{"id":27452127,"url":"https://github.com/beinsezii/quickdif","last_synced_at":"2025-04-15T11:40:13.099Z","repository":{"id":205391873,"uuid":"714131554","full_name":"Beinsezii/quickdif","owner":"Beinsezii","description":"Quick and easy Diffusers CLI","archived":false,"fork":false,"pushed_at":"2025-04-14T04:13:52.000Z","size":1794,"stargazers_count":13,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-04-14T05:25:11.993Z","etag":null,"topics":["cli","diffusers","diffusion","diffusion-models","image-generation","image2image","img2img","pytorch","stable-diffusion","text2image","txt2img"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"gpl-3.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Beinsezii.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2023-11-04T02:12:49.000Z","updated_at":"2025-04-14T04:13:43.000Z","dependencies_parsed_at":"2023-12-26T03:23:40.938Z","dependency_job_id":"997d27ef-eb3e-4cd8-96ff-2470f7d0e2f1","html_url":"https://github.com/Beinsezii/quickdif","commit_stats":null,"previous_names":["beinsezii/quickdif"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Beinsezii%2Fquickdif","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Beinsezii%2Fquickdif/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Beinsezii%2Fquickdif/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Beinsezii%2Fquickdif/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Beinsezii","download_url":"https://codeload.github.com/Beinsezii/quickdif/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":249062686,"owners_count":21206750,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["cli","diffusers","diffusion","diffusion-models","image-generation","image2image","img2img","pytorch","stable-diffusion","text2image","txt2img"],"created_at":"2025-04-15T11:40:11.221Z","updated_at":"2025-04-15T11:40:13.086Z","avatar_url":"https://github.com/Beinsezii.png","language":"Python","readme":"# quickdif\nQuick and easy CLI inference that just works™ for a variety of Diffusers models\n\n## Including\n  * Verified working across many models\n    + AuraFlow\n    + FLUX\n    + Hunyuan\n    + Kandinsky\n    + Kolors\n    + Lumina T2X\n    + Pixart (Alpha/Sigma)\n    + Sana\n    + Stable Cascade\n    + Stable Diffusion (1.5/2.1)\n    + Stable Diffusion 3\n    + Stable Diffusion XL\n      - No refiner\n    + Wuerstchen\n    * Other Diffusers format models will likely work to varying degrees\n  * Multi-lora fusion using `peft` for minimal performance loss\n  * Features all of the most common generation parameters\n  * Includes many advanced generation parameters\n    * 1, 2, 3, 4, 5, 6, 7, 8-bit quantization\n    * PAG / Perturbed-Attention Guidance\n    * Many latent augmentation params\n    * Many post processing effects\n    * Fully modular samplers/schedules, courtesy of [skrample](https://github.com/Beinsezii/skrample)\n  * Iterate over most parameters and create grids\n  * Expand prompts with `photo of {all|of|these}` or `photo of [one|of|these]`\n  * Load settings from JSON, TOML, PNG\n  * Extremely small 1-shot script using `accelerate` for hot loading models\n  * Targeted AMD/ROCm optimizations\n    * NVIDIA/CUDA does not need additional optimizations\n\n## Not Including\n  - ControlNet, Inpaint\n  - Multi-stage models: DFIF Stage 2, SDXL Refiner\n  - Server/API for a perpetual instance\n  - 100% maximum throughput\n  - 1-click installer\n  - MacOS/MPS and Intel OneAPI may not function properly\n\n## Installation\nThis project only supports `pip` based installs on Python 3.11+\n\n#### Basic setup\n```sh\n\u003e git clone https://github.com/Beinsezii/quickdif.git\n\u003e cd ./quickdif/\n```\n\n#### Create a venv in the quickdif folder\n```sh\n\u003e python3 -m venv ./venv\n# alternatively, it's recommended to use the full virtualenv if you have it\n\u003e virtualenv ./venv\n```\n\n#### Install dependencies\n```sh\n# replace with appropriate activation script for other shells\n\u003e source ./venv/bin/activate\n# It's recommended to first install torch using the recommended commands from https://pytorch.org/get-started/locally/\n\u003e pip install torch --index-url https://download.pytorch.org/whl/rocm6.0 # AMD example\n# finally\n\u003e pip install -e .\n\u003e deactivate\n\u003e ./quickdif.sh \"kitten\"\n# Windows users will instead have to invoke Python directly\n\u003e python quickdif.py \"kitten\"\n```\n\n### As a module\nAdditionally, you may use the project as a python module\n```sh\npip install git+https://github.com/Beinsezii/quickdif.git\npython -m quickdif \"kitten\"\n```\n\n## Usage\n```sh\n# See all options. Always refer to the script help over the other examples in this README\n\u003e ./quickdif.sh --help\n\n# Run with defaults\n\u003e ./quickdif.sh \"high resolution dslr photograph of pink roses in the misty rain\"\n# Custom model\n\u003e ./quickdif.sh \"analogue photograph of a kitteon on the beach in golden hour sun rays\" -m \"ptx0/terminus-xl-gamma-v1\"\n# Single files work for Stable Diffusion\n\u003e ./quickdif.sh \"colorful fantasy artwork side profile of a feminine robot in a dark cyberpunk city\" -m ./checkpoints/sd15/dreamshaper-v6.safetensors\n# Four dog and four cat images at twenty steps\n\u003e ./quickdif.sh \"photo of a dog\" \"painting of a cat\" -B 4 -s 20\n# Colored latent for an offset noise effect\n\u003e ./quickdif.sh \"underwater photograph of a dark cave full of bioluminescent glowing mushrooms\" -g 9.0 -s 30 -C black -c 0.8\n# Compile for a long job\n\u003e ./quickdif.sh $(cat prompts.txt) --compile\n# Export favorite settings to the defaults JSON\n\u003e ./quickdif.sh -m \"stabilityai/stable-cascade\" -s 20 -n \"blurry, noisy, cropped\" --json ./quickdif.json\n# Save a style to a custom JSON\n\u003e ./quickdif.sh \"fantasy artwork of a kitten wearing gothic plate armor\" -g 10 -G 0.5 --json ./epic_kitten.json\n# Merge multiple configs\n\u003e ./quickdif.sh -I underwater_cave.png epic_kitten.json\n```\n\n## F.A.Q.\nQuestion|Answer\n---|---\nWhy not X popular UI?|SD.Next's diffusers backend is extremely buggy/broken in multiple areas and InvokeAI (+non-diffusers UIs) only really supports Stable Diffusion.\nWindows?|The python script should work just fine but you'll need to set up the powershell/CMD stuff on your own.\nGradio?|No. If a UI ever gets made for this it'll be its own separate entity that interfaces via API. Cramming a bad gradio interface into the main script wont do anyone any favors\nFeature XYZ?|Maybe. Things in the *Not Including* list may come eventually™ if this script winds up being useful enough\n","funding_links":[],"categories":[],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbeinsezii%2Fquickdif","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fbeinsezii%2Fquickdif","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbeinsezii%2Fquickdif/lists"}