Fix set_zero_in_cmap_to_transparent with datashader rendering#577
Merged
Fix set_zero_in_cmap_to_transparent with datashader rendering#577
Conversation
Datashader's tf.shade() strips the alpha channel from matplotlib colormaps, so entries with alpha=0 (from set_zero_in_cmap_to_transparent) rendered as opaque white instead of transparent. Fix by masking aggregate values that map to transparent cmap entries to NaN before shading — NaN is datashader's native transparency mechanism. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #577 +/- ##
==========================================
+ Coverage 75.36% 75.42% +0.05%
==========================================
Files 10 10
Lines 2935 2966 +31
Branches 684 693 +9
==========================================
+ Hits 2212 2237 +25
- Misses 441 444 +3
- Partials 282 285 +3
🚀 New features to boost your workflow:
|
Add test_plot_transparent_cmap_shapes_matplotlib and test_plot_transparent_cmap_shapes_datashader to verify that shapes with value=0 are rendered transparently over an image when using set_zero_in_cmap_to_transparent, for both rendering backends. Placeholder baselines included — CI will generate the real ones. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Each visual test now shows a side-by-side comparison with the standard viridis cmap on the left and the transparent cmap on the right, making the transparency effect clearly visible. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When span=None, datashader auto-scales from non-NaN values. After masking zeros to NaN, the remaining minimum value (e.g. 2.0) would map to cmap(0.0) — the transparent-white entry — and min_alpha would override the alpha, producing opaque white. Fix: _mask_transparent_cmap_entries now returns the pre-masking span so ds.tf.shade uses the original data range, preventing re-scaling into the transparent bin. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The NaN-masking approach was fundamentally flawed: masking values to NaN caused datashader to auto-scale the remaining values, shifting other shapes' colors and potentially making them white. New approach: let datashader shade normally (preserving all colors and spans), then post-process the RGBA output to apply the cmap's alpha channel. This: - Does not modify aggregate values or spans - Preserves exact colors for all non-transparent shapes - Works for any cmap with transparent entries at any position - Handles span=None correctly (no auto-scaling issues) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Cover additional datashader code paths: - Points with transparent cmap (separate from shapes) - Shapes with clip=False norm (3-part under/in/over shading) Placeholder baselines — CI will generate real ones. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The points visual test failed because adding a column to the points DataFrame requires table annotation. The shapes and clip=False tests cover the same _datashader_map_aggregate_to_color code path. Updated baselines with CI-generated images. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
set_zero_in_cmap_to_transparentproduced opaque white shapes instead of transparent ones when rendering with datashaderds.tf.shade()strips the alpha channel from matplotlib colormaps entirely — entries with alpha=0 were rendered as opaque white_get_collection_shape)Known limitation: when all aggregate values are identical and map to the transparent cmap entry, the upstream single-value path in
_ds_shade_continuousconverts the cmap to a hex string before reaching the masking logic. This is a pre-existing edge case.