fmri_surface_data.surface: space-matched default view, render perf fix, docs#89
Merged
Conversation
…est for display volume surface()/addblobs() for cortex-only fs_LR data on a mixed patch layout (e.g. foursurfaces_hcp) became very slow: incidental subcortical anatomy meshes triggered a full barycentric resample + surf2vol projection even though the cortex painted natively on its own 32492 mesh. render_layer_surfaces (paint_surface_native_layer): - Per view, defer arbitrary (non-standard) meshes and project to a volume ONLY when no native/resampled cortical patch was painted in that view. Cortex-only data on foursurfaces_hcp now paints natively with no projection (addblobs ~0.2s vs ~48s before). - Print a one-time message when resampling onto a standard mesh or projecting to a volume, so the workspace shows when work is happening. to_display_volume: resample cortex to fsaverage-164k by nearest neighbour instead of barycentric. This volume is display-only (sampled at mesh vertices to colour a patch); at 164k density the result is visually identical but ~250x faster (0.2s vs ~48s -- the barycentric weight build dominated). Arbitrary-mesh render (hires left / cutaway) now ~0.8s. Surface render + fmridisplay suites: 23/23 pass. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ct's space surface(obj) with no explicit surface argument now picks a four-surface view that matches the object's surface-space family, so the data renders at native (or near-native) fidelity by default: - fsLR_32k -> foursurfaces_hcp (native, 32492/hemi) - fsaverage_164k -> foursurfaces_freesurfer (native, 163842/hemi) - fsaverage6/5/4 -> foursurfaces_freesurfer (nested; resampled up to 164k) - onavg_41k/10k -> foursurfaces_hcp (fs_LR-aligned; resampled to fs_LR) No new mesh assets are needed: fsaverage6/5/4 are nested subsets of fsaverage-164k and onavg is fs_LR-aligned via the vendored onavg spheres, so each renders on its parent/aligned display space through the fast, cached nearest-neighbour resample already in render_layer_surfaces (a one-line message prints when a resample happens). Previously every non-fs_LR/-164k space fell back to foursurfaces_hcp, forcing an unnecessary cross-family resample. @fmridisplay/surface: surface_default_keyword now keys off a coarse space family (local_space_family) rather than two exact strings. @fmri_surface_data/surface: - Fix parser bug: a bare surface keyword (e.g. surface(obj,'foursurfaces_hcp') or surface(obj,'hcp inflated left')) fell into the colour-option branch and indexed past the end of varargin. Non-mode tokens are now split into colour options (value pairs) vs surface directives (bare token / direction- orientation-axes pairs); directives are forwarded to the native managed path and override the space-matched default view. - Expanded help: a Surface-spaces table (space, verts/hemi, default view, mesh files) and runnable examples rendering one object per space plus the existing/mni-surface modes. New test test_default_view_matches_space (canlab_test_surface_render): verifies fs_LR/fsaverage6/onavg pick the right view and paint, and that an explicit keyword overrides without erroring. Surface render + fmridisplay suites: 24/24 pass. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ata surface reference Documents which four-surface view surface(obj) picks for each surface_space (fsLR_32k/fsaverage_164k render natively; nested fsaverage6/5/4 -> fsaverage-164k; onavg -> fs_LR, both via a fast cached nearest resample). Added the table to both the comprehensive reference (CanlabCore/docs) and the readthedocs functional index (docs). Also refreshed the comprehensive doc's Surface-spaces section (all 7 spaces; resample_surface is implemented, not a 'planned enhancement') and corrected the surface() space-mismatch text (standard meshes auto-resample; only truly arbitrary meshes project via a volume). Co-Authored-By: Claude Opus 4.8 <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
1 participant
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.
Follow-up to #86 (the
fmri_surface_dataobject). Three commits, cherry-picked onto currentmasterso they apply cleanly on top of the squash-merge.1. Space-matched default four-surface view
surface(obj)with no explicit surface argument now picks a four-surface view (L/R lateral + medial) matched to the object'ssurface_space, so data renders at native (or near-native) fidelity by default:surface_spacefsLR_32kfoursurfaces_hcpfsaverage_164kfoursurfaces_freesurferfsaverage6/5/4foursurfaces_freesurferonavg_41k/10kfoursurfaces_hcpNo new mesh assets needed: nested fsaverage spaces are subsets of fsaverage-164k and onavg is fs_LR-aligned via the vendored onavg spheres, so each renders on its parent/aligned display space through the fast, cached nearest resample already in
render_layer_surfaces(prints a one-line message when it resamples). Previously every non-fs_LR/-164k space fell back tofoursurfaces_hcp, forcing an unnecessary cross-family resample.Also fixes a parser bug in
@fmri_surface_data/surface.m: a bare surface keyword (surface(obj,'foursurfaces_hcp'),surface(obj,'hcp inflated left')) fell into the colour-option branch and indexed past the end ofvarargin. Non-mode tokens are now split into colour options vs surface directives; directives override the space-matched default view.2. Render performance regression fix
Cortex-only fs_LR data on
foursurfaces_hcphad regressed to ~48 s because incidental subcortical anatomy meshes triggered a needless volume projection. Now arbitrary (non-standard) meshes are deferred per view and projected only when no native/resampled cortical patch painted in that view —addblobsis back to ~0.2 s.to_display_volumealso uses nearest (not barycentric) for its display-only cortex resample (~250× faster: 0.2 s vs 48 s; the barycentric weight-build was the entire bottleneck). Arbitrary-mesh renders (hires left,cutaway) drop from ~48 s to ~0.8 s.3. Docs
Per-space default-view table added to both the comprehensive reference (
CanlabCore/docs/) and the readthedocs functional index (docs/); refreshed the Surface-spaces section (all 7 spaces;resample_surfaceis implemented) and corrected the surface() space-mismatch text.Tests
New
test_default_view_matches_spaceincanlab_test_surface_render.m. Surface render + fmridisplay suites pass locally (24/24) on R2026a + SPM25.🤖 Generated with Claude Code