Add AVIF image support via libavif#289
Merged
Merged
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
📝 WalkthroughSummary by CodeRabbit
WalkthroughAdds end-to-end AVIF support: CI/CMake/vcpkg updates, llimage AVIF codec (encode/decode/update), dimension extraction, snapshot/live-preview and UI wiring, file picker filters, and viewer upload/save integration. ChangesAVIF Image Format Support
Sequence Diagram(s)sequenceDiagram
participant UI as Snapshot/UI
participant Factory as LLImageFormatted Factory
participant LLImageAVIF
participant LibAVIF
participant Storage as File/Buffer
UI->>Factory: request formatted image (AVIF)
Factory->>LLImageAVIF: createFromType(IMG_CODEC_AVIF)
LLImageAVIF->>LibAVIF: avifDecoderCreate / avifPeekCompatibleFileType
LibAVIF-->>LLImageAVIF: metadata (width,height,alpha)
UI->>LLImageAVIF: decode() / encode()
LLImageAVIF->>LibAVIF: avifDecoderNextImage / avifImageYUVToRGB / avifImageRGBToYUV / avifEncoderWrite
LibAVIF-->>LLImageAVIF: image frames / bitstream
LLImageAVIF->>LLImageAVIF: vertical flip (in-place)
LLImageAVIF->>Storage: store encoded bitstream / fill raw buffer
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Warning There were issues while running some tools. Please review the errors and either fix the tool's configuration or disable the tool if it's a critical failure. 🔧 Infer (1.2.0)indra/llimage/llimageavif.cppindra/llimage/llimageavif.cpp:26:10: fatal error: 'linden_common.h' file not found ... [truncated 1081 characters] ... stall/lib/clang/18/include" Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
RyeMutt
force-pushed
the
feature/avif-support
branch
3 times, most recently
from
4d3a08b to
f153985
Compare
Add an AVIF codec (LLImageAVIF) to llimage backed by libavif with dav1d (decode) and aom (encode), and wire it into the image upload and save/snapshot paths, mirroring the existing WebP integration. - llimage: IMG_CODEC_AVIF and LLImageAVIF (lossy quality-based encode, 8-bit RGB/RGBA decode with vertical flip); extension, MIME, and dimensions-info registration - build: libavif[aom,dav1d] vcpkg dependency and AVIF.cmake module - upload: file picker load filters, IMAGE_EXTENSIONS, local bitmaps - save: FFSAVE_AVIF, snapshot format (combo box, encode switches, quality slider) and texture save-as Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
RyeMutt
force-pushed
the
feature/avif-support
branch
from
f153985 to
fc56e08
Compare
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@indra/llimage/llimageavif.cpp`:
- Around line 214-257: The AVIF encoder currently assumes non-4-component images
are RGB which breaks for 1-channel grayscale; update the logic after
avifRGBImageSetDefaults(&rgb, image) to handle components explicitly: set
rgb.format = AVIF_RGB_FORMAT_RGBA for components==4, rgb.format =
AVIF_RGB_FORMAT_GRAY (or the correct libavif constant for 1-channel gray) for
components==1 and keep rgb.depth/rowBytes as width*components, and for any other
components value (e.g., 2) call setLastError("LLImageAVIF::Unsupported component
count") and return false so we fail fast; ensure the change references
rgb.format, rgb.rowBytes, avifRGBImageSetDefaults, and setLastError.
In `@indra/llimage/llimageavif.h`:
- Around line 37-48: The header comment and getters/setters for LLImageAVIF
incorrectly describe quality=100 as "lossless"; update the wording to "best
quality" (or similar) instead of "lossless" in LLImageAVIF, mEncodeQuality,
setEncodeQuality, and getEncodeQuality declarations and the mEncodeQuality
comment to avoid implying a true lossless encode (since encode() currently uses
AVIF_PIXEL_FORMAT_YUV420 and encoder->quality/qualityAlpha). Alternatively, if
you want to support true lossless, implement a lossless encode path in
LLImageAVIF::encode() (select a lossless pixel format and set the encoder to
lossless mode) and adjust the header docs accordingly; otherwise change all
occurrences of "lossless" to "best quality" to reflect current behavior.
In `@indra/llimage/llimagedimensionsinfo.cpp`:
- Around line 203-212: The getImageDimensionsAVIF implementation currently reads
the entire file into memory (using LLFile::size, std::make_unique<U8[]>, and
mInfile.read) which breaks the lightweight dimensions-only contract; change
LLImageDimensionsInfo::getImageDimensionsAVIF to perform bounded metadata
parsing using file-backed IO instead of full-file buffering: open/seek/read only
the header/metadata regions needed (using mInfile and mSrcFilename) and feed
those limited buffers to the AVIF probe routine (or use a stream/reader API from
the AVIF parser that accepts a file callback), and apply the same bounded-read
approach to the related code in the 224-240 region so you never allocate the
full image in memory.
In `@indra/newview/llfilepicker.cpp`:
- Line 56: The material-texture aggregate filter constant
FFLOAD_MATERIAL_TEXTURE is missing the *.avif pattern so AVIF files are hidden
in the default material/texture picker; update the filter string(s) that build
FFLOAD_MATERIAL_TEXTURE (and the other occurrence noted) to include *.avif in
both the human-readable description and the semicolon-separated pattern list
(e.g., add ;*.avif to the pattern section and add .avif to the parenthesis list
in the display text) so AVIF files appear in the aggregate material-texture
picker on Windows/SDL.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: ef4019f1-ece5-48c2-a553-72544894e74d
📒 Files selected for processing (24)
.github/workflows/build.yamlindra/CMakeLists.txtindra/cmake/AVIF.cmakeindra/cmake/CMakeLists.txtindra/llimage/CMakeLists.txtindra/llimage/llimage.cppindra/llimage/llimage.hindra/llimage/llimageavif.cppindra/llimage/llimageavif.hindra/llimage/llimagedimensionsinfo.cppindra/llimage/llimagedimensionsinfo.hindra/newview/llfilepicker.cppindra/newview/llfilepicker.hindra/newview/lllocalbitmaps.cppindra/newview/lllocalbitmaps.hindra/newview/llpanelsnapshotlocal.cppindra/newview/llpreviewtexture.cppindra/newview/llsnapshotlivepreview.cppindra/newview/llsnapshotmodel.hindra/newview/llviewermenufile.cppindra/newview/llviewerwindow.cppindra/newview/skins/default/xui/en/panel_snapshot_local.xmlindra/newview/skins/default/xui/en/strings.xmlindra/vcpkg.json
LLImageAVIF::encode() now produces mathematically lossless output at quality 100 (identity matrix + YUV 4:4:4); the lossy path keeps 4:2:0 with a BT.601 matrix. Expose a distinct "AVIF (Lossless)" snapshot format alongside the lossy "AVIF" option, mirroring the WebP (Lossless) treatment: - SNAPSHOT_FORMAT_AVIF_LOSSLESS enum and combo box item - encode, size-estimate, and save-dispatch switches - no quality slider for the lossless variant Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Encode 1- and 2-component raw images as monochrome (4:0:0) AVIF via AVIF_RGB_FORMAT_GRAY / AVIF_RGB_FORMAT_GRAYA, selecting the pixel and RGB format by component count instead of assuming RGB/RGBA. Monochrome uses a BT.601 matrix (identity requires three planes); luma maps directly to Y so quality 100 remains lossless. 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.
Description
Adds support for the AVIF (AV1 Image File Format) image format, letting users upload
.avifimages as texture sources and save snapshots/textures as.avif. AVIF offers substantially better compression than JPEG/PNG/WebP at equivalent quality, with alpha support.A new
LLImageAVIFcodec is added tollimage, backed by libavif via vcpkg using dav1d (decode) and aom (encode). It mirrors the existing WebP integration point-for-point. Encoding is lossy with a quality control (libavif'squalityAPI, likeLLImageJPEG), since lossy is where AVIF's size advantage lies.Core (
llimage)IMG_CODEC_AVIF+LLImageAVIF: dav1d decode to 8-bit RGB/RGBA (with vertical flip for SL's OpenGL convention), aom lossy/quality encode (sRGB CICP, YUV420)image/avif), andLLImageDimensionsInforegistrationBuild
libavifvcpkg dependency withaom+dav1dfeatures; newindra/cmake/AVIF.cmake. Resolves from the existing pinned vcpkg baseline (libavif 1.4.1) — no custom port needed.Upload / load paths
IMAGE_EXTENSIONS, and local-bitmap loading. Factory-driven paths (upload preview,createUploadFile, bulk upload) work automatically via codec registration.Save / snapshot paths
FFSAVE_AVIFsave filters (all platforms),SNAPSHOT_FORMAT_AVIF, snapshot format combo box + encode switches + quality-slider visibility, viewer-window save dispatch, and texture "Save As".Related Issues
No tracking issue yet — happy to link one if desired.
Issue Link: N/A
Checklist
Please ensure the following before requesting review:
Additional Notes
Verification: Full viewer builds clean (
cmake --preset vs2026-os→--build --config Release, exit 0). vcpkg builtlibavif[aom,core,dav1d]@1.4.1;llimageavif.cppcompiles against the real libavif headers; all changednewviewTUs recompile andAlchemyTest.exelinks and packages. All pre-commit hooks pass.Design decisions:
libaviffeatures.MATERIAL_TEXTURES_FILTER, the GLTF-import filter, and the FFLOAD_ALL/EXE "Images" hint untouched (WebP isn't in them either); localized snapshot XUI needs no edits (name-merge surfaces theenAVIF combo item with its acronym label).Known limitation: HDR / >8-bit / wide-gamut AVIFs are down-converted to 8-bit sRGB on decode (LLImageRaw is 8-bit).