amotile/stable-diffusion-workshop

React Frontend for stable diffusion

Amotile's Workbench - Stable Diffusion

A browser based frontend to generate images with stable diffusion. Written in React.

Here's a video showing how it can be used: https://youtu.be/P9w57TiJWUM

Features

• txt2img & img2img
• history of generated images
• simple to try out variations
• no additional installation needed
• possibility to add new backend, currently only AUTOMATIC1111 supported.

How to run

Hosted

how

• Go to this url: https://amotile.github.io/stable-diffusion-workshop/
• Change the backend if needed, default points to a AUTOMATIC1111 installation on your own localhost
• That's it.

why

• Pro: No install required
• Con: Might break at any time if a new version is deployed

Your own copy

how

• Clone or download this repo.
• Host the /doc folder in any static website hosting system.
• Then configure like above.

why

• Pro: You decide when you update
• Con: Slightly more complicated (But if you got AUTOMATIC1111 running this is easy)

Development build

how

• It's pretty standard create-react-app project, so run:
• yarn
• yarn run start

why

• Pro: You can change the code
• Con: Yet more complicated

Why this project?

• I wanted a UI that better fit my workflow
• I know how to build things with react

stable-diffusion-workshop's Issues

Throwing errors with the latest version of the automatic webui

Thanks for building this tool. It's very exciting, and I'm looking forward to this amazing workflow.

I'm having an issue running this against the latest code in the automatic webui.

I've got the latest commit of the automatic webui installed https://github.com/AUTOMATIC1111/stable-diffusion-webui/tree/5a1951f17567022612aef5302fb650923a06fb54

There appears to be an additional parameter being passed through which is causing gradio to throw an error.

I get the following errors from the webui:

TypeError: string indices must be integers
Traceback (most recent call last):
  File "/home/eware/anaconda3/envs/sd-automatic/lib/python3.9/site-packages/gradio/routes.py", line 273, in run_predict
    output = await app.blocks.process_api(
  File "/home/eware/anaconda3/envs/sd-automatic/lib/python3.9/site-packages/gradio/blocks.py", line 750, in process_api
    inputs = self.preprocess_data(fn_index, inputs, state)
  File "/home/eware/anaconda3/envs/sd-automatic/lib/python3.9/site-packages/gradio/blocks.py", line 615, in preprocess_data
    processed_input.append(block.preprocess(raw_input[i]))
IndexError: list index out of range

Here's an example of a web request from the latest webui with 36 elements in it:

{
  "fn_index": 10,
  "data": [
    "an orange cat smoking a cigar, fat cat, business man, concept art, 4k",
    "",
    "None",
    "None",
    20,
    "Euler a",
    false,
    false,
    1,
    1,
    7,
    -1,
    -1,
    0,
    0,
    0,
    false,
    512,
    512,
    false,
    false,
    0.7,
    "None",
    "",
    false,
    null,
    "",
    false,
    "Seed",
    "",
    "Steps",
    "",
    true,
    null,
    "{\"prompt\": \"an orange cat smoking a cigar, fat cat, business man, concept art, 4k\", \"all_prompts\": [\"an orange cat smoking a cigar, fat cat, business man, concept art, 4k\"], \"negative_prompt\": \"\", \"seed\": 486859185, \"all_seeds\": [486859185], \"subseed\": 400618458, \"all_subseeds\": [400618458], \"subseed_strength\": 0, \"width\": 512, \"height\": 512, \"sampler_index\": 0, \"sampler\": \"Euler a\", \"cfg_scale\": 7, \"steps\": 20, \"batch_size\": 1, \"restore_faces\": false, \"face_restoration_model\": null, \"sd_model_hash\": \"7460a6fa\", \"seed_resize_from_w\": 0, \"seed_resize_from_h\": 0, \"denoising_strength\": null, \"extra_generation_params\": {}, \"index_of_first_image\": 0}",
    "<p>an orange cat smoking a cigar, fat cat, business man, concept art, 4k<br>\nSteps: 20, Sampler: Euler a, CFG scale: 7, Seed: 486859185, Size: 512x512</p><div class='performance'><p class='time'>Time taken: <wbr>3.40s</p><p class='vram'>Torch active/reserved: 3161/3688 MiB, <wbr>Sys VRAM: 5393/24269 MiB (22.22%)</p></div>"
  ],
  "session_hash": "4umtmkku44w"
}

The payload from the workshop front-end is passing through 35 elements:

{
  "fn_index": 10,
  "data": [
    "an orange cat smoking a cigar, fat cat, business man, concept art, 4k",
    "",
    "None",
    "None",
    15,
    "LMS",
    false,
    false,
    1,
    1,
    7.5,
    "1468895597",
    -1,
    0,
    0,
    0,
    false,
    512,
    512,
    false,
    false,
    0.7,
    "None",
    false,
    null,
    "",
    false,
    "Seed",
    "",
    "Steps",
    "",
    true,
    null,
    "",
    ""
  ],
  "session_hash": "d1h1f518vfu"
}

Sorry not an issue, but question.

Hi great work, excellent concept.

Not sure if this still works or you have abandoned, can you kindly point me to what I would need to modify to support the API_AUTH parameter to support the automatic1111 username:password.

Thank-you