ":
+            np.savez(str(path), field=np.asarray(field.data))
+            return
+        if format_name == "PNG":
+            from PIL import Image
+            Image.fromarray(datafield_to_uint8(field, field.colormap)).save(str(path))
+            return
+        raise ValueError(f"Format {format_name} is not supported for DATA_FIELD.")
+
+    def _save_image_or_array(self, path: Path, image: np.ndarray, format_name: str):
+        arr = np.asarray(image)
+        if format_name == "PNG":
+            from PIL import Image
+            Image.fromarray(image_to_uint8(arr)).save(str(path))
+            return
+        if format_name == "TIFF":
+            import tifffile
+            tifffile.imwrite(str(path), image_to_uint8(arr))
+            return
+        if format_name == "NPZ":
+            np.savez(str(path), image=arr)
+            return
+        raise ValueError(f"Format {format_name} is not supported for IMAGE.")
+
+    def _save_line(self, path: Path, line: LineData, format_name: str):
+        y = np.asarray(line.data, dtype=np.float64).ravel()
+        x = np.asarray(line.x_axis, dtype=np.float64).ravel()[: len(y)] if line.x_axis is not None else np.arange(len(y), dtype=np.float64)
+        if format_name == "CSV":
+            with path.open("w", newline="", encoding="utf-8") as fh:
+                writer = csv.writer(fh)
+                writer.writerow(["x", "y", "x_unit", "y_unit"])
+                for xv, yv in zip(x, y):
+                    writer.writerow([xv, yv, line.x_unit, line.y_unit])
+            return
+        if format_name == "NPZ":
+            np.savez(str(path), x=x, y=y)
+            return
+        if format_name == "JSON":
+            path.write_text(json.dumps({
+                "x": x.tolist(),
+                "y": y.tolist(),
+                "x_unit": line.x_unit,
+                "y_unit": line.y_unit,
+            }, indent=2), encoding="utf-8")
+            return
+        raise ValueError(f"Format {format_name} is not supported for LINE.")
+
+    def _save_table(self, path: Path, rows: list, format_name: str):
+        if format_name == "JSON":
+            path.write_text(json.dumps(rows, indent=2), encoding="utf-8")
+            return
+        if format_name == "CSV":
+            columns: list[str] = []
+            for row in rows:
+                if isinstance(row, dict):
+                    for key in row.keys():
+                        if key not in columns:
+                            columns.append(str(key))
+            with path.open("w", newline="", encoding="utf-8") as fh:
+                writer = csv.DictWriter(fh, fieldnames=columns)
+                writer.writeheader()
+                for row in rows:
+                    writer.writerow(row if isinstance(row, dict) else {"value": row})
+            return
+        raise ValueError(f"Format {format_name} is not supported for table inputs.")
+
+    def _save_scalar(self, path: Path, value: float, format_name: str):
+        if format_name == "TXT":
+            path.write_text(f"{value}\n", encoding="utf-8")
+            return
+        if format_name == "JSON":
+            path.write_text(json.dumps({"value": value}, indent=2), encoding="utf-8")
+            return
+        raise ValueError(f"Format {format_name} is not supported for scalar values.")
+
+    def _save_mesh(self, path: Path, mesh: MeshModel, format_name: str):
+        if format_name == "OBJ":
+            self._save_obj(path, mesh)
+            return
+        if format_name == "STL":
+            self._save_stl(path, mesh)
+            return
+        raise ValueError(f"Format {format_name} is not supported for MESH_MODEL.")
+
+    def _save_obj(self, path: Path, mesh: MeshModel):
+        lines = []
+        for vertex in mesh.vertices:
+            lines.append(f"v {vertex[0]} {vertex[1]} {vertex[2]}")
+        for face in mesh.faces:
+            lines.append(f"f {int(face[0]) + 1} {int(face[1]) + 1} {int(face[2]) + 1}")
+        path.write_text("\n".join(lines) + "\n", encoding="utf-8")
+
+    def _save_stl(self, path: Path, mesh: MeshModel):
+        def normal(a, b, c):
+            n = np.cross(b - a, c - a)
+            length = float(np.linalg.norm(n))
+            return n / length if length > 0 else np.array([0.0, 1.0, 0.0], dtype=np.float32)
+
+        lines = ["solid argonode"]
+        vertices = np.asarray(mesh.vertices, dtype=np.float32)
+        for face in np.asarray(mesh.faces, dtype=np.int32):
+            a, b, c = vertices[int(face[0])], vertices[int(face[1])], vertices[int(face[2])]
+            n = normal(a, b, c)
+            lines.append(f"  facet normal {n[0]} {n[1]} {n[2]}")
+            lines.append("    outer loop")
+            lines.append(f"      vertex {a[0]} {a[1]} {a[2]}")
+            lines.append(f"      vertex {b[0]} {b[1]} {b[2]}")
+            lines.append(f"      vertex {c[0]} {c[1]} {c[2]}")
+            lines.append("    endloop")
+            lines.append("  endfacet")
+        lines.append("endsolid argonode")
+        path.write_text("\n".join(lines) + "\n", encoding="utf-8")
+
+    def _send_warning(self, message: str):
+        fn = Save._broadcast_warning_fn
+        nid = Save._current_node_id
+        if fn and nid:
+            fn(nid, message)
diff --git a/backend/nodes/save_image.py b/backend/nodes/save_image.py
index 887d602..f122c5e 100644
--- a/backend/nodes/save_image.py
+++ b/backend/nodes/save_image.py
@@ -49,11 +49,11 @@ class SaveImage:
     OUTPUT_NODE = True
     MANUAL_TRIGGER = True
     DESCRIPTION = (
-        "Save one or more layers to a single file. "
+        "Save one or more image/field layers to a single file. "
         "Each layer input accepts either a DATA_FIELD or an IMAGE, including annotated images. "
         "Optionally drive the output directory from a folder/path node, while keeping the filename widget for the file name. "
         "A new slot appears as each one is filled, with a matching per-layer name field. "
-        "TIFF writes multi-page data and stores layer names as page descriptions; "
+        "Use this for composing multi-channel stacks. TIFF writes multi-page data and stores layer names as page descriptions; "
         "NPZ writes named arrays using those layer names as keys. "
         "Click Save to write (does not auto-run)."
     )
diff --git a/backend/nodes/view_3d.py b/backend/nodes/view_3d.py
index 9ffe623..8bfcef1 100644
--- a/backend/nodes/view_3d.py
+++ b/backend/nodes/view_3d.py
@@ -1,37 +1,119 @@
 from __future__ import annotations
+import base64
+import io
 import numpy as np
 from backend.node_registry import register_node
 from backend.data_types import (
     COLORMAPS,
     DataField,
+    ImageData,
+    MeshModel,
+    _annotation_context_from_field,
     colormap_to_uint8,
     normalize_for_colormap,
     resolve_colormap_input,
 )
 
 
+def _darken_colors(colors: np.ndarray, factor: float) -> np.ndarray:
+    return np.clip(np.rint(colors.astype(np.float32) * factor), 0, 255).astype(np.uint8)
+
+
+def _grid_triangle_indices(nx: int, ny: int, *, reverse: bool = False) -> list[list[int]]:
+    faces: list[list[int]] = []
+    for iy in range(ny - 1):
+        for ix in range(nx - 1):
+            a = iy * nx + ix
+            b = a + 1
+            c = a + nx
+            d = c + 1
+            if reverse:
+                faces.append([a, b, c])
+                faces.append([b, d, c])
+            else:
+                faces.append([a, c, b])
+                faces.append([b, c, d])
+    return faces
+
+
+def _build_mesh_model(z: np.ndarray, colors_u8: np.ndarray, z_scale: float, make_solid: bool) -> MeshModel:
+    ny, nx = z.shape
+    zmin = float(z.min())
+    zmax = float(z.max())
+    z_range = zmax - zmin if zmax != zmin else 1.0
+
+    top_vertices = np.empty((nx * ny, 3), dtype=np.float32)
+    top_colors = colors_u8.reshape(-1, 3).astype(np.uint8)
+    for iy in range(ny):
+        py = iy / max(ny - 1, 1) - 0.5
+        for ix in range(nx):
+            idx = iy * nx + ix
+            px = ix / max(nx - 1, 1) - 0.5
+            pz = ((float(z[iy, ix]) - zmin) / z_range - 0.5) * z_scale
+            top_vertices[idx] = (px, pz, py)
+
+    faces = _grid_triangle_indices(nx, ny)
+    if not make_solid:
+        return MeshModel(vertices=top_vertices, faces=np.asarray(faces, dtype=np.int32), colors=top_colors)
+
+    base_y = float(top_vertices[:, 1].min())
+    bottom_vertices = top_vertices.copy()
+    bottom_vertices[:, 1] = base_y
+    bottom_colors = _darken_colors(top_colors, 0.35)
+
+    vertices = np.vstack([top_vertices, bottom_vertices]).astype(np.float32)
+    colors = np.vstack([top_colors, bottom_colors]).astype(np.uint8)
+
+    bottom_offset = len(top_vertices)
+    faces.extend([[a + bottom_offset, b + bottom_offset, c + bottom_offset] for a, b, c in _grid_triangle_indices(nx, ny, reverse=True)])
+
+    def _add_wall(a: int, b: int):
+        faces.append([a, a + bottom_offset, b])
+        faces.append([b, a + bottom_offset, b + bottom_offset])
+
+    for ix in range(nx - 1):
+        _add_wall(ix, ix + 1)
+        top_row = (ny - 1) * nx
+        _add_wall(top_row + ix + 1, top_row + ix)
+    for iy in range(ny - 1):
+        _add_wall((iy + 1) * nx, iy * nx)
+        _add_wall(iy * nx + (nx - 1), (iy + 1) * nx + (nx - 1))
+
+    return MeshModel(vertices=vertices, faces=np.asarray(faces, dtype=np.int32), colors=colors)
+
+
 @register_node(display_name="3D View")
 class View3D:
+    _CUSTOM_PREVIEW = True
+
     @classmethod
     def INPUT_TYPES(cls):
         return {
             "required": {
-                "field": ("DATA_FIELD",),
+                "field": ("DATA_FIELD", {"label": "mesh"}),
                 "colormap": (["auto"] + list(COLORMAPS), {"hide_when_input_connected": "colormap_map"}),
                 "z_scale": ("FLOAT", {"default": 1, "min": 0.1, "max": 10.0, "step": 0.05}),
                 "resolution": ("INT", {"default": 128, "min": 32, "max": 512, "step": 16}),
+                "make_solid": ("BOOLEAN", {"default": False}),
+                "camera_azimuth": ("FLOAT", {"default": 0.0, "hidden": True}),
+                "camera_polar": ("FLOAT", {"default": 1.1, "hidden": True}),
+                "camera_distance": ("FLOAT", {"default": 1.8, "hidden": True}),
+                "viewport_snapshot": ("STRING", {"default": "", "hidden": True}),
             },
             "optional": {
+                "map_field": ("DATA_FIELD", {"label": "map"}),
                 "colormap_map": ("COLORMAP", {"label": "colormap"}),
             },
         }
 
-    RETURN_TYPES = ()
+    RETURN_TYPES = ("MESH_MODEL", "IMAGE")
+    RETURN_NAMES = ("mesh", "viewport")
     FUNCTION = "render"
 
     OUTPUT_NODE = True
     DESCRIPTION = (
         "Interactive 3D surface view of a DATA_FIELD. "
+        "Use the mesh input for geometry and optionally a second map input for coloring. "
         "Drag to rotate, scroll to zoom. z_scale exaggerates height."
     )
 
@@ -40,9 +122,12 @@ class View3D:
 
     def render(
         self, field: DataField,
-        colormap: str, z_scale: float, resolution: int, colormap_map=None,
+        colormap: str, z_scale: float, resolution: int, make_solid: bool = False,
+        camera_azimuth: float = 0.0, camera_polar: float = 1.1, camera_distance: float = 1.8,
+        viewport_snapshot: str = "",
+        map_field: DataField | None = None, colormap_map=None,
     ) -> tuple:
-        import base64
+        from scipy.ndimage import map_coordinates
 
         data = field.data
         yres, xres = data.shape
@@ -53,33 +138,75 @@ class View3D:
         ny, nx = z.shape
 
         zmin, zmax = float(z.min()), float(z.max())
+        color_field = map_field if map_field is not None else field
+        color_data = color_field.data
+
+        if color_field is field and color_data.shape == z.shape:
+            color_samples = z
+        elif color_field is field:
+            color_samples = color_data[::step_y, ::step_x].astype(np.float32)
+        else:
+            x_phys = np.linspace(field.xoff, field.xoff + field.xreal, nx, dtype=np.float64)
+            y_phys = np.linspace(field.yoff, field.yoff + field.yreal, ny, dtype=np.float64)
+            grid_y, grid_x = np.meshgrid(y_phys, x_phys, indexing="ij")
+
+            map_x = np.clip(
+                (grid_x - color_field.xoff) / max(color_field.xreal, 1e-12) * max(color_field.xres - 1, 0),
+                0.0,
+                max(color_field.xres - 1, 0),
+            )
+            map_y = np.clip(
+                (grid_y - color_field.yoff) / max(color_field.yreal, 1e-12) * max(color_field.yres - 1, 0),
+                0.0,
+                max(color_field.yres - 1, 0),
+            )
+            color_samples = map_coordinates(
+                color_data.astype(np.float64),
+                [map_y, map_x],
+                order=1,
+                mode="nearest",
+            ).astype(np.float32)
+
         z_norm = normalize_for_colormap(
-            z,
-            offset=field.display_offset,
-            scale=field.display_scale,
-            data_min=float(field.data.min()),
-            data_max=float(field.data.max()),
+            color_samples,
+            offset=color_field.display_offset,
+            scale=color_field.display_scale,
+            data_min=float(color_field.data.min()),
+            data_max=float(color_field.data.max()),
         )
 
         resolved_colormap = resolve_colormap_input(
             colormap,
             colormap_input=colormap_map,
-            inherited=field.colormap,
+            inherited=color_field.colormap,
             default="gray",
         )
         colors_u8 = colormap_to_uint8(z_norm, resolved_colormap)
+        mesh_model = _build_mesh_model(z, colors_u8, float(z_scale * 0.1), bool(make_solid))
 
         z_b64 = base64.b64encode(z.tobytes()).decode()
         colors_b64 = base64.b64encode(colors_u8.tobytes()).decode()
+        positions_b64 = base64.b64encode(np.asarray(mesh_model.vertices, dtype=np.float32).tobytes()).decode()
+        indices_b64 = base64.b64encode(np.asarray(mesh_model.faces, dtype=np.uint32).tobytes()).decode()
+        mesh_colors_b64 = None
+        if mesh_model.colors is not None:
+            mesh_colors_b64 = base64.b64encode(np.asarray(mesh_model.colors, dtype=np.uint8).tobytes()).decode()
 
         mesh_data = {
             "width": nx,
             "height": ny,
             "z_data": z_b64,
             "colors": colors_b64,
+            "positions": positions_b64,
+            "indices": indices_b64,
+            "vertex_colors": mesh_colors_b64,
             "z_min": zmin,
             "z_max": zmax,
             "z_scale": float(z_scale * 0.1),
+            "make_solid": bool(make_solid),
+            "camera_azimuth": float(camera_azimuth),
+            "camera_polar": float(camera_polar),
+            "camera_distance": float(camera_distance),
             "x_range": [float(field.xoff), float(field.xoff + field.xreal)],
             "y_range": [float(field.yoff), float(field.yoff + field.yreal)],
         }
@@ -87,4 +214,32 @@ class View3D:
         if View3D._broadcast_mesh_fn is not None:
             View3D._broadcast_mesh_fn(View3D._current_node_id, mesh_data)
 
-        return ()
+        annotation_context = _annotation_context_from_field(color_field, resolved_colormap)
+        annotation_context["xreal"] = float(field.xreal)
+        annotation_context["si_unit_xy"] = str(field.si_unit_xy)
+        viewport_image = ImageData(
+            self._decode_viewport_snapshot(viewport_snapshot),
+            metadata={
+                "annotation_context": annotation_context,
+                "viewport_camera": {
+                    "azimuth": float(camera_azimuth),
+                    "polar": float(camera_polar),
+                    "distance": float(camera_distance),
+                },
+            },
+        )
+        return (mesh_model, viewport_image)
+
+    def _decode_viewport_snapshot(self, snapshot: str) -> np.ndarray:
+        text = str(snapshot or "").strip()
+        if not text.startswith("data:image/"):
+            return np.zeros((1, 1, 3), dtype=np.uint8)
+
+        try:
+            header, payload = text.split(",", 1)
+            raw = base64.b64decode(payload)
+            from PIL import Image
+            image = Image.open(io.BytesIO(raw)).convert("RGB")
+            return np.asarray(image, dtype=np.uint8)
+        except Exception:
+            return np.zeros((1, 1, 3), dtype=np.uint8)
diff --git a/frontend/src/App.jsx b/frontend/src/App.jsx
index 2d6cd2c..a89b236 100644
--- a/frontend/src/App.jsx
+++ b/frontend/src/App.jsx
@@ -16,6 +16,12 @@ import { embedWorkflow, extractWorkflow } from './pngMetadata';
 import { captureViewportBlob as captureWorkflowViewportBlob } from './workflowCapture';
 import { hydrateWorkflowState } from './workflowHydration';
 import { serializeWorkflowState } from './workflowSerialization';
+import {
+  buildNodeClipboardPayload,
+  instantiateNodeClipboardPayload,
+  NODE_CLIPBOARD_MIME,
+  parseNodeClipboardPayload,
+} from './nodeClipboard';
 import { loadDefaultWorkflowAsset } from './defaultWorkflow';
 import {
   serializeExecutionGraph,
@@ -49,6 +55,12 @@ function sameStringArray(a = [], b = []) {
   return a.every((item, index) => item === b[index]);
 }
 
+function isEditableTarget(target) {
+  if (!target || !(target instanceof Element)) return false;
+  if (target.closest('input, textarea, select')) return true;
+  return target.closest('[contenteditable="true"]') !== null;
+}
+
 function compareMenuNodes(a, b) {
   const orderA = Number.isFinite(a?.def?.menu_order) ? a.def.menu_order : Number.MAX_SAFE_INTEGER;
   const orderB = Number.isFinite(b?.def?.menu_order) ? b.def.menu_order : Number.MAX_SAFE_INTEGER;
@@ -254,7 +266,11 @@ function ContextMenu({ x, y, nodeDefs, onAdd, onClose, filterType, filterDirecti
           });
           if (!hasMatch) continue;
         } else {
-          if (!def.output.some((type) => socketTypesCompatible(type, filterType))) continue;
+          const hasMatch = def.output.some((type) =>
+            socketTypesCompatible(type, filterType)
+            || (type === 'ANNOTATION_SOURCE' && (filterType === 'DATA_FIELD' || filterType === 'IMAGE'))
+          );
+          if (!hasMatch) continue;
         }
       }
       const cat = def.category || 'uncategorized';
@@ -454,6 +470,8 @@ function Flow() {
   const autoRunTimer = useRef(null);
   const autoRunRef = useRef(null);
   const defaultWorkflowLoadAttemptedRef = useRef(false);
+  const lastPastedClipboardTextRef = useRef('');
+  const pasteRepeatCountRef = useRef(0);
   const reactFlow = useReactFlow();
 
   // ── WebSocket ───────────────────────────────────────────────────────
@@ -554,6 +572,24 @@ function Flow() {
     }
   }, [reactFlow, refreshLoadNodeOutputs, setNodeOutputs]);
 
+  const refreshAnnotationNodeOutputs = useCallback((nodeId) => {
+    const node = reactFlow.getNode(nodeId);
+    if (!node) return;
+
+    const inputEdge = reactFlow.getEdges().find(
+      (edge) => edge.target === nodeId && getInputName(edge.targetHandle) === 'input'
+    );
+    const outputType = inputEdge ? getHandleType(inputEdge.sourceHandle) : 'ANNOTATION_SOURCE';
+    setNodeOutputs(nodeId, [outputType], ['Output']);
+
+    if (!inputEdge || outputType === 'ANNOTATION_SOURCE') return;
+
+    setEdges((prev) => prev.filter((edge) => {
+      if (edge.source !== nodeId) return true;
+      return socketTypesCompatible(outputType, getHandleType(edge.targetHandle));
+    }));
+  }, [reactFlow, setEdges, setNodeOutputs]);
+
   useEffect(() => {
     api.setMessageHandler((msg) => {
       console.log('[argonode] WS:', msg.type, msg.data?.node_id || msg.data?.node || '');
@@ -639,14 +675,21 @@ function Flow() {
         refreshLoadNodeOutputs(params.target);
       }, 0);
     }
+    const targetNode = reactFlow.getNode(params.target);
+    if (targetNode && (targetNode.data.className === 'Annotations' || targetNode.data.className === 'Markup')) {
+      setTimeout(() => {
+        refreshAnnotationNodeOutputs(params.target);
+      }, 0);
+    }
     scheduleAutoRun();
-  }, [refreshLoadNodeOutputs, setEdges]); // scheduleAutoRun is stable (no deps)
+  }, [reactFlow, refreshAnnotationNodeOutputs, refreshLoadNodeOutputs, setEdges]); // scheduleAutoRun is stable (no deps)
 
   const handleEdgesChange = useCallback((changes) => {
     const currentEdges = reactFlow.getEdges();
     onEdgesChange(changes);
 
     const affectedPathTargets = new Set();
+    const affectedAnnotationTargets = new Set();
     for (const change of changes) {
       if (change.type !== 'remove') continue;
       const removedEdge = currentEdges.find((edge) => edge.id === change.id);
@@ -654,6 +697,12 @@ function Flow() {
       if (getInputName(removedEdge.targetHandle) === 'path') {
         affectedPathTargets.add(removedEdge.target);
       }
+      if (getInputName(removedEdge.targetHandle) === 'input') {
+        const targetNode = reactFlow.getNode(removedEdge.target);
+        if (targetNode && (targetNode.data.className === 'Annotations' || targetNode.data.className === 'Markup')) {
+          affectedAnnotationTargets.add(removedEdge.target);
+        }
+      }
     }
 
     if (affectedPathTargets.size > 0) {
@@ -663,7 +712,14 @@ function Flow() {
         });
       }, 0);
     }
-  }, [onEdgesChange, reactFlow, refreshLoadNodeOutputs]);
+    if (affectedAnnotationTargets.size > 0) {
+      setTimeout(() => {
+        affectedAnnotationTargets.forEach((nodeId) => {
+          refreshAnnotationNodeOutputs(nodeId);
+        });
+      }, 0);
+    }
+  }, [onEdgesChange, reactFlow, refreshAnnotationNodeOutputs, refreshLoadNodeOutputs]);
 
   // ── Drop-on-blank: open filtered context menu ──────────────────────
 
@@ -749,12 +805,6 @@ function Flow() {
     });
   }, [reactFlow]);
 
-  const contextValue = useMemo(() => ({
-    onWidgetChange,
-    openFileBrowser,
-    onManualTrigger,
-  }), [onWidgetChange, openFileBrowser, onManualTrigger]);
-
   // ── Add node from context menu ──────────────────────────────────────
 
   const addNode = useCallback((className, def) => {
@@ -789,6 +839,7 @@ function Flow() {
         className,
         definition: def,
         widgetValues,
+        runtimeValues: {},
         previewImage: null,
         tableRows: null,
         meshData: null,
@@ -842,9 +893,12 @@ function Flow() {
         }
       } else {
         // Dragged from an input → connect from the first matching output on the new node
-        const outputIdx = def.output.findIndex((type) => socketTypesCompatible(type, filterType));
+        const outputIdx = def.output.findIndex((type) =>
+          socketTypesCompatible(type, filterType)
+          || (type === 'ANNOTATION_SOURCE' && (filterType === 'DATA_FIELD' || filterType === 'IMAGE'))
+        );
         if (outputIdx !== -1) {
-          const outputType = def.output[outputIdx];
+          const outputType = def.output[outputIdx] === 'ANNOTATION_SOURCE' ? filterType : def.output[outputIdx];
           const sourceHandle = `output::${outputIdx}::${outputType}`;
           const color = TYPE_COLORS[outputType] || 'var(--fallback-type)';
           setEdges((eds) => addEdge({
@@ -907,6 +961,101 @@ function Flow() {
     autoRunTimer.current = setTimeout(() => autoRunRef.current?.(), 300);
   }, []);
 
+  const onRuntimeValuesChange = useCallback((nodeId, patch, { scheduleRun = false } = {}) => {
+    if (!patch || typeof patch !== 'object') return;
+
+    setNodes((ns) => ns.map((n) => {
+      if (n.id !== nodeId) return n;
+      return {
+        ...n,
+        data: {
+          ...n.data,
+          runtimeValues: { ...(n.data.runtimeValues || {}), ...patch },
+        },
+      };
+    }));
+
+    if (scheduleRun) {
+      scheduleAutoRun();
+    }
+  }, [setNodes, scheduleAutoRun]);
+
+  const pasteClipboardSelection = useCallback((clipboardText) => {
+    const payload = parseNodeClipboardPayload(clipboardText);
+    if (!payload) return false;
+
+    if (clipboardText === lastPastedClipboardTextRef.current) {
+      pasteRepeatCountRef.current += 1;
+    } else {
+      lastPastedClipboardTextRef.current = clipboardText;
+      pasteRepeatCountRef.current = 1;
+    }
+
+    const offsetAmount = 36 * pasteRepeatCountRef.current;
+    const pasted = instantiateNodeClipboardPayload(
+      payload,
+      nodeDefsRef.current,
+      nextIdRef.current,
+      { x: offsetAmount, y: offsetAmount },
+    );
+
+    if (pasted.nodes.length === 0) return false;
+
+    nextIdRef.current = pasted.nextNodeId;
+
+    setNodes((existing) => [
+      ...existing.map((node) => ({ ...node, selected: false })),
+      ...pasted.nodes,
+    ]);
+    setEdges((existing) => [
+      ...existing.map((edge) => ({ ...edge, selected: false })),
+      ...pasted.edges,
+    ]);
+
+    setTimeout(() => {
+      pasted.nodes.forEach((node) => {
+        if (node.data.className === 'Folder' && node.data.widgetValues?.folder) {
+          refreshFolderNodeOutputs(node.id, node.data.widgetValues.folder);
+        }
+      });
+      pasted.nodes.forEach((node) => {
+        if (node.data.className === 'Image' || node.data.className === 'ImageDemo') {
+          refreshLoadNodeOutputs(node.id);
+        }
+      });
+      pasted.nodes.forEach((node) => {
+        if (node.data.className === 'Annotations' || node.data.className === 'Markup') {
+          refreshAnnotationNodeOutputs(node.id);
+        }
+      });
+      pasted.nodes.forEach((node) => {
+        reactFlow.updateNodeInternals(node.id);
+      });
+    }, 0);
+
+    setStatus({
+      text: `Pasted ${pasted.nodes.length} node${pasted.nodes.length === 1 ? '' : 's'}.`,
+      level: 'info',
+    });
+    scheduleAutoRun();
+    return true;
+  }, [
+    reactFlow,
+    refreshAnnotationNodeOutputs,
+    refreshFolderNodeOutputs,
+    refreshLoadNodeOutputs,
+    scheduleAutoRun,
+    setEdges,
+    setNodes,
+  ]);
+
+  const contextValue = useMemo(() => ({
+    onWidgetChange,
+    onRuntimeValuesChange,
+    openFileBrowser,
+    onManualTrigger,
+  }), [onRuntimeValuesChange, onWidgetChange, openFileBrowser, onManualTrigger]);
+
   const clearGraph = useCallback(() => {
     setNodes([]);
     setEdges([]);
@@ -930,8 +1079,13 @@ function Flow() {
           refreshLoadNodeOutputs(node.id);
         }
       });
+      hydrated.nodes.forEach((node) => {
+        if (node.data.className === 'Annotations' || node.data.className === 'Markup') {
+          refreshAnnotationNodeOutputs(node.id);
+        }
+      });
     }, 0);
-  }, [refreshFolderNodeOutputs, refreshLoadNodeOutputs, setNodes, setEdges]);
+  }, [refreshAnnotationNodeOutputs, refreshFolderNodeOutputs, refreshLoadNodeOutputs, setNodes, setEdges]);
 
   const loadDefaultWorkflow = useCallback(async () => {
     if (defaultWorkflowLoadAttemptedRef.current) return;
@@ -1168,6 +1322,45 @@ function Flow() {
     return () => window.removeEventListener('keydown', handler);
   }, [runWorkflow]);
 
+  useEffect(() => {
+    const handleCopy = (event) => {
+      if (isEditableTarget(event.target)) return;
+
+      const payload = buildNodeClipboardPayload(reactFlow.getNodes(), reactFlow.getEdges());
+      if (!payload) return;
+
+      const serialized = JSON.stringify(payload);
+      event.preventDefault();
+      event.clipboardData?.setData(NODE_CLIPBOARD_MIME, serialized);
+      event.clipboardData?.setData('text/plain', serialized);
+      setStatus({
+        text: `Copied ${payload.nodes.length} node${payload.nodes.length === 1 ? '' : 's'}.`,
+        level: 'info',
+      });
+    };
+
+    const handlePaste = (event) => {
+      if (isEditableTarget(event.target)) return;
+
+      const clipboardText = event.clipboardData?.getData(NODE_CLIPBOARD_MIME)
+        || event.clipboardData?.getData('text/plain')
+        || '';
+      if (!clipboardText) return;
+
+      const pasted = pasteClipboardSelection(clipboardText);
+      if (pasted) {
+        event.preventDefault();
+      }
+    };
+
+    window.addEventListener('copy', handleCopy);
+    window.addEventListener('paste', handlePaste);
+    return () => {
+      window.removeEventListener('copy', handleCopy);
+      window.removeEventListener('paste', handlePaste);
+    };
+  }, [pasteClipboardSelection, reactFlow]);
+
   // ── Context menu ────────────────────────────────────────────────────
 
   const onPaneContextMenu = useCallback((event) => {
diff --git a/frontend/src/CustomNode.jsx b/frontend/src/CustomNode.jsx
index 37d1b00..b5b4666 100644
--- a/frontend/src/CustomNode.jsx
+++ b/frontend/src/CustomNode.jsx
@@ -1075,7 +1075,13 @@ function CustomNode({ id, data }) {
         {data.meshData && (
           <CollapsibleSection title="3D View" defaultOpen={true}>
             <Suspense fallback={<div className="node-preview" style={{color:'var(--text-muted)',padding:4}}>Loading 3D...</div>}>
-              <SurfaceView meshData={data.meshData} />
+              <SurfaceView
+                meshData={data.meshData}
+                nodeId={id}
+                widgetValues={data.widgetValues}
+                runtimeValues={data.runtimeValues}
+                onRuntimeValuesChange={ctx.onRuntimeValuesChange}
+              />
             </Suspense>
           </CollapsibleSection>
         )}
diff --git a/frontend/src/SurfaceView.jsx b/frontend/src/SurfaceView.jsx
index 8247ed7..df8e53e 100644
--- a/frontend/src/SurfaceView.jsx
+++ b/frontend/src/SurfaceView.jsx
@@ -8,9 +8,13 @@ import { OrbitControls } from 'three/examples/jsm/controls/OrbitControls.js';
  *   meshData: { width, height, z_data (b64 float32), colors (b64 uint8 RGB),
  *               z_min, z_max, z_scale, x_range, y_range }
  */
-export default function SurfaceView({ meshData }) {
+export default function SurfaceView({ meshData, nodeId, widgetValues, runtimeValues, onRuntimeValuesChange }) {
   const containerRef = useRef(null);
   const threeRef = useRef(null); // { renderer, scene, camera, controls, mesh }
+  const syncTimerRef = useRef(null);
+  const lastSnapshotRef = useRef('');
+  const lastAnglesRef = useRef({ azimuth: null, polar: null, distance: null });
+  const hasSyncedInitialSnapshotRef = useRef(false);
 
   // Decode base64 to typed arrays
   const decode = useCallback((b64, ArrayType) => {
@@ -20,6 +24,52 @@ export default function SurfaceView({ meshData }) {
     return new ArrayType(bytes.buffer);
   }, []);
 
+  const syncViewportState = useCallback((scheduleRun = false) => {
+    const state = threeRef.current;
+    if (!state || !nodeId || !onRuntimeValuesChange) return;
+    const { renderer, controls } = state;
+    const azimuth = Number(controls.getAzimuthalAngle().toFixed(4));
+    const polar = Number(controls.getPolarAngle().toFixed(4));
+    const distance = Number(controls.getDistance().toFixed(4));
+    const snapshot = renderer.domElement.toDataURL('image/png');
+    const previous = lastAnglesRef.current;
+    const patch = {};
+    if (previous.azimuth !== azimuth) patch.camera_azimuth = azimuth;
+    if (previous.polar !== polar) patch.camera_polar = polar;
+    if (previous.distance !== distance) patch.camera_distance = distance;
+    if (snapshot !== lastSnapshotRef.current) patch.viewport_snapshot = snapshot;
+    if (Object.keys(patch).length > 0) {
+      onRuntimeValuesChange(nodeId, patch, { scheduleRun });
+      lastAnglesRef.current = { azimuth, polar, distance };
+      lastSnapshotRef.current = snapshot;
+    }
+  }, [nodeId, onRuntimeValuesChange]);
+
+  const scheduleViewportSync = useCallback((delay = 120, scheduleRun = false) => {
+    if (syncTimerRef.current) {
+      clearTimeout(syncTimerRef.current);
+    }
+    syncTimerRef.current = setTimeout(() => {
+      syncTimerRef.current = null;
+      syncViewportState(scheduleRun);
+    }, delay);
+  }, [syncViewportState]);
+
+  const applyCameraState = useCallback((azimuth, polar, distance) => {
+    const state = threeRef.current;
+    if (!state) return;
+    const { camera, controls } = state;
+    const target = controls.target.clone();
+    const spherical = new THREE.Spherical(
+      Math.max(0.3, Number.isFinite(distance) ? distance : 1.8),
+      THREE.MathUtils.clamp(Number.isFinite(polar) ? polar : 1.1, 0.01, Math.PI - 0.01),
+      Number.isFinite(azimuth) ? azimuth : 0.0,
+    );
+    const offset = new THREE.Vector3().setFromSpherical(spherical);
+    camera.position.copy(target).add(offset);
+    controls.update();
+  }, []);
+
   // Initialize Three.js scene once
   useEffect(() => {
     const container = containerRef.current;
@@ -48,6 +98,8 @@ export default function SurfaceView({ meshData }) {
     controls.dampingFactor = 0.1;
     controls.minDistance = 0.3;
     controls.maxDistance = 10;
+    const handleControlsEnd = () => scheduleViewportSync(0, true);
+    controls.addEventListener('end', handleControlsEnd);
 
     // Lighting
     const ambient = new THREE.AmbientLight(0xffffff, 0.4);
@@ -69,6 +121,11 @@ export default function SurfaceView({ meshData }) {
     animate();
 
     threeRef.current = { renderer, scene, camera, controls, mesh: null, animId };
+    applyCameraState(
+      Number(runtimeValues?.camera_azimuth ?? widgetValues?.camera_azimuth),
+      Number(runtimeValues?.camera_polar ?? widgetValues?.camera_polar),
+      Number(runtimeValues?.camera_distance ?? widgetValues?.camera_distance),
+    );
 
     // Resize observer to maintain 1:1 aspect when node width changes
     const ro = new ResizeObserver((entries) => {
@@ -86,6 +143,8 @@ export default function SurfaceView({ meshData }) {
     return () => {
       ro.disconnect();
       cancelAnimationFrame(animId);
+      if (syncTimerRef.current) clearTimeout(syncTimerRef.current);
+      controls.removeEventListener('end', handleControlsEnd);
       controls.dispose();
       renderer.dispose();
       if (container.contains(renderer.domElement)) {
@@ -93,18 +152,24 @@ export default function SurfaceView({ meshData }) {
       }
       threeRef.current = null;
     };
-  }, []);
+  }, [applyCameraState, scheduleViewportSync]);
 
   // Update mesh when data changes
   useEffect(() => {
     if (!threeRef.current || !meshData) return;
 
     const { scene, camera, controls } = threeRef.current;
-    const { width: nx, height: ny, z_data, colors, z_min, z_max, z_scale, x_range, y_range } = meshData;
+    const {
+      width: nx, height: ny, z_data, colors, z_min, z_max, z_scale,
+      positions, indices, vertex_colors, camera_azimuth, camera_polar, camera_distance,
+    } = meshData;
 
     // Decode arrays
-    const zArr = decode(z_data, Float32Array);
-    const colArr = decode(colors, Uint8Array);
+    const zArr = z_data ? decode(z_data, Float32Array) : null;
+    const colArr = colors ? decode(colors, Uint8Array) : null;
+    const posArr = positions ? decode(positions, Float32Array) : null;
+    const indexArr = indices ? decode(indices, Uint32Array) : null;
+    const vertexColorArr = vertex_colors ? decode(vertex_colors, Uint8Array) : null;
 
     // Remove old mesh
     if (threeRef.current.mesh) {
@@ -115,45 +180,51 @@ export default function SurfaceView({ meshData }) {
 
     // Build geometry
     const geom = new THREE.BufferGeometry();
-    const positions = new Float32Array(nx * ny * 3);
-    const colorAttr = new Float32Array(nx * ny * 3);
+    const positionsArray = posArr ?? new Float32Array(nx * ny * 3);
+    const colorAttr = new Float32Array((vertexColorArr ? vertexColorArr.length : (nx * ny * 3)));
 
-    // Normalize coordinates to roughly [-0.5, 0.5] for good camera framing
-    const zRange = z_max - z_min || 1;
+    if (!posArr) {
+      const zRange = z_max - z_min || 1;
+      for (let iy = 0; iy < ny; iy++) {
+        for (let ix = 0; ix < nx; ix++) {
+          const idx = iy * nx + ix;
+          const px = ix / (nx - 1) - 0.5;
+          const py = iy / (ny - 1) - 0.5;
+          const pz = ((zArr[idx] - z_min) / zRange - 0.5) * z_scale;
 
-    for (let iy = 0; iy < ny; iy++) {
-      for (let ix = 0; ix < nx; ix++) {
-        const idx = iy * nx + ix;
-        const px = ix / (nx - 1) - 0.5;  // [-0.5, 0.5]
-        const py = iy / (ny - 1) - 0.5;
-        const pz = ((zArr[idx] - z_min) / zRange - 0.5) * z_scale;
-
-        positions[idx * 3] = px;
-        positions[idx * 3 + 1] = pz; // height on Y axis
-        positions[idx * 3 + 2] = py;
-
-        colorAttr[idx * 3] = colArr[idx * 3] / 255;
-        colorAttr[idx * 3 + 1] = colArr[idx * 3 + 1] / 255;
-        colorAttr[idx * 3 + 2] = colArr[idx * 3 + 2] / 255;
+          positionsArray[idx * 3] = px;
+          positionsArray[idx * 3 + 1] = pz;
+          positionsArray[idx * 3 + 2] = py;
+        }
       }
     }
 
-    geom.setAttribute('position', new THREE.BufferAttribute(positions, 3));
+    const sourceColors = vertexColorArr ?? colArr;
+    if (sourceColors) {
+      for (let i = 0; i < sourceColors.length; i += 1) {
+        colorAttr[i] = sourceColors[i] / 255;
+      }
+    }
+
+    geom.setAttribute('position', new THREE.BufferAttribute(positionsArray, 3));
     geom.setAttribute('color', new THREE.BufferAttribute(colorAttr, 3));
 
-    // Build index (triangles from grid)
-    const indices = [];
-    for (let iy = 0; iy < ny - 1; iy++) {
-      for (let ix = 0; ix < nx - 1; ix++) {
-        const a = iy * nx + ix;
-        const b = a + 1;
-        const c = a + nx;
-        const d = c + 1;
-        indices.push(a, c, b);
-        indices.push(b, c, d);
+    if (indexArr) {
+      geom.setIndex(Array.from(indexArr));
+    } else {
+      const gridIndices = [];
+      for (let iy = 0; iy < ny - 1; iy++) {
+        for (let ix = 0; ix < nx - 1; ix++) {
+          const a = iy * nx + ix;
+          const b = a + 1;
+          const c = a + nx;
+          const d = c + 1;
+          gridIndices.push(a, c, b);
+          gridIndices.push(b, c, d);
+        }
       }
+      geom.setIndex(gridIndices);
     }
-    geom.setIndex(indices);
     geom.computeVertexNormals();
 
     const mat = new THREE.MeshPhongMaterial({
@@ -169,8 +240,16 @@ export default function SurfaceView({ meshData }) {
 
     // Reset camera target to center of mesh
     controls.target.set(0, 0, 0);
-    controls.update();
-  }, [meshData, decode]);
+    if (!hasSyncedInitialSnapshotRef.current) {
+      applyCameraState(
+        Number.isFinite(camera_azimuth) ? camera_azimuth : Number(runtimeValues?.camera_azimuth ?? widgetValues?.camera_azimuth),
+        Number.isFinite(camera_polar) ? camera_polar : Number(runtimeValues?.camera_polar ?? widgetValues?.camera_polar),
+        Number.isFinite(camera_distance) ? camera_distance : Number(runtimeValues?.camera_distance ?? widgetValues?.camera_distance),
+      );
+      hasSyncedInitialSnapshotRef.current = true;
+    }
+    scheduleViewportSync(0, false);
+  }, [meshData, decode, applyCameraState, runtimeValues, scheduleViewportSync, widgetValues]);
 
   // Prevent scroll events from propagating to React Flow
   const onWheel = useCallback((e) => {
diff --git a/frontend/src/constants.js b/frontend/src/constants.js
index cf398aa..691da4d 100644
--- a/frontend/src/constants.js
+++ b/frontend/src/constants.js
@@ -2,8 +2,8 @@
 
 export const DATA_TYPES = new Set([
   'DATA_FIELD', 'IMAGE', 'LINE', 'MEASURE_TABLE', 'RECORD_TABLE', 'ANY_TABLE',
-  'COORD', 'STATS_SOURCE', 'CURSOR_SOURCE', 'VALUE_SOURCE', 'COLORMAP',
-  'SAVE_LAYER', 'FONT', 'FILE_PATH', 'DIRECTORY', 'COORDPAIR',
+  'COORD', 'STATS_SOURCE', 'CURSOR_SOURCE', 'VALUE_SOURCE', 'ANNOTATION_SOURCE', 'COLORMAP',
+  'SAVE_LAYER', 'SAVE_VALUE', 'MESH_MODEL', 'FONT', 'FILE_PATH', 'DIRECTORY', 'COORDPAIR',
 ]);
 
 export const SOCKET_WIDGET_TYPES = new Set(['FLOAT', 'INT']);
@@ -22,8 +22,11 @@ export const TYPE_COLORS = {
   STATS_SOURCE:  '#c084fc',
   CURSOR_SOURCE: '#a78bfa',
   VALUE_SOURCE:  '#60a5fa',
+  ANNOTATION_SOURCE: '#06b6d4',
   COLORMAP:      '#f472b6',
   SAVE_LAYER:    '#22c55e',
+  SAVE_VALUE:    '#4ade80',
+  MESH_MODEL:    '#14b8a6',
   FONT:          '#fb7185',
   FILE_PATH:     '#f59e0b',
   DIRECTORY:     '#f97316',
@@ -44,7 +47,9 @@ export const SOCKET_COMPATIBILITY = {
   CURSOR_SOURCE: new Set(['DATA_FIELD', 'LINE']),
   ANY_TABLE:     new Set(['MEASURE_TABLE', 'RECORD_TABLE']),
   VALUE_SOURCE:  new Set(['FLOAT', 'MEASURE_TABLE']),
+  ANNOTATION_SOURCE: new Set(['DATA_FIELD', 'IMAGE']),
   SAVE_LAYER:    new Set(['DATA_FIELD', 'IMAGE']),
+  SAVE_VALUE:    new Set(['DATA_FIELD', 'IMAGE', 'LINE', 'MEASURE_TABLE', 'RECORD_TABLE', 'MESH_MODEL', 'FLOAT']),
   FLOAT:         new Set(['INT']),
   INT:           new Set(['FLOAT']),
   LINE:          new Set(['COORDPAIR']),
diff --git a/frontend/src/executionGraph.js b/frontend/src/executionGraph.js
index 25e9b2f..1bc0cf7 100644
--- a/frontend/src/executionGraph.js
+++ b/frontend/src/executionGraph.js
@@ -52,11 +52,12 @@ export function serializeExecutionGraph(nodes, edges, { excludeManualTrigger = f
   for (const node of nodes) {
     if (!runnableNodeIds.has(node.id)) continue;
 
-    const { className, definition, widgetValues } = node.data;
+    const { className, definition, widgetValues, runtimeValues } = node.data;
     if (!definition) continue;
     if (excludeManualTrigger && definition.manual_trigger) continue;
 
     const inputs = {};
+    const valueBag = { ...(widgetValues || {}), ...(runtimeValues || {}) };
 
     const allWidgets = {
       ...(definition.input.required || {}),
@@ -66,8 +67,8 @@ export function serializeExecutionGraph(nodes, edges, { excludeManualTrigger = f
       const [type] = Array.isArray(spec) ? spec : [spec];
       if (DATA_TYPES.has(type)) continue;
       if (type === 'BUTTON') continue;
-      if (widgetValues[name] !== undefined) {
-        inputs[name] = widgetValues[name];
+      if (valueBag[name] !== undefined) {
+        inputs[name] = valueBag[name];
       }
     }
 
diff --git a/frontend/src/nodeClipboard.js b/frontend/src/nodeClipboard.js
new file mode 100644
index 0000000..b87e533
--- /dev/null
+++ b/frontend/src/nodeClipboard.js
@@ -0,0 +1,128 @@
+export const NODE_CLIPBOARD_KIND = 'argonode/node-selection';
+export const NODE_CLIPBOARD_MIME = 'application/x-argonode-node-selection';
+
+function cloneValue(value) {
+  if (value == null) return value;
+  if (typeof structuredClone === 'function') {
+    try {
+      return structuredClone(value);
+    } catch {
+      // Fall through to JSON clone for simple plain data.
+    }
+  }
+  return JSON.parse(JSON.stringify(value));
+}
+
+function clonePlainObject(value) {
+  if (!value || typeof value !== 'object' || Array.isArray(value)) return {};
+  return cloneValue(value) || {};
+}
+
+export function buildNodeClipboardPayload(nodes, edges) {
+  const selectedNodes = Array.isArray(nodes) ? nodes.filter((node) => node?.selected) : [];
+  if (selectedNodes.length === 0) return null;
+
+  const selectedIds = new Set(selectedNodes.map((node) => String(node.id)));
+  const internalEdges = Array.isArray(edges)
+    ? edges.filter((edge) => selectedIds.has(String(edge.source)) && selectedIds.has(String(edge.target)))
+    : [];
+
+  return {
+    kind: NODE_CLIPBOARD_KIND,
+    version: 1,
+    nodes: selectedNodes.map((node) => ({
+      id: String(node.id),
+      type: node.type || 'custom',
+      position: {
+        x: Number(node.position?.x) || 0,
+        y: Number(node.position?.y) || 0,
+      },
+      dragHandle: node.dragHandle || '.drag-handle',
+      data: {
+        label: node.data?.label || node.data?.className || 'Node',
+        className: node.data?.className || '',
+        widgetValues: clonePlainObject(node.data?.widgetValues),
+        runtimeValues: clonePlainObject(node.data?.runtimeValues),
+      },
+    })),
+    edges: internalEdges.map((edge) => ({
+      source: String(edge.source),
+      sourceHandle: edge.sourceHandle,
+      target: String(edge.target),
+      targetHandle: edge.targetHandle,
+      ...(edge.style ? { style: { ...edge.style } } : {}),
+    })),
+  };
+}
+
+export function parseNodeClipboardPayload(text) {
+  if (typeof text !== 'string' || !text.trim()) return null;
+
+  try {
+    const parsed = JSON.parse(text);
+    if (parsed?.kind !== NODE_CLIPBOARD_KIND) return null;
+    if (!Array.isArray(parsed.nodes) || !Array.isArray(parsed.edges)) return null;
+    return parsed;
+  } catch {
+    return null;
+  }
+}
+
+export function instantiateNodeClipboardPayload(payload, defs = {}, nextNodeId = 1, offset = { x: 40, y: 40 }) {
+  if (!payload || !Array.isArray(payload.nodes) || payload.nodes.length === 0) {
+    return { nodes: [], edges: [], nextNodeId };
+  }
+
+  const idMap = new Map();
+  let currentId = Number(nextNodeId) || 1;
+
+  const nodes = payload.nodes.map((node) => {
+    const newId = String(currentId++);
+    idMap.set(String(node.id), newId);
+    const className = node.data?.className || '';
+    const definition = className ? defs[className] || null : null;
+
+    return {
+      id: newId,
+      type: node.type || 'custom',
+      position: {
+        x: (Number(node.position?.x) || 0) + (Number(offset?.x) || 0),
+        y: (Number(node.position?.y) || 0) + (Number(offset?.y) || 0),
+      },
+      dragHandle: node.dragHandle || '.drag-handle',
+      selected: true,
+      data: {
+        label: node.data?.label || className || 'Node',
+        className,
+        widgetValues: clonePlainObject(node.data?.widgetValues),
+        runtimeValues: clonePlainObject(node.data?.runtimeValues),
+        definition,
+        previewImage: null,
+        tableRows: null,
+        meshData: null,
+        overlay: null,
+        scalarValue: null,
+        processingTimeMs: null,
+        warning: null,
+      },
+    };
+  });
+
+  const edges = payload.edges
+    .filter((edge) => idMap.has(String(edge.source)) && idMap.has(String(edge.target)))
+    .map((edge, index) => ({
+      id: `e${idMap.get(String(edge.source))}-${idMap.get(String(edge.target))}-${index}`,
+      source: idMap.get(String(edge.source)),
+      sourceHandle: edge.sourceHandle,
+      target: idMap.get(String(edge.target)),
+      targetHandle: edge.targetHandle,
+      selected: false,
+      ...(edge.style ? { style: { ...edge.style } } : {}),
+    }));
+
+  return {
+    nodes,
+    edges,
+    nextNodeId: currentId,
+  };
+}
diff --git a/frontend/src/pngMetadata.js b/frontend/src/pngMetadata.js
index 716d3e7..2e76afa 100644
--- a/frontend/src/pngMetadata.js
+++ b/frontend/src/pngMetadata.js
@@ -3,7 +3,7 @@
  *
  * PNG files are composed of chunks: [4-byte length][4-byte type][data][4-byte CRC].
  * We add an iTXt chunk with key "workflow" containing the JSON-serialised graph,
- * inserted just before the IEND chunk. We still read legacy tEXt chunks.
+ * inserted just before the IEND chunk.
  */
 
 // ── CRC32 (PNG uses CRC-32/ISO 3309) ────────────────────────────────
@@ -71,10 +71,6 @@ function parseTextChunk(type, chunkData) {
   const keyword = decoder.decode(chunkData.subarray(0, keywordEnd));
   if (keyword !== 'workflow') return null;
 
-  if (type === 'tEXt') {
-    return JSON.parse(decoder.decode(chunkData.subarray(keywordEnd + 1)));
-  }
-
   if (type !== 'iTXt') return null;
 
   const compressionFlagIdx = keywordEnd + 1;
@@ -139,7 +135,7 @@ export async function embedWorkflow(pngBlob, workflow) {
 }
 
 /**
- * Extract the workflow object from a PNG blob's iTXt/tEXt chunks.
+ * Extract the workflow object from a PNG blob's iTXt chunks.
  * Returns the parsed object, or null if no "workflow" key is found.
  */
 export async function extractWorkflow(pngBlob) {
@@ -154,7 +150,7 @@ export async function extractWorkflow(pngBlob) {
     if (pos + 12 + len > data.length) break;
     const type = chunkType(data, pos);
 
-    if (type === 'tEXt' || type === 'iTXt') {
+    if (type === 'iTXt') {
       const chunkData = data.subarray(pos + 8, pos + 8 + len);
       const parsed = parseTextChunk(type, chunkData);
       if (parsed) found = parsed;
diff --git a/frontend/src/workflowHydration.js b/frontend/src/workflowHydration.js
index 39691ee..3e97032 100644
--- a/frontend/src/workflowHydration.js
+++ b/frontend/src/workflowHydration.js
@@ -1,25 +1,7 @@
 function mergeDefinition(nodeData, defs) {
   const savedData = nodeData || {};
-  const savedDefinition = savedData.definition && typeof savedData.definition === 'object'
-    ? savedData.definition
-    : null;
   const registryDefinition = savedData.className ? defs[savedData.className] : null;
-  const definition = registryDefinition || savedDefinition;
-
-  if (!definition) return null;
-
-  const output = Array.isArray(savedData.output)
-    ? savedData.output
-    : (Array.isArray(savedDefinition?.output) ? savedDefinition.output : null);
-  const outputName = Array.isArray(savedData.output_name)
-    ? savedData.output_name
-    : (Array.isArray(savedDefinition?.output_name) ? savedDefinition.output_name : null);
-
-  return {
-    ...definition,
-    ...(output ? { output } : {}),
-    ...(outputName ? { output_name: outputName } : {}),
-  };
+  return registryDefinition || null;
 }
 
 function getSocketType(inputDef) {
@@ -28,12 +10,6 @@ function getSocketType(inputDef) {
   return Array.isArray(type) ? type[0] : type;
 }
 
-function getInputType(definition, inputName) {
-  const required = definition?.input?.required || {};
-  const optional = definition?.input?.optional || {};
-  return getSocketType(required[inputName] ?? optional[inputName]);
-}
-
 function getInputEntries(definition) {
   return [
     ...Object.entries(definition?.input?.required || {}),
@@ -54,31 +30,6 @@ function sanitizeWidgetValues(widgetValues, definition) {
   return nextValues;
 }
 
-function remapLegacyHandle(handleId, kind, nodeData) {
-  if (typeof handleId !== 'string') return handleId;
-
-  const parts = handleId.split('::');
-  if (parts.length !== 3 || parts[2] !== 'TABLE') return handleId;
-
-  if (kind === 'source' && parts[0] === 'output') {
-    const outputSlot = Number.parseInt(parts[1], 10);
-    const outputType = nodeData?.definition?.output?.[outputSlot];
-    if (typeof outputType === 'string' && outputType !== 'TABLE') {
-      return `output::${outputSlot}::${outputType}`;
-    }
-    return handleId;
-  }
-
-  if (kind === 'target' && parts[0] === 'input') {
-    const inputType = getInputType(nodeData?.definition, parts[1]);
-    if (typeof inputType === 'string' && inputType !== 'TABLE') {
-      return `input::${parts[1]}::${inputType}`;
-    }
-  }
-
-  return handleId;
-}
-
 export function hydrateWorkflowState(data, defs = {}) {
   const loadedNodes = Array.isArray(data?.nodes) ? data.nodes : [];
   const loadedEdges = Array.isArray(data?.edges) ? data.edges : [];
@@ -94,6 +45,7 @@ export function hydrateWorkflowState(data, defs = {}) {
         ...node.data,
         label: node.data?.label || node.data?.className || 'Node',
         widgetValues: sanitizeWidgetValues(node.data?.widgetValues, definition),
+        runtimeValues: {},
         definition,
         previewImage: null,
         tableRows: null,
@@ -104,13 +56,7 @@ export function hydrateWorkflowState(data, defs = {}) {
     };
   });
 
-  const nodeById = new Map(nodes.map((node) => [String(node.id), node.data]));
-
-  const edges = loadedEdges.map((edge) => ({
-    ...edge,
-    sourceHandle: remapLegacyHandle(edge.sourceHandle, 'source', nodeById.get(String(edge.source))),
-    targetHandle: remapLegacyHandle(edge.targetHandle, 'target', nodeById.get(String(edge.target))),
-  }));
+  const edges = loadedEdges.map((edge) => ({ ...edge }));
 
   const nextNodeId = Math.max(0, ...loadedNodes.map((node) => parseInt(node.id, 10) || 0)) + 1;
 
diff --git a/frontend/tests/nodeClipboard.test.mjs b/frontend/tests/nodeClipboard.test.mjs
new file mode 100644
index 0000000..7ea25b0
--- /dev/null
+++ b/frontend/tests/nodeClipboard.test.mjs
@@ -0,0 +1,179 @@
+import test from 'node:test';
+import assert from 'node:assert/strict';
+
+import {
+  buildNodeClipboardPayload,
+  instantiateNodeClipboardPayload,
+  NODE_CLIPBOARD_KIND,
+  parseNodeClipboardPayload,
+} from '../src/nodeClipboard.js';
+
+test('buildNodeClipboardPayload keeps only selected nodes and internal edges', () => {
+  const nodes = [
+    {
+      id: '1',
+      selected: true,
+      type: 'custom',
+      position: { x: 10, y: 20 },
+      data: {
+        label: 'Image',
+        className: 'Image',
+        widgetValues: { filename: 'scan.ibw' },
+        runtimeValues: { layerIndex: 2 },
+      },
+    },
+    {
+      id: '2',
+      selected: true,
+      position: { x: 100, y: 200 },
+      data: {
+        label: 'Preview',
+        className: 'Preview',
+        widgetValues: { mode: 'auto' },
+      },
+    },
+    {
+      id: '3',
+      selected: false,
+      position: { x: 500, y: 600 },
+      data: {
+        label: 'Save',
+        className: 'Save',
+      },
+    },
+  ];
+
+  const edges = [
+    {
+      id: 'e1-2',
+      source: '1',
+      sourceHandle: 'output::0::DATA_FIELD',
+      target: '2',
+      targetHandle: 'input::field::DATA_FIELD',
+      style: { stroke: '#fff', strokeWidth: 2 },
+    },
+    {
+      id: 'e2-3',
+      source: '2',
+      sourceHandle: 'output::0::IMAGE',
+      target: '3',
+      targetHandle: 'input::value::SAVE_VALUE',
+    },
+  ];
+
+  const payload = buildNodeClipboardPayload(nodes, edges);
+
+  assert.equal(payload.kind, NODE_CLIPBOARD_KIND);
+  assert.equal(payload.nodes.length, 2);
+  assert.deepEqual(payload.nodes.map((node) => node.id), ['1', '2']);
+  assert.equal(payload.edges.length, 1);
+  assert.deepEqual(payload.edges[0], {
+    source: '1',
+    sourceHandle: 'output::0::DATA_FIELD',
+    target: '2',
+    targetHandle: 'input::field::DATA_FIELD',
+    style: { stroke: '#fff', strokeWidth: 2 },
+  });
+
+  const reparsed = parseNodeClipboardPayload(JSON.stringify(payload));
+  assert.deepEqual(reparsed, payload);
+});
+
+test('instantiateNodeClipboardPayload remaps ids, offsets positions, and hydrates node shells', () => {
+  const payload = {
+    kind: NODE_CLIPBOARD_KIND,
+    version: 1,
+    nodes: [
+      {
+        id: '1',
+        position: { x: 10, y: 20 },
+        data: {
+          label: 'Image',
+          className: 'Image',
+          widgetValues: { filename: 'scan.ibw', colormap: 'viridis' },
+          runtimeValues: { layerIndex: 1 },
+        },
+      },
+      {
+        id: '2',
+        position: { x: 100, y: 200 },
+        data: {
+          label: 'Preview',
+          className: 'Preview',
+          widgetValues: { colormap: 'gray' },
+        },
+      },
+    ],
+    edges: [
+      {
+        source: '1',
+        sourceHandle: 'output::0::DATA_FIELD',
+        target: '2',
+        targetHandle: 'input::field::DATA_FIELD',
+        style: { stroke: '#abc', strokeWidth: 2 },
+      },
+    ],
+  };
+
+  const defs = {
+    Image: { output: ['DATA_FIELD'], output_name: ['field'] },
+    Preview: { output: ['IMAGE'], output_name: ['preview'] },
+  };
+
+  const instantiated = instantiateNodeClipboardPayload(payload, defs, 12, { x: 32, y: 48 });
+
+  assert.equal(instantiated.nextNodeId, 14);
+  assert.deepEqual(instantiated.nodes.map((node) => node.id), ['12', '13']);
+  assert.deepEqual(instantiated.nodes.map((node) => node.position), [
+    { x: 42, y: 68 },
+    { x: 132, y: 248 },
+  ]);
+  assert.equal(instantiated.nodes[0].selected, true);
+  assert.deepEqual(instantiated.nodes[0].data.widgetValues, { filename: 'scan.ibw', colormap: 'viridis' });
+  assert.deepEqual(instantiated.nodes[0].data.runtimeValues, { layerIndex: 1 });
+  assert.equal(instantiated.nodes[0].data.previewImage, null);
+  assert.deepEqual(instantiated.nodes[0].data.definition, defs.Image);
+
+  assert.deepEqual(instantiated.edges, [
+    {
+      id: 'e12-13-0',
+      source: '12',
+      sourceHandle: 'output::0::DATA_FIELD',
+      target: '13',
+      targetHandle: 'input::field::DATA_FIELD',
+      selected: false,
+      style: { stroke: '#abc', strokeWidth: 2 },
+    },
+  ]);
+});
+
+test('clipboard payload deep-copies local widget and runtime fields', () => {
+  const nodes = [
+    {
+      id: '9',
+      selected: true,
+      position: { x: 0, y: 0 },
+      data: {
+        label: 'Markup',
+        className: 'Markup',
+        widgetValues: {
+          stroke_width: 3,
+          markup_shapes: [
+            { kind: 'line', points: [0.1, 0.2, 0.3, 0.4] },
+          ],
+        },
+        runtimeValues: {
+          camera: { azimuth: 15, polar: 60 },
+        },
+      },
+    },
+  ];
+
+  const payload = buildNodeClipboardPayload(nodes, []);
+
+  nodes[0].data.widgetValues.markup_shapes[0].points[0] = 0.9;
+  nodes[0].data.runtimeValues.camera.azimuth = 90;
+
+  assert.equal(payload.nodes[0].data.widgetValues.markup_shapes[0].points[0], 0.1);
+  assert.equal(payload.nodes[0].data.runtimeValues.camera.azimuth, 15);
+});
diff --git a/frontend/tests/pngMetadata.test.mjs b/frontend/tests/pngMetadata.test.mjs
index 16831fe..cad6065 100644
--- a/frontend/tests/pngMetadata.test.mjs
+++ b/frontend/tests/pngMetadata.test.mjs
@@ -9,63 +9,6 @@ function makePngBlob() {
   return new Blob([Buffer.from(PNG_BASE64, 'base64')], { type: 'image/png' });
 }
 
-function crc32(bytes) {
-  const table = new Uint32Array(256);
-  for (let i = 0; i < 256; i++) {
-    let c = i;
-    for (let j = 0; j < 8; j++) {
-      c = (c & 1) ? (0xEDB88320 ^ (c >>> 1)) : (c >>> 1);
-    }
-    table[i] = c;
-  }
-
-  let crc = 0xFFFFFFFF;
-  for (let i = 0; i < bytes.length; i++) {
-    crc = table[(crc ^ bytes[i]) & 0xFF] ^ (crc >>> 8);
-  }
-  return (crc ^ 0xFFFFFFFF) >>> 0;
-}
-
-function buildChunk(type, payload) {
-  const typeBytes = new TextEncoder().encode(type);
-  const crcInput = new Uint8Array(4 + payload.length);
-  crcInput.set(typeBytes, 0);
-  crcInput.set(payload, 4);
-
-  const chunk = new Uint8Array(12 + payload.length);
-  const view = new DataView(chunk.buffer);
-  view.setUint32(0, payload.length);
-  chunk.set(typeBytes, 4);
-  chunk.set(payload, 8);
-  view.setUint32(8 + payload.length, crc32(crcInput));
-  return chunk;
-}
-
-async function insertTextChunk(blob, workflow) {
-  const png = new Uint8Array(await blob.arrayBuffer());
-  const encoder = new TextEncoder();
-  const key = encoder.encode('workflow');
-  const text = encoder.encode(JSON.stringify(workflow));
-  const payload = new Uint8Array(key.length + 1 + text.length);
-  payload.set(key, 0);
-  payload.set(text, key.length + 1);
-  const chunk = buildChunk('tEXt', payload);
-
-  let pos = 8;
-  while (pos < png.length) {
-    const len = new DataView(png.buffer, pos, 4).getUint32(0);
-    const type = String.fromCharCode(png[pos + 4], png[pos + 5], png[pos + 6], png[pos + 7]);
-    if (type === 'IEND') break;
-    pos += 12 + len;
-  }
-
-  const out = new Uint8Array(png.length + chunk.length);
-  out.set(png.subarray(0, pos), 0);
-  out.set(chunk, pos);
-  out.set(png.subarray(pos), pos + chunk.length);
-  return new Blob([out], { type: 'image/png' });
-}
-
 test('embedWorkflow roundtrips workflow data through an iTXt chunk', async () => {
   const workflow = {
     version: 1,
@@ -88,15 +31,6 @@ test('embedWorkflow roundtrips workflow data through an iTXt chunk', async () =>
   assert.match(Buffer.from(bytes).toString('latin1'), /iTXt/);
 });
 
-test('extractWorkflow still supports legacy tEXt metadata chunks', async () => {
-  const workflow = { version: 1, legacy: true, nodes: [], edges: [] };
-  const legacyBlob = await insertTextChunk(makePngBlob(), workflow);
-
-  const extracted = await extractWorkflow(legacyBlob);
-
-  assert.deepEqual(extracted, workflow);
-});
-
 test('extractWorkflow returns the last workflow chunk when an image is re-saved', async () => {
   const first = { version: 1, name: 'old', nodes: [], edges: [] };
   const second = { version: 1, name: 'new', nodes: [], edges: [] };
diff --git a/frontend/tests/workflowSerialization.test.mjs b/frontend/tests/workflowSerialization.test.mjs
index 0e0e154..0bf80e4 100644
--- a/frontend/tests/workflowSerialization.test.mjs
+++ b/frontend/tests/workflowSerialization.test.mjs
@@ -95,7 +95,7 @@ test('serializeWorkflowState keeps only stable workflow fields needed for reload
   assert.equal('selected' in serialized.edges[0], false);
 });
 
-test('hydrateWorkflowState clears shared path widgets while restoring saved dynamic outputs', () => {
+test('hydrateWorkflowState clears shared path widgets and uses registry definitions', () => {
   const saved = {
     version: 1,
     nodes: [
@@ -142,12 +142,12 @@ test('hydrateWorkflowState clears shared path widgets while restoring saved dyna
   assert.equal(hydrated.nodes[0].data.previewImage, null);
   assert.equal(hydrated.nodes[0].data.widgetValues.filename, '');
   assert.equal(hydrated.nodes[0].data.widgetValues.colormap, 'viridis');
-  assert.deepEqual(hydrated.nodes[0].data.definition.output, ['DATA_FIELD', 'DATA_FIELD']);
-  assert.deepEqual(hydrated.nodes[0].data.definition.output_name, ['Height', 'Phase']);
+  assert.deepEqual(hydrated.nodes[0].data.definition.output, ['DATA_FIELD']);
+  assert.deepEqual(hydrated.nodes[0].data.definition.output_name, ['field']);
   assert.deepEqual(hydrated.nodes[0].data.definition.input, defs.Image.input);
 });
 
-test('serializeWorkflowState and hydrateWorkflowState clear path-like widgets but preserve other metadata', () => {
+test('serializeWorkflowState and hydrateWorkflowState clear path-like widgets without restoring saved outputs', () => {
   const nodes = [
     {
       id: '7',
@@ -188,8 +188,8 @@ test('serializeWorkflowState and hydrateWorkflowState clear path-like widgets bu
   const hydrated = hydrateWorkflowState(serialized, defs);
 
   assert.deepEqual(hydrated.nodes[0].data.widgetValues, { filename: '', colormap: 'gray' });
-  assert.deepEqual(hydrated.nodes[0].data.definition.output, ['DATA_FIELD', 'DATA_FIELD', 'DATA_FIELD']);
-  assert.deepEqual(hydrated.nodes[0].data.definition.output_name, ['Topography', 'Error', 'Mask']);
+  assert.deepEqual(hydrated.nodes[0].data.definition.output, ['DATA_FIELD']);
+  assert.deepEqual(hydrated.nodes[0].data.definition.output_name, ['field']);
   assert.deepEqual(hydrated.edges, edges);
 });
 
@@ -223,6 +223,6 @@ test('hydrateWorkflowState clears saved folder selections on shared workflows',
   const hydrated = hydrateWorkflowState(saved, defs);
 
   assert.equal(hydrated.nodes[0].data.widgetValues.folder, '');
-  assert.deepEqual(hydrated.nodes[0].data.definition.output, ['PATH', 'PATH']);
-  assert.deepEqual(hydrated.nodes[0].data.definition.output_name, ['scan1.png', 'scan2.png']);
+  assert.deepEqual(hydrated.nodes[0].data.definition.output, ['PATH']);
+  assert.deepEqual(hydrated.nodes[0].data.definition.output_name, ['path']);
 });
diff --git a/tests/test_nodes.py b/tests/test_nodes.py
index 0c8b6bd..114a17e 100644
--- a/tests/test_nodes.py
+++ b/tests/test_nodes.py
@@ -1107,9 +1107,13 @@ def test_annotations():
     print("=== Test: Annotations ===")
     from backend.nodes.annotations import Annotations
     from backend.nodes.font_node import Font
+    from backend.data_types import ImageData
 
     node = Annotations()
     font_node = Font()
+    warnings = []
+    Annotations._broadcast_warning_fn = lambda nid, msg: warnings.append(msg)
+    Annotations._current_node_id = "test"
     field = DataField(
         data=np.linspace(0.0, 1.0, 64 * 64, dtype=np.float64).reshape(64, 64),
         xreal=1e-6,
@@ -1123,7 +1127,7 @@ def test_annotations():
     plain_preview = render_datafield_preview(field, "viridis")
     assert np.array_equal(plain_preview, base)
 
-    plain_field, = node.render(field, colormap="auto", show_scale_bar=False, show_color_map=False)
+    plain_field, = node.render(input=field, colormap="auto", show_scale_bar=False, show_color_map=False)
     assert isinstance(plain_field, DataField)
     assert np.array_equal(plain_field.data, field.data)
     assert plain_field.colormap == "viridis"
@@ -1132,19 +1136,19 @@ def test_annotations():
     assert plain.shape == base.shape
     assert np.array_equal(plain, base)
 
-    with_scale_field, = node.render(field, colormap="auto", show_scale_bar=True, show_color_map=False)
+    with_scale_field, = node.render(input=field, colormap="auto", show_scale_bar=True, show_color_map=False)
     with_scale = render_datafield_preview(with_scale_field, with_scale_field.colormap)
     assert with_scale.shape == base.shape
     assert not np.array_equal(with_scale, base)
 
-    with_legend_field, = node.render(field, colormap="auto", show_scale_bar=False, show_color_map=True)
+    with_legend_field, = node.render(input=field, colormap="auto", show_scale_bar=False, show_color_map=True)
     with_legend = render_datafield_preview(with_legend_field, with_legend_field.colormap)
     assert with_legend.shape[0] == base.shape[0]
     assert with_legend.shape[1] > base.shape[1]
     assert with_legend.shape[2] == 3
 
     larger_legend_field, = node.render(
-        field,
+        input=field,
         colormap="auto",
         show_scale_bar=False,
         show_color_map=True,
@@ -1156,7 +1160,7 @@ def test_annotations():
 
     annotation_font, = font_node.build("Arial")
     with_font_field, = node.render(
-        field,
+        input=field,
         colormap="auto",
         show_scale_bar=False,
         show_color_map=True,
@@ -1167,18 +1171,62 @@ def test_annotations():
     with_font = render_datafield_preview(with_font_field, with_font_field.colormap)
     assert with_font.shape == with_legend.shape
 
-    with_both_field, = node.render(field, colormap="auto", show_scale_bar=True, show_color_map=True)
+    with_both_field, = node.render(input=field, colormap="auto", show_scale_bar=True, show_color_map=True)
     with_both = render_datafield_preview(with_both_field, with_both_field.colormap)
     assert with_both.shape == with_legend.shape
     assert not np.array_equal(with_both[:, :base.shape[1]], base)
 
+    viewport_image = ImageData(
+        np.zeros((48, 64, 3), dtype=np.uint8),
+        metadata={
+            "annotation_context": {
+                "xreal": 2e-6,
+                "si_unit_xy": "m",
+                "legend_min": -1.5,
+                "legend_mid": 0.0,
+                "legend_max": 1.5,
+                "legend_unit": "V",
+                "colormap": "viridis",
+            },
+        },
+    )
+    annotated_image, = node.render(
+        input=viewport_image,
+        colormap="auto",
+        show_scale_bar=True,
+        show_color_map=True,
+        text_size=18.0,
+    )
+    assert isinstance(annotated_image, ImageData)
+    assert annotated_image.shape[0] == viewport_image.shape[0]
+    assert annotated_image.shape[1] > viewport_image.shape[1]
+    assert annotated_image.metadata["annotation_context"]["legend_unit"] == "V"
+    assert not np.array_equal(np.asarray(annotated_image)[:, :viewport_image.shape[1]], np.asarray(viewport_image))
+    assert warnings == []
+
+    plain_image = ImageData(np.zeros((32, 40, 3), dtype=np.uint8))
+    passthrough_image, = node.render(
+        input=plain_image,
+        colormap="auto",
+        show_scale_bar=True,
+        show_color_map=True,
+        text_size=18.0,
+    )
+    assert isinstance(passthrough_image, ImageData)
+    assert passthrough_image.shape == plain_image.shape
+    assert np.array_equal(np.asarray(passthrough_image), np.asarray(plain_image))
+    assert len(warnings) == 1
+    assert "no scale metadata" in warnings[0]
+
+    Annotations._broadcast_warning_fn = None
+
     print("  PASS\n")
 
 
 def test_markup():
     print("=== Test: Markup ===")
     from backend.nodes.markup import Markup
-    from backend.data_types import _preview_markup_stroke_width
+    from backend.data_types import ImageData, _preview_markup_stroke_width
 
     node = Markup()
     field = make_field(data=np.linspace(0.0, 1.0, 48 * 48, dtype=np.float64).reshape(48, 48))
@@ -1192,7 +1240,7 @@ def test_markup():
     Markup._current_node_id = "test"
 
     plain_field, = node.process(
-        field=field,
+        input=field,
         shape="line",
         stroke_color="#ffd54f",
         stroke_width=3,
@@ -1212,7 +1260,7 @@ def test_markup():
         {"kind": "arrow", "x1": 0.15, "y1": 0.85, "x2": 0.85, "y2": 0.2, "width": 4, "color": "#ffffff"},
     ])
     marked_field, = node.process(
-        field=field,
+        input=field,
         shape="arrow",
         stroke_color="#ffffff",
         stroke_width=4,
@@ -1222,6 +1270,23 @@ def test_markup():
     assert marked.shape == base.shape
     assert not np.array_equal(marked, base)
 
+    viewport_image = ImageData(
+        np.zeros((48, 48, 3), dtype=np.uint8),
+        metadata={"annotation_context": {"xreal": 1e-6, "si_unit_xy": "m"}},
+    )
+    image_markup, = node.process(
+        input=viewport_image,
+        shape="line",
+        stroke_color="#ff0000",
+        stroke_width=4,
+        markup_shapes=json.dumps([
+            {"kind": "line", "x1": 0.1, "y1": 0.2, "x2": 0.9, "y2": 0.8, "width": 4, "color": "#ff0000"},
+        ]),
+    )
+    assert isinstance(image_markup, ImageData)
+    assert image_markup.metadata["annotation_context"]["si_unit_xy"] == "m"
+    assert not np.array_equal(np.asarray(image_markup), np.asarray(viewport_image))
+
     Markup._broadcast_overlay_fn = None
     print("  PASS\n")
 
@@ -1326,6 +1391,36 @@ def test_load_file_npz():
     print("  PASS\n")
 
 
+def test_load_file_cache():
+    print("=== Test: Image cache ===")
+    from unittest.mock import patch
+    from backend.nodes.image import Image
+
+    node = Image()
+    Image._load_fields_cached.cache_clear()
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        data = np.arange(16, dtype=np.float64).reshape(4, 4)
+        path = os.path.join(tmpdir, "cached.npy")
+        np.save(path, data)
+
+        with patch.object(Image, "_load_image_or_array", wraps=Image._load_image_or_array) as loader:
+            first, = node.load(filename=path)
+            second, = node.load(filename=path)
+            assert loader.call_count == 1
+
+        assert np.allclose(first.data, data)
+        assert np.allclose(second.data, data)
+        assert first is not second
+        first.data[0, 0] = -999.0
+
+        third, = node.load(filename=path)
+        assert third.data[0, 0] == data[0, 0]
+
+    Image._load_fields_cached.cache_clear()
+    print("  PASS\n")
+
+
 def test_load_file_not_found():
     print("=== Test: Image not found ===")
     from backend.nodes.image import Image
@@ -1487,6 +1582,31 @@ def test_load_demo():
     print("  PASS\n")
 
 
+def test_load_demo_cache():
+    print("=== Test: ImageDemo cache ===")
+    from unittest.mock import patch
+    from backend.nodes.image import Image
+    from backend.nodes.image_demo import ImageDemo
+
+    node = ImageDemo()
+    Image._load_fields_cached.cache_clear()
+
+    with patch.object(Image, "_load_image_or_array", wraps=Image._load_image_or_array) as loader:
+        first, = node.load(name="nanoparticles.npy")
+        second, = node.load(name="nanoparticles.npy")
+        assert loader.call_count == 1
+
+    assert np.allclose(first.data, second.data)
+    assert first is not second
+    first.data[0, 0] = -999.0
+
+    third, = node.load(name="nanoparticles.npy")
+    assert third.data[0, 0] != -999.0
+
+    Image._load_fields_cached.cache_clear()
+    print("  PASS\n")
+
+
 def test_load_demo_multi_layer_preview_payload():
     print("=== Test: ImageDemo multi-layer preview payload ===")
     from backend.execution import ExecutionEngine
@@ -1849,6 +1969,10 @@ def test_stats():
 def test_view3d():
     print("=== Test: View3D ===")
     from backend.nodes.view_3d import View3D
+    from backend.data_types import ImageData, MeshModel
+    import base64
+    import io
+    from PIL import Image
 
     node = View3D()
     field = make_field()
@@ -1857,8 +1981,25 @@ def test_view3d():
     View3D._broadcast_mesh_fn = lambda nid, mesh: captured.append(mesh)
     View3D._current_node_id = "test"
 
-    result = node.render(field, colormap="viridis", z_scale=2.0, resolution=64)
-    assert result == ()
+    preview_image = Image.new("RGB", (12, 10), (255, 0, 0))
+    preview_buffer = io.BytesIO()
+    preview_image.save(preview_buffer, format="PNG")
+    viewport_snapshot = "data:image/png;base64," + base64.b64encode(preview_buffer.getvalue()).decode()
+
+    result = node.render(
+        field,
+        colormap="viridis",
+        z_scale=2.0,
+        resolution=64,
+        make_solid=False,
+        viewport_snapshot=viewport_snapshot,
+    )
+    assert len(result) == 2
+    assert isinstance(result[0], MeshModel)
+    assert isinstance(result[1], ImageData)
+    assert result[1].shape == (10, 12, 3)
+    assert np.all(result[1][0, 0] == np.array([255, 0, 0], dtype=np.uint8))
+    assert result[1].metadata["annotation_context"]["si_unit_xy"] == field.si_unit_xy
     assert len(captured) == 1
 
     mesh = captured[0]
@@ -1873,7 +2014,6 @@ def test_view3d():
     assert mesh["z_min"] < mesh["z_max"]
 
     # Verify base64 data can be decoded
-    import base64
     z_bytes = base64.b64decode(mesh["z_data"])
     assert len(z_bytes) == mesh["width"] * mesh["height"] * 4  # float32
 
@@ -1883,14 +2023,178 @@ def test_view3d():
     # High-res input should be downsampled
     big_field = make_field(shape=(256, 256))
     captured.clear()
-    node.render(big_field, colormap="hot", z_scale=1.0, resolution=64)
+    node.render(big_field, colormap="hot", z_scale=1.0, resolution=64, make_solid=False)
     assert captured[0]["width"] <= 64
     assert captured[0]["height"] <= 64
 
+    # Separate map input should affect colors without changing mesh geometry
+    mesh_field = make_field(data=np.zeros((64, 64), dtype=np.float64), xreal=2.0, yreal=3.0)
+    map_field = make_field(data=np.tile(np.linspace(0.0, 1.0, 64, dtype=np.float64), (64, 1)), xreal=2.0, yreal=3.0)
+    captured.clear()
+    mapped_result = node.render(mesh_field, map_field=map_field, colormap="viridis", z_scale=1.0, resolution=32, make_solid=False)
+    mapped_mesh = captured[0]
+    assert mapped_mesh["x_range"] == [float(mesh_field.xoff), float(mesh_field.xoff + mesh_field.xreal)]
+    assert mapped_mesh["y_range"] == [float(mesh_field.yoff), float(mesh_field.yoff + mesh_field.yreal)]
+    mapped_z = np.frombuffer(base64.b64decode(mapped_mesh["z_data"]), dtype=np.float32)
+    assert np.allclose(mapped_z, 0.0)
+    mapped_colors = np.frombuffer(base64.b64decode(mapped_mesh["colors"]), dtype=np.uint8)
+
+    captured.clear()
+    node.render(mesh_field, colormap="viridis", z_scale=1.0, resolution=32, make_solid=False)
+    mesh_only = captured[0]
+    mesh_only_colors = np.frombuffer(base64.b64decode(mesh_only["colors"]), dtype=np.uint8)
+    assert not np.array_equal(mapped_colors, mesh_only_colors)
+
+    # make_solid should add extra geometry beyond the top surface grid
+    solid_mesh = mapped_result[0]
+    assert isinstance(solid_mesh, MeshModel)
+    captured.clear()
+    solid_result = node.render(mesh_field, colormap="viridis", z_scale=1.0, resolution=16, make_solid=True)
+    assert len(solid_result[0].vertices) > 16 * 16
+    assert len(solid_result[0].faces) > (15 * 15 * 2)
+    solid_payload = captured[0]
+    assert solid_payload["make_solid"] is True
+    assert "positions" in solid_payload
+    assert "indices" in solid_payload
+    assert "vertex_colors" in solid_payload
+
     View3D._broadcast_mesh_fn = None
     print("  PASS\n")
 
 
+def test_save_generic():
+    print("=== Test: Save ===")
+    from backend.nodes.save import Save
+    from backend.data_types import DataField, LineData, MeasureTable, MeshModel, RecordTable
+    import tifffile
+    from PIL import Image as PILImage
+
+    node = Save()
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        # Save scalar as TXT and JSON
+        node.save(filename="scalar", directory_path=tmpdir, format="TXT", value=3.5)
+        assert Path(tmpdir, "scalar.txt").read_text(encoding="utf-8").strip() == "3.5"
+        node.save(filename="scalar_json", directory_path=tmpdir, format="JSON", value=3.5)
+        assert json.loads(Path(tmpdir, "scalar_json.json").read_text(encoding="utf-8")) == {"value": 3.5}
+
+        # Save line as CSV, NPZ, and JSON
+        line = LineData(data=np.array([1.0, 2.0, 3.0]), x_axis=np.array([0.0, 0.5, 1.0]), x_unit="um", y_unit="nm")
+        node.save(filename="profile", directory_path=tmpdir, format="CSV", value=line)
+        csv_text = Path(tmpdir, "profile.csv").read_text(encoding="utf-8")
+        assert "x,y,x_unit,y_unit" in csv_text
+        assert "um" in csv_text and "nm" in csv_text
+        node.save(filename="profile_npz", directory_path=tmpdir, format="NPZ", value=line)
+        line_npz = np.load(Path(tmpdir, "profile_npz.npz"))
+        assert np.allclose(line_npz["x"], line.x_axis)
+        assert np.allclose(line_npz["y"], line.data)
+        node.save(filename="profile_json", directory_path=tmpdir, format="JSON", value=line)
+        line_json = json.loads(Path(tmpdir, "profile_json.json").read_text(encoding="utf-8"))
+        assert line_json["x_unit"] == "um"
+        assert line_json["y_unit"] == "nm"
+        assert line_json["x"] == [0.0, 0.5, 1.0]
+        assert line_json["y"] == [1.0, 2.0, 3.0]
+
+        # Save DATA_FIELD as TIFF, PNG, and NPZ
+        field = DataField(
+            data=np.array([[1.0, 2.0], [3.0, 4.5]], dtype=np.float64),
+            xreal=2e-6,
+            yreal=1e-6,
+            si_unit_xy="m",
+            si_unit_z="m",
+            colormap="viridis",
+        )
+        node.save(filename="field_tiff", directory_path=tmpdir, format="TIFF", value=field)
+        field_tiff = tifffile.imread(Path(tmpdir, "field_tiff.tiff"))
+        assert field_tiff.shape == field.data.shape
+        assert field_tiff.dtype == np.float32
+        assert np.allclose(field_tiff, field.data.astype(np.float32))
+
+        node.save(filename="field_png", directory_path=tmpdir, format="PNG", value=field)
+        field_png = np.asarray(PILImage.open(Path(tmpdir, "field_png.png")))
+        assert field_png.shape == (2, 2, 3)
+        assert field_png.dtype == np.uint8
+
+        node.save(filename="field_npz", directory_path=tmpdir, format="NPZ", value=field)
+        field_npz = np.load(Path(tmpdir, "field_npz.npz"))
+        assert np.allclose(field_npz["field"], field.data)
+
+        # Save IMAGE as PNG, TIFF, and NPZ
+        image = np.array(
+            [
+                [[255, 0, 0], [0, 255, 0]],
+                [[0, 0, 255], [255, 255, 0]],
+            ],
+            dtype=np.uint8,
+        )
+        node.save(filename="image_png", directory_path=tmpdir, format="PNG", value=image)
+        image_png = np.asarray(PILImage.open(Path(tmpdir, "image_png.png")))
+        assert image_png.shape == image.shape
+        assert np.array_equal(image_png, image)
+
+        node.save(filename="image_tiff", directory_path=tmpdir, format="TIFF", value=image)
+        image_tiff = tifffile.imread(Path(tmpdir, "image_tiff.tiff"))
+        assert image_tiff.shape == image.shape
+        assert image_tiff.dtype == np.uint8
+        assert np.array_equal(image_tiff, image)
+
+        node.save(filename="image_npz", directory_path=tmpdir, format="NPZ", value=image)
+        image_npz = np.load(Path(tmpdir, "image_npz.npz"))
+        assert np.array_equal(image_npz["image"], image)
+
+        # Save tables as CSV and JSON
+        measure_table = MeasureTable([
+            {"quantity": "Rq", "value": 1.23, "unit": "nm"},
+            {"quantity": "Ra", "value": 0.98, "unit": "nm"},
+        ])
+        node.save(filename="measurements_csv", directory_path=tmpdir, format="CSV", value=measure_table)
+        measure_csv = Path(tmpdir, "measurements_csv.csv").read_text(encoding="utf-8")
+        assert "quantity,value,unit" in measure_csv
+        assert "Rq,1.23,nm" in measure_csv
+        node.save(filename="measurements_json", directory_path=tmpdir, format="JSON", value=measure_table)
+        assert json.loads(Path(tmpdir, "measurements_json.json").read_text(encoding="utf-8")) == list(measure_table)
+
+        record_table = RecordTable([
+            {"label": "particle-1", "height": 12.0, "area": 44.0},
+            {"label": "particle-2", "height": 8.0, "area": 21.0},
+        ])
+        node.save(filename="records_csv", directory_path=tmpdir, format="CSV", value=record_table)
+        record_csv = Path(tmpdir, "records_csv.csv").read_text(encoding="utf-8")
+        assert "label,height,area" in record_csv
+        assert "particle-1,12.0,44.0" in record_csv
+        node.save(filename="records_json", directory_path=tmpdir, format="JSON", value=record_table)
+        assert json.loads(Path(tmpdir, "records_json.json").read_text(encoding="utf-8")) == list(record_table)
+
+        # Save mesh as OBJ and STL
+        mesh = MeshModel(
+            vertices=np.array([[0, 0, 0], [1, 0, 0], [0, 1, 0]], dtype=np.float32),
+            faces=np.array([[0, 1, 2]], dtype=np.int32),
+        )
+        node.save(filename="triangle", directory_path=tmpdir, format="OBJ", value=mesh)
+        obj_text = Path(tmpdir, "triangle.obj").read_text(encoding="utf-8")
+        assert "v 0.0 0.0 0.0" in obj_text
+        assert "f 1 2 3" in obj_text
+
+        node.save(filename="triangle", directory_path=tmpdir, format="STL", value=mesh)
+        stl_text = Path(tmpdir, "triangle.stl").read_text(encoding="utf-8")
+        assert stl_text.startswith("solid argonode")
+        assert "facet normal" in stl_text
+
+        try:
+            node.save(filename="triangle", directory_path=tmpdir, format="PNG", value=mesh)
+            assert False, "Mesh should only be saveable as OBJ or STL"
+        except ValueError:
+            pass
+
+        try:
+            node.save(filename="field_bad", directory_path=tmpdir, format="CSV", value=field)
+            assert False, "DATA_FIELD should reject unsupported save formats"
+        except ValueError:
+            pass
+
+    print("  PASS\n")
+
+
 # =========================================================================
 # Run all tests
 # =========================================================================
@@ -1940,6 +2244,7 @@ if __name__ == "__main__":
     test_load_demo()
     test_coordinate()
     test_range_slider()
+    test_save_generic()
     test_save_image()
 
     # Display
diff --git a/tests/test_numpy_compat.py b/tests/test_numpy_compat.py
deleted file mode 100644
index a13faa1..0000000
--- a/tests/test_numpy_compat.py
+++ /dev/null
@@ -1,8 +0,0 @@
-import backend  # noqa: F401
-import numpy as np
-
-
-def test_numpy_compat_aliases_are_available_after_backend_import():
-    assert np.complex is complex
-    assert np.float is float
-    assert np.int is int