Free and Open Source Machine Learning Plugin for Houdini developed by Ambrosiussen Holding and Entagma, Licensed and Distributed by Bismuth Consultancy B.V. _By downloading or using the plugin (or any of its contents), you are agreeing to the LICENSE found in this repository and Terms of Service of Bismuth Consultancy B.V._
Paul Ambrosiussen | Entagma | Discord |
---|---|---|
<img src="http://img.youtube.com/vi/izLicMTBYUg/0.jpg" alt="2.0 Release Promo" width="640" height="480" border="0" />
To install the plugin for the first time, follow these steps:
MLOPs.json
file found in the repository root, and paste it in the $HOUDINI_USER_PREF_DIR/packages/ folder.MLOPs.json
file you just pasted, and modify the $MLOPS
path found inside. Set the path to where you cloned the repository to in step one.git
. Follow the instructions for your relevant OS here.MLOPs
shelf. Click the Install Dependencies
shelf button. Restart Houdini once complete.MLOPs
shelf. Click the Download Model
button. Optionally change the Model Name
parameter to a custom model, or just leave as is and hit Download
to work with the default Stable Diffusion Model.[type] Model
parameters to select a downloaded model to use. You can also provide a repo name from the Huggingface Library, and the nodes will download it for you. For example runwayml/stable-diffusion-v1-5
.$MLOPS_SD_MODEL
is the path to a SINGLE model used by all Stable Diffusion nodes by default. You can set this to be your preferred default model.$MLOPS_MODELS
. (Notice the S at the end) This will make them show up in the dropdowns for the model paths on the nodes.
Both of the above varibles can be changed in the MLOPS.json
to suit your preference.Install Dependencies
shelf button again. PYTORCH_ENABLE_MPS_FALLBACK=1
$MLOPS_MODELS/cache/[your model]
and try again. This is likely caused by a corrupt cache.PYTHONPATH
environment variable set. If that is the case remove it and restart Houdini (And the Launcher if you use it)hip/
folder.requirements.txt
.