Repository navigation
Releases
Releases · deepratna-awale/Emotica
Emotica v2.3.0
Changed
- AutoExpress is now Emotica. The app, the WebUI extension, the ComfyUI and InvokeAI nodes, the SillyTavern extension, the MCP server and the Docker image all carry the new name:
python -m emotica,python -m emotica_mcp,EMOTICA_*settings andghcr.io/deepratna-awale/emotica. Nothing set up under the old name breaks:AUTOEXPRESS_*environment variables,python -m autoexpress_mcp, ComfyUI workflows withAutoExpress*nodes, the InvokeAIautoexpress_expression_promptsnode, the WebUI's saved output folder and Civitai key, SillyTavern extension settings, browser-saved choices,autoexpress_overrides.jsonfiles, puppets made asAutoExpress Puppetand olderautoexpress-frontend.zipreleases are all still read. The expression LoRA files keep theirAutoExpress-*names, and the Comfy Registry id staysautoexpressso existing ComfyUI installs keep updating. - The WebUI connections run on sdmux. Detecting the WebUI, the transport (timeouts, retries, logins, tokens, extra headers), and each WebUI's session, login, upload, queue, download, progress and interrupt steps now come from the sdmux library (0.2.0), which started out as this code. Emotica keeps its face edit, instruction edit, animation and Live2D workflows and the capability checks with fixes on top. Detection also recognises Forge 2.x, and names Draw Things and stable-diffusion.cpp when they answer instead of calling them unknown. Interrupting a ComfyUI job only touches that job: a queued prompt is dropped, a running one interrupted.
- The original interface at
/classiccatches up with the web interface. It now has the Login section, model defaults with notes when a setting is changed to match the WebUI, the Illustration/Photo style, Change pose too, the intensity slider, Remove background, Animate with idle-motion suggestions, a progress bar and missing-LoRA notes, Retry and per-expression settings with export and import, the Live2D model kit, pasting an image, and a working Interrupt button.
Docs
- README and docs brought up to date, with new screenshots. The README lists login for protected WebUIs, retrying one expression, the Live2D undress chain, touch reactions and the talking mouth, and drops a roadmap item that already shipped (poses). New screenshots of the current web interface, the Login section, the per-expression Retry dialog and the Live2D model kit dialog. The Live2D guide says the kit redraws the character in a T-pose by default (
--arm-angle) and documents the puppet's scripting API (expressions, reactions, IK,pose(), visemes). - Animated sprites in the README and docs. The README has an Animated sprites section with looping examples, and docs/animated-sprites.md shows more clips, frames from two loops, and how the endlessly looping WebP files keep playing in SillyTavern, browsers and other tools. The SillyTavern and RunPod guides explain where the looping sprites go, and the HTTP API table lists
lightningwith the current sampling defaults (20 steps, cfg 3.5).
Fixed
- Secrets stay out of logs and browser storage. A URL with
user:password@in it is no longer written to the log on a failed connect or kept in the browser's saved settings, error tracebacks no longer print local variables, and the connection settings hide passwords, tokens and headers fromrepr(). Extra headers are dropped when a redirect leaves the WebUI's host, and SwarmUI'sswarm_tokencookie only goes to the WebUI's host. - Live2D: the coat really comes off. Qwen-Image-Edit sometimes recoloured the blazer (a white one where the beige one was) instead of taking it off. The coat-off edit now asks for the whole garment gone, collar, lapels and sleeves included, with the shirt's own sleeves showing, and the vision language model on ComfyUI (Qwen3-VL) checks the result: while it still sees an outer layer the edit is drawn again, up to twice. The RunPod setup starts ComfyUI with
--disable-dynamic-vram, because ComfyUI's dynamic VRAM loading left Qwen3-VL's weights out when it writes text. - Live2D: shoes painted all over. Stockings carried down into the shoes were drawn over them when the split put the footwear under the legwear, which hid the inner half of each shoe. Shoes are now always drawn over the stockings.
- Live2D: no shadow streak under a raised arm. The shade an arm casts on the jacket or shirt beside it is lifted onto its own layer: the cloth underneath is painted lit, and the shade fades out as the arm moves away from the body.
- Live2D: the mouth and blush stay on the face as the head turns. They took their depth from their own small layers, so turning the head left them behind (about 40 px for the mouth). They now take the face's depth under them.
- Live2D: tongue and teeth in the visemes. For th, d, t, l and n the tongue now shows raised between the teeth (it was lifted out of the opening and hidden). The upper teeth sit back a little for small openings, and the f, th, d and s shapes are retuned. Grief, relief, surprise and realization have corrected eye, brow and mouth settings, and the swap between the open and the closed eye is quicker, so no ghost of both lids shows.
- ComfyUI: a download the proxy in front of ComfyUI drops (a 404 or 5xx on a file that is there, as RunPod's proxy does now and then) is tried again, up to four times, instead of failing the whole job. A Live2D layer split downloads dozens of files, and one dropped file used to throw the split away.
- Qwen-Image-Edit no longer picks Qwen-Image-Layered's VAE. With both installed in ComfyUI, the edit took
qwen_image_layered_vae(it sorts first), which burned any big change, such as a turned view, to near white. The edit now takes the Qwen-Image VAE the setup downloads.
New
- MCP server and an assistant skill.
python -m emotica_mcplets any MCP-capable assistant or agent drive the standalone app over stdio or streamable HTTP: check and connect the WebUI, make, retry and fetch expressions, animate them, and build Live2D kits. skills/emotica/SKILL.md is a vendor-neutral guide that teaches any assistant the workflow and the rules, with or without MCP. Setup for common clients is in docs/mcp.md; install withuv sync --extra mcp. - Live2D: the undress chain, in the picture's own pose. An opt-in way to build a kit's clothing (Undress chain, picture's pose,
"undress_chain": true,--undress): the vision model on ComfyUI tags the visible garments outermost first (slot, colour, the regions they cover) and imagines the inner wear under them, and the undress order, the states, the half states and the layer count are worked out from that (a character without outer wear gets fewer layers). Each state is a Qwen-Image-Edit run that takes one garment off, lined up with the state before, colour-locked and pixel-locked: outside the removed garment's region every pixel is copied from the state before, and revealed skin is colour-locked to the face. Each garment, the inner top and bottom (each with its own switch) and the body become layers, and the half states (jacket slipped, shirt open, skirt lifted) keep only the garment's own pixels. The puppet keeps the picture's pose: no arm motion, breathing, idle sway and cursor follow, the 28 emotions with head-and-body reactions, and a boop on the nose, a pat on the head, a poke on the cheek and a tap on the body.attach --fixed-poseadds the same runtime to a Cubism model. The PSD hides the half-off art by default, and every image that shows inner wear or less is namedINTERNAL_and kept out of previews and the contact sheet. See docs/live2d.md. - Retry a single expression, with its own prompt and settings. A Retry button above the gallery makes the expression on screen again with a new seed. Its dialog can change that expression's prompt (the tags, or the whole edit instruction) and its seed, intensity, denoising, steps, CFG and negative prompt. With Remember on, they are saved with the character (
Output/<name>/emotica_overrides.json) and every later run uses them. Advanced: per-expression prompts and settings edits all 28 at once, with Export and Import of the file. The API takes"overrides"and"save_overrides"onPOST /generate, and hasGET/PUT /expression-overridesandPOST /expression-prompt. The SillyTavern extension redoes a sprite the character already has with a new seed. - Login for protected WebUIs, right in the app. The WebUI Connection panel has a Login section under the URL: username (or InvokeAI email), password, token and Extra headers (for login proxies such as Cloudflare Access service tokens, also
EMOTICA_BACKEND_HEADERS). It opens by itself when the WebUI refuses the request. The login is kept on the Emotica server and the page only ever learns which fields are set, never their values. Reconnecting to the same host (from the app or the SillyTavern extension) keeps it; it is never sent to a different address, and a login from the environment only goes toEMOTICA_BACKEND_URL's host. ComfyUI's progress websocket now uses the same login, so the progress bar moves behind basic auth or a login proxy too. - Live2D puppets: real IK with elbow and knee hints.
puppet.ik(limb, target, hint)reaches a hand or foot to a point with a two-bone solver in 3D, the elbow or knee bending towards the hint point (by default out and down for an elbow, forward for a knee), within each joint's limits;puppet.ikHint(limb, hint)moves just the elbow or knee. Dragging a hand or foot uses it, and dragging the upper arm or thigh moves the hint. The joint limits are wider so the arms can wave, cheer and rest on the hips. The foot stays level as the leg moves. - **Live2D puppets: motions move the whole body....
AutoExpress v2.2.0
New
- SillyTavern extension: install AutoExpress in SillyTavern from this repo's URL and press Generate expression sprites to make a character's sprites from its avatar or another image. It drives the standalone app (and whichever WebUI that's connected to, local or remote), and each finished expression goes straight into the character's Character Expressions folder. Options for the checkpoint, prompt style, method, intensity, background removal and poses, a Missing button that picks only the expressions the character has no sprite for, progress with Cancel, and a
/autoexpresscommand. See docs/sillytavern.md. - The standalone app's API can be called from other web pages listed in
AUTOEXPRESS_CORS_ORIGINS, which allows SillyTavern's default address (http://localhost:8000,http://127.0.0.1:8000) out of the box. - Poses: change the pose along with each expression, for visual novel and chatbot sprites. Each of the 28 expressions has a small gesture that starts from the character's current pose (hands on the hips for anger, leaning in with the hands behind the back for curiosity), keeping the outfit, framing and background. Made with Qwen-Image-Edit on ComfyUI: in one edit with the instruction edit mode, or as a pose edit before face inpainting. A
poseinput on the Edit Instructions node, Change pose too in the standalone app and"pose": trueonPOST /generate.
Changed
- The Next.js web interface is always the default. A source install without Node.js now downloads the prebuilt interface from the GitHub release (
autoexpress-frontend.zip, published with each release) on first start, instead of falling back to the classic page. A build made withinstallis kept, and a downloaded one is replaced when AutoExpress is updated.AUTOEXPRESS_DOWNLOAD_FRONTEND=falseturns it off. - Anime expressions: the Danbooru, Pony and Illustrious tags and the sprite poses were tuned against a hand-drawn reference sprite set. Love no longer forces heart-shaped pupils, gratitude keeps the eyes open, grief shows teary eyes instead of sobbing, fear and realization get wavy open mouths, and small manga marks (anger vein, sparkle, flying sweatdrops) appear where they fit. Poses keep the hands off the face, so face inpainting still finds it after the pose edit.
Security
- minimatch 11.0.3 for the frontend's lint tools (GHSA-g3hv-75wg-m249, a denial of service through brace expansion). It is only used while developing and is not in the built app.
- Pillow 13.0.0 and rembg 2.0.86 in
uv.lock, fixing 41 Pillow advisories (heap overflows in the JPEG, TIFF and FlashPix decoders, the Ghostscript argument injection, decompression bomb bypasses and others) and 3 rembg advisories (SAM model injection, SSRF and path traversal in custom model loading). - The standalone app now needs Python 3.11 or newer, since the fixed Pillow and rembg releases no longer support 3.10. The WebUI extension and nodes still work on Python 3.10.
Downloads
- ComfyUI: install AutoExpress from ComfyUI-Manager or the Comfy Registry.
- AUTOMATIC1111 / Forge / reForge / SD.Next: Extensions > Install from URL, paste
https://github.com/deepratna-awale/AutoExpress. - SillyTavern: Extensions > Install extension, paste
https://github.com/deepratna-awale/AutoExpress, and run the standalone app below alongside it. - Standalone app, Windows: download
autoexpress-v2.2.0-Windows_Release.zip, extract it and runrun.bat. - Standalone app, Linux: download
autoexpress-v2.2.0-Linux_Release.zip, extract it and run./run.sh. - Standalone app, macOS: download
autoexpress-v2.2.0-macOS_Release.zip(Apple silicon, macOS 14+) orautoexpress-v2.2.0-macOS-Intel_Release.zip(no Remove background), extract it and run./run.shin Terminal. - Standalone app, Docker:
docker run -p 5000:5000 ghcr.io/deepratna-awale/autoexpress:2.2.0 - From source: see the README.
Full changelog: v2.1.1...v2.2.0
AutoExpress v2.1.1
- macOS releases:
autoexpress-<version>-macOS_Release.zipfor Apple silicon (macOS 14+) andautoexpress-<version>-macOS-Intel_Release.zipfor Intel Macs, started with./run.sh. The Intel build has no Remove background, since onnxruntime isn't built for Intel Macs. - Dependencies live in
pyproject.tomlandrequirements.txtis gone:dependenciesfor the WebUI extension and nodes, astandaloneextra for the app and adevgroup for tests.uv.lockpins every package with hashes and platform markers.- The release zips, the Docker image and
install(when uv is installed) install exactly the versions inuv.lock, picking the right wheels for each OS and CPU. Without uv,installusespip install -e ".[standalone]". - Installing AutoExpress into ComfyUI from git no longer pulls the standalone app's dependencies (Flask, rembg, ...) into ComfyUI.
- The release zips, the Docker image and
- Release zips bundle their own Python (3.13, in
runtime/), each built on its own platform from prebuilt wheels only, so a missing wheel fails the build instead of shipping a zip that doesn't start. Before, the zips carried a virtual environment that pointed at the build machine's Python. runandrun_debuguse the bundled Python when it's there, and.venvotherwise. On macOS they clear the download quarantine on the bundled Python.- Each release zip is smoke tested (the app starts, serves the web interface, and Remove background is available where it should be) before it's published.
- Zip names keep the version as tagged, e.g.
autoexpress-v2.1.1-Linux_Release.zip. - The dependency audit checks the versions pinned in
uv.lock, and Dependabot updatesuv.lock.
Downloads
- ComfyUI: install AutoExpress from ComfyUI-Manager or the Comfy Registry.
- AUTOMATIC1111 / Forge / reForge / SD.Next: Extensions > Install from URL, paste
https://github.com/deepratna-awale/AutoExpress. - Standalone app, Windows: download
autoexpress-v2.1.1-Windows_Release.zip, extract it and runrun.bat. - Standalone app, Linux: download
autoexpress-v2.1.1-Linux_Release.zip, extract it and run./run.sh. - Standalone app, macOS: download
autoexpress-v2.1.1-macOS_Release.zip(Apple silicon, macOS 14+) orautoexpress-v2.1.1-macOS-Intel_Release.zip(no Remove background), extract it and run./run.shin Terminal. - Standalone app, Docker:
docker run -p 5000:5000 ghcr.io/deepratna-awale/autoexpress:2.1.1 - From source: see the README.
Full changelog: v2.1.0...v2.1.1
AutoExpress v2.1.0
New
- Instruction edit mode on ComfyUI: Qwen-Image-Edit changes the expression from one sentence instead of inpainting the face. Tuned illustration and photo wordings for all 28 expressions, an Edit Instructions node, readiness checks and an example workflow that needs only core nodes.
- Photos use Qwen-Image-Edit-2511 at 40 steps and CFG 4, and only the edited face is pasted back onto the original through a feathered mask, so the rest of the photo is untouched.
- Expression intensity slider (0 to 1) with a default per model family: inpainting with expression LoRAs for anime families, instruction edit with photo wording for realistic and natural-language models.
- The standalone app works with a WebUI hosted anywhere (LAN, RunPod and other hosts, https, reverse-proxy sub-paths, logins). Nothing assumes it runs on this machine.
- Detect automatically (the new default) reads the WebUI type from its API: A1111, Forge, SD.Next, ComfyUI, SwarmUI or InvokeAI.
- After connecting, a setup check lists what the WebUI has and what's missing (ADetailer and its face model, Impact Pack and Subpack, face detection models, checkpoints, samplers, LoRAs). Each missing piece comes with steps for that WebUI, commands to copy, links, and a Re-check button. Also at
GET /backend/capabilities. - Connection errors say what to try:
--listen, the host's proxy URL, or the login. - A RunPod proxy URL for a port the pod doesn't expose says to add the port under Expose HTTP Ports.
- Expression LoRAs you renamed on the WebUI are recognised by the hashes in their metadata (A1111 family, InvokeAI with sha256 hashing).
- Expression LoRAs are written to the LoRA folder only when it's on this machine. InvokeAI installs missing ones itself through its Model Manager. For other WebUIs, the run lists each missing LoRA with its download link (
noteson the job).
- Remove background in the standalone app saves every expression with a transparent background, only the character is kept. Uses rembg (
isnet-animefor drawn characters,u2net_human_segfor photos), installed fromrequirements.txt. Also"remove_background": trueonPOST /generate, andGET /background-removal.
Changed
- The expression LoRA base (SD 1.5, SD 2, SDXL, Flux) is read from the model itself instead of guessed from the checkpoint name, so SD 1.5 checkpoints with unusual names no longer get SDXL LoRAs. The name still tells Pony and Illustrious apart from plain SDXL.
- LoRA entries can set a weight per expression (0 drops the LoRA and its trigger). The Illustrious expressions LoRA is skipped for anger, and the Illustrious family recommends NoobAI-XL, WAI and other finetunes over base Illustrious-XL 0.1.
install.sh/install.batbuild the web interface too (Node.js 20.9+, pnpm or corepack), check for Python 3.10+, reuse.venvunless--cleanis given and accept--skip-frontend.install.shno longer runsapt update/upgrade, so it works on macOS.run/run_debugcheck that the app is installed and start from their own folder.run_debuglistens on 127.0.0.1 by default, since the Werkzeug debugger can run code.- The Docker image runs as an unprivileged user (uid 1000). A mounted results folder needs to be writable by it.
--build-arg PRELOAD_REMBG_MODELS="..."bakes background removal models into the image. - The frontend, release zips and Docker image build on Node 26 (the Docker image installs pnpm with npm, since corepack is gone).
- Release notes come from this changelog.
- Each release publishes the Docker image to
ghcr.io/deepratna-awale/autoexpress. The Docker Hub image is no longer updated. - matplotlib is no longer a dependency.
- README and docs restructured, with a page per integration under docs/.
Security
- Option values and labels are escaped in the classic interface, the LoRA tag parser runs in linear time, and release jobs get minimal permissions.
- Dependabot and a weekly dependency audit (pip-audit and pnpm audit); sd-parsers 0.6 and higher requirement floors.
Downloads
- ComfyUI: install AutoExpress from ComfyUI-Manager or the Comfy Registry.
- AUTOMATIC1111 / Forge / reForge / SD.Next: Extensions > Install from URL, paste
https://github.com/deepratna-awale/AutoExpress. - Standalone app, Windows: download
autoexpress-v2.1.0-Windows_Release.zip, extract it and runrun.bat. - Standalone app, Linux: download
autoexpress-v2.1.0-Linux_Release.zip, extract it and run./run.sh. - Standalone app, Docker:
docker run -p 5000:5000 ghcr.io/deepratna-awale/autoexpress:2.1.0 - From source: see the README.
Full changelog: v1.0.0...v2.1.0
AutoExpress v1.0.0
AutoExpress v1.0.0
AutoExpress automatically creates 28 different expressions from a given image using Automatic1111 Stable Diffusion WebUI API.
What's Changed
- Adds debug argument to run_debug files (f9cf2e1)
- tries to remove recursively copying packages (4c3a0d7)
- exclude stuff inside folder linux (e41f744)
- Add release folder to ignore list in rsync (cc51598)
- Adds virtual env dir to copy list for both envs (f7f1a98)
- Adds correct tag version using _ (cf6f3a7)
- Windows release path issue fix (875278c)
- create release folder (2e5ff62)
- Windows build workflow force move readme (13d9a13)
- Go back to copying files using rsync in linux build (ed6ec4d)
- Fix linux pathing issue (99a608a)
- Use repo name instead of workdir var (c828edc)
- ls workdir (ea6891a)
- Print PWD Linux (2873d8a)
- Linux PWD fix (2b1aa7c)
- Linux fix (a82d125)
- Fixes .venv creation (52a4d6e)
- Edit workflows for easier release (a23bd21)
- Makes the release workflows use _ instead of . in directory name. Instead of copying all to a new dir, the same dir is compressed as a different name (6cb1741)
- Fixes paths for linux/mac systems. (3a75a32)
- Adds link to live website (4640200)
- Adds online UI support. (114ea93)
- Removed thread lock (1652183)
- Added apt update, upgrade, python install in installer shell script (080dc8b)
- Interruption functionality (64f720a)
- Fixed ASCII header parsing error (d238ef2)
- Fixed typo in ASCII header (6d2d88d)
- Added installation instructions using install scripts. (9771545)
- Fixed click propogation when clicking info icon (59d2576)
- Fixed click propogation when clicking info icon (61a4b95)
- Fixed parameters dissapearance if image is loaded first and then connect is pressed. (c57ef9d)
- Added .dockerignore and updated Dockerfile (ab79ed7)
- This should have been the commit message for the last commit: - This fixes the Lora MultiSelect - Fixes Installer typo. - Reduces Image size in viewer Last Commit Log: - Added multiple Lora Support. - Refactored A1111 Client and added docstrings - In /autoexpress/modules/image_parser.py - Added width, height from pillow image metadata - Enabled two pass and normalize parameters. - Changed params keys to target ad_* - Added function to export uncleaned parameters - In /autoexpress/resources/payload.json_handler - Updated json to match the new A1111 WebUI. - In /autoexpress/static/js/connect.js - Removed stray semicolon - Added Scheduler dropdown. - Added Interrupt functionality. - Added installer scripts for windows and linux. - Added run scripts for windows and linux. (removed startup.bat and startupDebug.bat) (d0c7e87)
- Added install_logs.txt to git ignore. (521a6a4)
- Readme updated with docker image instructions. Changelog updated with latest changes. (deb2613)
- added pip upgrade to dockerfile, deleted frozen requirements and added library names instead in requirements.txt (26ae298)
- Added linux requirements (29bc59f)
- Removed Rust from Dockerfile (0e2d9e8)
- Linux requirements file (927f3b1)
- Generate Button color change (5327502)
- Changed Dockerfile and requirements.txt (912fe7b)
- Added authors to index.html (6c5f25e)
- Major UI update by Avery - Variable based CSS allows easily changing the theme from \autoexpress\static\stylesheet\main.css - Option to pause Gradient Animation. Changes/Contribs of Deep: - Sets default settings for UI theming. - Introduces a modal for GUI settings and other future settings. - Adds image info icon which lets user see the raw generation data when hovering over the icon or uploaded image. - Scheduler added to list of parameters. - If element does not exist in dropdown (Model, Sampler, Scheduler, Lora), a new dropdown element is created. (a13bcae)
- Huge UI Update (22d84e1)
- Removed comma at end of clip_expressions. (a3ac436)
- Update README.md (c25aab9)
- Exp fixes, new .bat (f12c531)
- Update README.md (f75af5e)
- Adds Dockerfile. (8529466)
- Removed streamlit from enviorenment dependencies. (ccf25e4)
- Update Changelog.md (b5dfd1a)
- Update README.md (9794bfe)
- Update README.md (50d5c3c)
- Update clip_expressions.json (b68d497)
- Update clip_expressions.json (c5fb5a8)
- Added sd-parsers to enviorenment requirements. (a50c342)
- Update expressions.json (7163218)
- Update expressions.json (dee9f06)
- Update clip_expressions.json (db118b7)
- Update README.md (9b3a9e3)
- Fixed Issue #10 (984f078)
- created getURL and setURL instead of editing url variable directly. (50887d8)
- Changed Border Radius for buttons, changed positions of fns in a1111_client.py. Image_parser.py now uses latest sd_parser (ceb25f6)
- Renamed a1111_api.py to a1111_client.py (2d421b3)
- Changed a1111api to be object oriented. (b4fed2a)
- Changed name of generate to expression generator, simplified image parser module. (17c1a23)
- Corrected paths for image and changelog files (2ac79e9)
- Corrected paths for image and changelog files (2a2f218)
- Removed termcolor support to make logs readable. (b3e2f9b)
- Removed in-house logger and added loguru for logging. (196617c)
- New UI Image Source added. (63492ed)
- UI Overhaul. (7d71d5b)
- Updated requirements file to include loguru (b43b9c8)
- adds create_sprite() functionality to create_grid.py (8f7242f)
- Bump tqdm from 4.66.1 to 4.66.3 (fb50653)
- Bump idna from 3.6 to 3.7 (816e1a0)
- Bump pillow from 10.1.0 to 10.3.0 (0faaa43)
- Bump requests from 2.31.0 to 2.32.2 (bf2ed57)
- Fixed namespace issue with generate module and generate route. (4ef2e2a)
- generate in views.py now returns a json if adetailer is not found (719035c)
- Update generate.py (dd57ced)
- Adds Anime /Realistic switch. (86b9a31)
- Update README.md (14aa739)
- Update README.md (2667d18)
- Update README.md (c60eb10)
- Updated README.md (b80c159)
- Update README.md (ba458ff)
- Update README.md (3ee31d0)
- Added favicon rouded (4d963ac)
- Corrected img src tag in readme (a130557)
- Removed .png from .gitignore (9844c63)
- Update README.md (5216322)
- Renamed package to autoexpress (d7cd689)
- Sub divided modules and refactored. Fixed Logger duplication issue. (40fc2b3)
- Switched to WebUI (68ef1a1)
- Major Refactor. Renamed and moved all source files. (fc884ea)
- Update .gitignore (022ac8a)
- init (07cd2da)
- Initial commit (afadbf8)
Downloads
🚀 Ready-to-Run Releases (Recommended)
- Windows: Download
autoexpress-v1.0.0-Windows_Release.zip, extract and runrun.bat - Linux: Download
autoexpress-v1.0.0-Linux_Release.zip, extract and run./run.sh
These releases include all dependencies pre-installed in a virtual environment. No additional setup required!
📦 Installation Options
Option 1: Docker (Cross-platform)
docker pull deepratna/autoexpress:v1.0.0
docker run -p 7860:7860 deepratna/autoexpress:v1.0.0Option 2: From Source
- Download and extract the source code (
autoexpress-v1.0.0-source.zip) - Install dependencies:
pip install -r requirements.txt - Run the application:
python -m autoexpress
Requirements
- Automatic1111 Stable Diffusion WebUI with API enabled
- After Detailer Extension installed
Links
Full Changelog: ...v1.0.0