PaddleOCR prints download/init messages to stdout which corrupts the
JSON result that the bridge expects. Same risk with basicsr/realesrgan.
Applied the same fd-level stdout redirect pattern already used in
remove_bg.py: redirect fd 1 to stderr during ML work, restore for
the JSON result. Also added show_log=False to PaddleOCR constructor.
node --import tsx requires tsx to be directly in node_modules/, but
pnpm hoists it differently. npx tsx works because it resolves through
pnpm's bin links. tini as PID 1 handles signal forwarding regardless.
basicsr has a known torchvision.transforms.functional_tensor compat
issue on arm64 with newer torchvision. On arm64, upscale.py falls back
to Lanczos via ImportError anyway. Smoke test still verifies the model
weights file exists on all platforms.
PaddleOCR uses its own language codes (ch, japan, korean, latin) not
ISO codes (zh, ja, ko, de, fr, es). The download script and ocr.py
now map API language codes to PaddleOCR codes correctly. German,
French, and Spanish all use the "latin" script model.
Rewrite docker-tags.md for single image with GPU auto-detection.
Update deployment.md to remove variant table and lite/cuda references.
Replace LaMa Cleaner references with OpenCV in architecture and AI docs.
Add migration notes for users on :lite and :cuda tags.
Remove 3-variant matrix (full/lite/cuda). Single build produces
a multi-arch manifest (amd64 + arm64) pushed to Docker Hub and GHCR.
Tags: latest, X.Y.Z, X.Y, X. CI builds native platform only (amd64)
for speed. Multi-arch only on release.
- Add 8s shutdown timeout to prevent indefinite hang when app.close()
stalls. Stays under Docker's default 10s stop_grace_period.
- Health endpoint now checks database connectivity, returns 503 when
DB is unreachable so Docker marks container unhealthy.
- Removed variant field from health response (single image now).
GPU is activated at runtime via --gpus all, not a separate compose file.
Added log rotation (10MB x 3 files) to prevent disk fill on long-running
instances. Removed docker-compose.gpu.yml.
- Remove VARIANT/GPU build args, single image for all platforms
- amd64: nvidia/cuda base with GPU Python packages
- arm64: node base with CPU Python packages
- Add tini as PID 1 for proper signal handling
- Replace npx tsx with node --import tsx
- Split pip install into base + tool layers for better caching
- Add NVIDIA_VISIBLE_DEVICES env vars for container toolkit
- Suppress Python ML library log noise
- Increase healthcheck start-period to 60s
- Remove STIRLING_VARIANT env var
- Remove lama-cleaner from pip installs
Downloads all rembg models (6), RealESRGAN_x4plus.pth weights,
PaddleOCR models for all 7 supported languages, verifies MediaPipe
bundles its face detection models. Runs a final smoke test importing
every ML library. Any failure exits non-zero, failing the Docker build.
lama-cleaner is pip-installed but never imported in any Python script.
inpaint.py uses OpenCV TELEA. Removing saves ~100+ MB of image size.
Also added seam-carving to requirements-gpu.txt where it was missing.
model_path was None, so the model had random weights and always fell back
to Lanczos. Now loads RealESRGAN_x4plus.pth from /opt/models/realesrgan/
(configurable via REALESRGAN_MODEL_PATH env var). Only falls back to
Lanczos on ImportError, not blanket Exception.
User-uploaded files were stored in /app/data/files (container writable layer)
instead of /data/files (persistent volume) because the env var was not set in
the Dockerfile. Files were lost on container recreation.
Replace static llms.txt and llms-full.txt with auto-generated versions
that stay in sync with docs on every build. The plugin also generates
per-page .md files for individual page fetching by LLMs.
New tool for joining images horizontally or vertically,
distinct from the grid-based collage tool. Preserves aspect
ratios with fit/original resize modes, optional gap, and
multi-format output.
Users running lite mode had no way to tell why AI tools were greyed out.
Now the public health endpoint reports the variant, and a visible banner
appears in the tool panel when running in lite mode.
Prevent silent data corruption when the API is called directly
(bypassing UI guards). Binary/complex EXIF fields like MakerNote
are now filtered from fieldsToRemove in the image-engine operation.
Move sanitizeValue, parseExif, parseGps, parseXmp into the shared
image-engine package so both strip-metadata and edit-metadata can
reuse them. Includes 13 unit tests covering all four functions.
Design spec for new edit-metadata tool (issue #15). Covers common EXIF
field editing, GPS clearing, granular per-field stripping, and shared
metadata infrastructure extracted from strip-metadata.
Clickable localhost:1349 links are misleading on the GitHub Pages site
since users may not have Stirling Image running locally. Use plain path
references instead so it is clear these live on their own instance.
The links used /Stirling-Image/ (wrong casing) which VitePress then
prefixed with the base /stirling-image/, producing a double-prefixed
404 path. Remove the manual base so VitePress prepends it automatically.
Keep the main docker run command front and center. Lite and CUDA
variants are in a collapsible details block so the quick start
section stays scannable.
onnxruntime-gpu reports CUDAExecutionProvider as "available" just
because the library was compiled with CUDA support, even on machines
with no GPU. This made gpu_available() return True incorrectly,
causing upscale.py to try torch.device("cuda") and fall back to
Lanczos instead of running Real-ESRGAN on CPU.
torch.cuda.is_available() actually probes the hardware. Use it as
the single source of truth for GPU detection.
Verified: CUDA image on Apple Silicon (no GPU) now correctly reports
gpu: false and all AI tools run on CPU without crashes.
The STIRLING_GPU=true env var was baked into the :cuda Dockerfile,
which made gpu_available() return True without checking actual
hardware. On machines without a GPU, this would crash upscale.py
(torch.device("cuda") fails) and ocr.py (PaddleOCR use_gpu=True).
Fix: the env var can only disable GPU (set to false/0), never
force-enable it. Hardware detection always runs. Removed the
baked env var from the Dockerfile since it adds no value now.
- README: add CUDA docker run example alongside full and lite
- Getting started: add GPU acceleration tip with speedup numbers
- Deployment: add CUDA row to variants table
- Docker tags: expand benchmarks with warm + cold start tables
- Add :cuda tag to Docker Tags docs with setup, benchmarks, compose example
- Add GPU acceleration tip to AI engine docs
- Include benchmark table from RTX 4070 testing
Add a :cuda Docker image tag that auto-detects NVIDIA GPU at runtime
and falls back gracefully to CPU. Same pattern as Immich.
- New gpu.py shared utility for cached CUDA detection
- Background removal (rembg): pass CUDAExecutionProvider to ONNX Runtime
- Upscaling (Real-ESRGAN): use CUDA device + FP16 when GPU available
- OCR (PaddleOCR): enable use_gpu when CUDA detected
- Dispatcher reports GPU status at startup via readiness signal
- Admin health endpoint exposes GPU availability
- Dockerfile uses ARG GPU=false with conditional NVIDIA CUDA base image
- docker-compose.gpu.yml override for GPU users
- CI/CD workflows build and publish :cuda tag (amd64 only)
Three tags: :latest (CPU), :lite (no AI), :cuda (GPU with CPU fallback)
Centralize duplicated getToken() + Bearer header logic into a single
formatHeaders() helper in lib/api.ts. When no token exists, the
Authorization header is omitted entirely instead of sending an empty
Bearer token, which breaks forward-auth proxies like Authelia behind
Caddy.
Changes:
- Add formatHeaders() with try-catch around localStorage access
- Replace 20+ duplicated getToken() definitions across tool components
- Migrate all call sites including file-details, settings, change-password
- Update tests to verify header omission on empty token
Based on the fix proposed by @jules2689 in #6, with improvements:
file placement (lib/api.ts vs components), localStorage error handling,
simplified truthiness check, and complete call-site coverage.
Co-Authored-By: Julian Nadeau <julian@jnadeau.ca>
Both llms.txt and llms-full.txt incorrectly said MIT. Also added lite
variant mention to the description and a warning callout on the AI
engine docs page noting AI tools are unavailable in the lite image.
README Quick Start now shows both :latest and :lite commands.
Getting Started adds a tip callout about the lite image.
Deployment page lists both variants with a comparison table and
updates the CI/CD description to mention both are built.
Developer guide adds the lite build command.
Linux libheif packages provide heif-convert instead of heif-dec (which
is macOS-only). The decoder now tries heif-convert first, then falls
back to heif-dec. Both accept the same argument syntax.
Ubuntu 24.04 uses plugin-based libheif codecs. Added libheif-plugin-x265
(HEVC encoder) and libheif-plugin-libde265 (HEVC decoder) to the CI test
job. Debian bookworm (Docker) bundles these in libheif1 directly.
Adds a new mode that trims uniform-color borders around the subject,
like GIMP's "Crop to Content." Includes configurable tolerance threshold
and optional pad-to-square with target size for e-commerce workflows.
The original attention-based crop is preserved as "Focus Crop" mode.
Closes#7
New users on dark-mode systems were seeing dark theme on first visit.
The default is now explicitly light, matching the API's DEFAULT_THEME.
Users can still switch to dark or system in settings.
- Health endpoint returns "healthy" instead of "ok" for consistency
- MAX_USERS now configurable via env var (default 5)
- People API returns team names instead of UUIDs in register/list
- PUT user update accepts team names (name-first lookup, fallback to ID)
- Login rate limit follows global rate limit when RATE_LIMIT_PER_MIN > 1000
- Strip-metadata preserves original format encoding instead of always PNG
- Fix e2e tests: rotate/crop/border button selectors match actual UI
- Fix e2e tests: create Engineering/Design teams in people test setup
- Fix e2e tests: people UI uses select for team field, not text input
- Update visual regression baseline for tablet home page
Without the fallback, every Copy button is dead on HTTP. The
execCommand approach is deprecated but works in all current browsers
and does not require a secure context.
Addresses issues #4 and #5 - crypto.randomUUID() and
navigator.clipboard.writeText() fail over plain HTTP on
non-localhost addresses, breaking all tool operations.
Worker thread initialization imports the tool registry which reads SQLite.
Under Docker volume filesystems, this can deadlock silently on SQLITE_BUSY,
causing APPLY to spin at 0% forever. Sharp operations complete in milliseconds
and don't need worker offloading. Added 30s AbortSignal timeout as defense
in depth for future re-enablement.
Adds a gosu-based entrypoint that starts as root, fixes ownership of
/data and /tmp/workspace for the stirling user, then drops privileges.
This fixes "SQLITE database not found" errors when users bind-mount
host directories.
Remove socialLinks config and three-dots menu. Add appearance toggle
and combined GitHub + Star button directly in the navbar via layout
slot. The button shows the Octocat icon, "Star" label, and live
stargazer count.
Replace MIT with AGPLv3 + commercial dual-license. Add copyright header
and dual-license notice to LICENSE, add CLA to CONTRIBUTING.md, update
README license section, and update all package.json license fields.
SVG files were rejected by the convert endpoint because
validateImageBuffer only recognized raster magic bytes. This adds
text-based SVG detection, sanitization in the tool factory, and
proper Sharp density handling so SVG-to-raster conversion works
through the standard convert route.
Update all references across docs, workflows, UI components, and config
to point to the new GitHub org (stirling-image/stirling-image) and Docker
Hub account (stirlingimage/stirling-image) ahead of repo transfer.
Add a privacy policy page accessible at /privacy (public, no auth required).
Relax Content-Security-Policy for /api/docs route to allow Scalar's inline
script initialization, fixing blank docs page in production.
Remove DB probe from public health endpoint - it only needs to confirm
the process is alive. Add test for non-admin user getting 403 on admin
health endpoint.
Public GET /api/v1/health now returns only status and version.
Full diagnostics (uptime, storage, database, queue) moved to
GET /api/v1/admin/health which requires admin authentication.
Validation now runs on all entries before any database writes.
Previously, clean entries could be written before a later malicious
entry triggered a 400 response.
PUT /api/v1/settings now returns 400 if any key or value contains HTML
tags. Settings are configuration values - there is no legitimate use
case for HTML in them.
The processed image in the BeforeAfterSlider was layered on top of the
original with a semi-transparent background. When the processed result
had transparency (e.g. after remove-background), the original image
showed through, making it look like the background was not removed.
Added an opaque checkerboard background behind the processed image so
transparent areas are clearly visible instead of showing the original.
Two issues caused intermittent step addition failures in the automation
pipeline:
1. RemoveBgControls had onChange in its useEffect deps. Since onChange is
a new function reference on every parent render, this created an
infinite re-render loop (effect -> setState -> render -> effect).
Fixed by using the onChangeRef pattern matching other settings
components.
2. All step mutation callbacks read from stepsRef.current and passed
values to setSteps. Concurrent callbacks (e.g. addStep + a settings
effect) would overwrite each other. Fixed by switching to functional
state updates (setSteps(prev => ...)) and removing stepsRef.
Also rewrites automate e2e tests to use manual step addition instead of
referencing templates that no longer exist in the UI.
Siddharth Kumar Sah