Skip to content

wdshin/aiimages

 
 

Repository files navigation

aiimag.es

GitHub Discord GitHub Repo stars

DOWNLOAD: aiimag.es Download on itch.io: aiimag.es

Installation for Users

  1. Go to aiimag.es and click "download"
  2. Unzip it (at a path that has no space in it)
  3. Run aiimages.exe

aiimages tool

About

Goals

  1. Make installation as easy as possible
  2. Accessible to everyone
  3. Easy to iterate on images

Features

  • Inpainting! You got a broken face or a missing arm? Just paint over it in the app and let it be redrawn!
  • Variations! Found an almost-good image? Lock the seed and create slight variations of it!
  • Better faces! Got slightly broken face? Let the face AI update it!
  • Upscaling! You fixed the face, the missing arms and got the right variation? Time to upscale it with just another AI!
  • Seamless option. Wanna create beautiful tiling textures? Apply tiling!
  • History view. Just go back to your old stuff!
  • Drag'n'drop. Want to use a picture as input? Drag it there. Want to use a specific seed? Drag it on the seed. Etc.
  • Templates. Save good style/content prompts and quickly load them. Comes with some pre-installed templates!
  • Open Source! Yeah, I guess you already know that...

Used Software

Future Features

  1. Networking (aka "Borrow a GPU")
    1. Friends can connect to your PC and run their prompts on your GPU
    2. Using a custom solution, and Mirror and Light Reflective Mirror
  2. Android / WebGL versions (networking only)
  3. User/project managment. Different users on one PC, different histories, ...
  4. Save/share style palettes. Also just drop a generated image in to use it's style.
  5. Mac/Linux builds. Should be easy with Unity, but I cannot test it at home. :(
  6. Multi-GPU support. Start the backend twice, get twice as many pictures!
  7. GPU clustering. Connect to friends via network, prompts are calculated on the free GPUs.
  8. Outpainting window.
  9. UI improvements. "Generate X pictures" button, better error feedback, ui animations,...
  10. Lexica.art wishlist. Simply input nice prompts from lexica.art to test them later (and create templates from it).
  11. Stats view. E.g. containing the GPU time, estimating power usage and costs, etc.

For Developers

Installation

  1. Alternative: Instead of installing the LStein environment, you could use the one from the build. But you'll have to edit GeneratiorConnection.cs (basically reverse the #ifdefs). Sorry that I didn't allow for that yet. :X
  2. !!! Install everything that follows on C:
  3. Follow the tutorial of Lstein (don't worry, it's easy)
  4. Open the repo in Unity 2020.3.16f1

Deploying your solution

If you only change stuff on the Unity side:

  1. Build your project
  2. The downloads of dependencies should still work

If you also changed the conda environment:

  1. (Coming soon. For now see this)

How does it work interally?

  1. The Unity project is zipped with all prerequisites: conda environment (with conda-pack), model, python, sd-repo, ai cache.
  2. Unity starts an invisible command line, runs the dream.py and sends it prompts
  3. ...once an image appears, Unity displays it

About

A Unity UI and easy installer for Stable Diffusion.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C# 66.2%
  • ShaderLab 28.7%
  • HLSL 5.0%
  • Batchfile 0.1%