low pri features

docs/nodes/Super Resolution.md

@@ -0,0 +1,32 @@
+# Super Resolution
+
+Combine multiple aligned scans to produce a super-resolved image with higher spatial resolution. Sub-pixel shifts between inputs are estimated via FFT cross-correlation and used to reconstruct a finer grid. Equivalent to Gwyddion's superresolution.c.
+
+## Inputs
+
+| Name | Type | Required | Description |
+|------|------|----------|-------------|
+| field1 | DATA_FIELD | Yes | Reference image |
+| field2 | DATA_FIELD | No | Second image |
+| field3 | DATA_FIELD | No | Third image |
+| field4 | DATA_FIELD | No | Fourth image |
+
+## Outputs
+
+| Name | Type | Description |
+|------|------|-------------|
+| result | DATA_FIELD | Super-resolved image |
+
+## Controls
+
+| Name | Type | Default | Description |
+|------|------|---------|-------------|
+| upscale | INT | 2 | Upscaling factor (2, 3, or 4) |
+
+## Notes
+
+- When only one field is provided the image is upsampled using cubic interpolation (no multi-image enhancement).
+- Additional fields are aligned to the reference using FFT-based cross-correlation with parabolic sub-pixel refinement, then averaged on the high-resolution grid.
+- Providing more images generally improves the result because each scan samples slightly different sub-pixel positions.
+- All input fields should have the same pixel dimensions and physical size for correct alignment.
+- The upscale factor controls the output resolution multiplier (2x, 3x, or 4x the input dimensions).