Grasshopper Developing
StabilityAI (Stable Diffusion) inside Grasshopper with Python – Advanced version

πŸ“£ Now the AI components of Ambrosinus-Toolkit are about two months old [today is 25/01/2023]

Although this, with Build-106 of the StabilityAI-GH component I have introduced and included the last features of “stability-sdk” Python library.

So, summarized you can run/apply the following features:

  • Text-to-Image mode
  • Image-to-Image mode
  • Image-to-Image masking mode (with grayscale and blurred images as masks) 
  • Apply CLIP guidance presets (None, Simple, FastBlue, FastGreen, Slow, Slower and Slowest)

by using a couple of new/updated components like GrayGaussMask, FileNamer and SdINinfo

it is possible to:

  • Generate a grayscale image and a blurred one thanks to some python libraries – You have full control over black/white areas 
  • Generate a filename without incurring the overwriting action and put also the mask images into “IMGs” folder
  • Read all useful parameters used in generating your AI art image
  • Track all runs in a unique CSV file

About GrayGaussMask component In order to run it you need to install the Python library as shown here:
pip install torchvision
This library is ~160MB so wait until appear the installation successful message.

Below are these components in action

 

I have always used to say: 

A Computational Designer is also (but not exclusively) a data storyteller

While you are exploring your design and pushing forward your creativity, the StabilityAI-GH component (the same happens for OpenAI) stores all data in this way:

  • Images are stored in the “YourCustomFolder\IMGs” folder
  • Metadata (for each run) is stored in the “YourCustomFolder\TXTs” folder
  • All your runs are stored in a unique CSV file

Here below the video DEMO

 

Download this component from HERE