stylegan3 VS CLIP

StyleGAN3 by Nvidia Open Source Software »

StyleGAN by Nvidia (Open Source) - GitHub » StyleGAN on GitHUB

Not unless you are Nvidia Corporation: https://github.com/NVlabs/stylegan3/blob/main/LICENSE.txt

I found the follow script but unsure of how to utilize it in this context. https://github.com/NVlabs/stylegan3/blob/main/torch_utils/persistence.py

full dataset.py file that I'm changing is here: https://github.com/NVlabs/stylegan3/blob/main/training/dataset.py

The neural network on which the training was carried out - https://github.com/NVlabs/stylegan3

I trained a Generative Adversarial Network stylegan2-ada model with NVIDIA’s StyleGAN3 algorithm with a RTX 3090 GPU for a few nights. This is an interpolation video of the generator model’s random-walk datapoints divided into 4 different windows. Initially I web crawled some 2000 images of any Formidable images and cropped them with nagadomi’s lbpcascade_animeface anime face detector, with a setting that I attempted to also include her assets in the image. Previously I have done by transfer-learning from Gwern’s ThisWaifuDoesNotExist, which only included heads of Emilia from Re:Zero, which was quite good. This time I wanted to see if the model can also handle having something more than just a head. Having Formidable’s chest also in the image made some angles perform pretty bad, as there are as many ways of making anatomy as there are artists. Because of this, I removed all swimsuit and party skin images, as making her features was hard enough with her default skin, making the final dataset size some 1500 images. In the end, I’m pretty satisfied with the results, but I could prune the dataset even more and crop the images more homogenously as well as try a bit different hyperparameters (most importantly gamma) and stylegan3-t. However, I want to move into trying out Stable Diffusion model, so I will wrap this project up at least for now and post this. There is a psi hyperparameter used in this video generation, that determines how “creative” the generator might be, i.e. how far from an optimal statistical distribution it can go at any given datapoint (video time in this case). With psi=0 you have almost static video, and with psi=1 wildly varying results of which half aren’t even recognizable, and some are really good. I settled for 0.65, which I think has some nice variety with a reasonable amount of bad morphs.

Refer to the training configurations to choose the correct values for your hardware, config and dataset. Reducing the --batch value typically warrants increasing --gamma and/or lowering the D and G learning rates (--dlr and --glr). Think of --gamma as the “randomness” or “creativity” of the model.

Heck to train StyleGAN, an AI image style transfer model, requires 1-8 GPU’s with a minimum of 12GB of VRAM to train. Using a single Tesla V100, which is about £6-10k, it still can take upwards of one minute to process 1000 256x256 images.

We will also need two more libraries: OpenAI’s CLIP GitHub repo, enabling us to generate image features with the CLIP model, and the umap-learn library, which will let us apply a dimensionality reduction technique called Uniform Manifold Approximation and Projection (UMAP) to those features to visualize them in 2D:

Biggest missing feature for all these self hosted photo hosting is the lack of a real search. Being able to search for things like "beach at night" is a time saver instead of browsing through hundreds or thousands of photos. There are trained neural networks out there like https://github.com/openai/CLIP which are quite good.

In computer vision, this is known as zero-shot learning, or zero-shot prediction, because the goal is to generate predictions without explicitly being given any example predictions to learn from. With the advent of high quality multimodal models like CLIP and foundation models like Segment Anything, it is now possible to generate remarkably good zero-shot predictions for a variety of computer vision tasks, including:

(Github Repo | Most Popular Model | Paper | Project Page)

the first thing that comes to mind is CLIP: https://github.com/openai/CLIP

Whenever I’m working on semantic search applications that connect images and text, I start with a family of models known as contrastive language image pre-training (CLIP). These models are trained on image-text pairs to generate similar vector representations or embeddings for images and their captions, and dissimilar vectors when images are paired with other text strings. There are multiple CLIP-style models, including OpenCLIP and MetaCLIP, but for simplicity we’ll focus on the original CLIP model from OpenAI. No model is perfect, and at a fundamental level there is no right way to compare images and text, but CLIP certainly provides a good starting point.

in the modal card it says: pretrained text encoders (OpenCLIP-ViT/G and CLIP-ViT/L).

Command: "C:\StabilityMatrix\Packages\stable-diffusion-webui\venv\Scripts\python.exe" -m pip install https://github.com/openai/CLIP/archive/d50d76daa670286dd6cacf3bcd80b5e4823fc8e1.zip --prefer-binary

CLIP might work for your needs.

pip install deepfloyd_if==1.0.2rc0 pip install xformers==0.0.16 pip install git+https://github.com/openai/CLIP.git --no-deps pip install huggingface_hub --upgrade