ExponentialML

Okay, so I came up with a solution to train multiple subjects into one model. I had an idea where you assign a placeholder token to a specific image, and...

The basis for a lot of your questions are addressed at this [issue](https://github.com/rinongal/textual_inversion/issues/7).

One thing I found that helps and/or fixes this scenario is adding periods to your prompts, not commas like in original SD repo. This may or may not be a...

The two things that had the most success for me are: 1. Replace the template string with a single `{}` 2. Make sure you're using the `sd-v1-4-full-ema.ckpt ` I'm almost...

W̶e̶l̶l̶,̶ ̶t̶h̶i̶s̶ ̶c̶e̶r̶t̶a̶i̶n̶l̶y̶ ̶i̶s̶ ̶a̶n̶ ̶i̶n̶t̶e̶r̶e̶s̶t̶i̶n̶g̶ ̶d̶i̶s̶c̶o̶v̶e̶r̶y̶.̶ ̶ ̶ ̶ S̶o̶ ̶t̶h̶i̶s̶ ̶c̶o̶u̶l̶d̶ ̶t̶h̶e̶o̶r̶e̶t̶i̶c̶a̶l̶l̶y̶ ̶p̶r̶o̶v̶e̶ ̶t̶h̶a̶t̶ ̶y̶o̶u̶ ̶n̶e̶e̶d̶ ̶t̶o̶ ̶f̶i̶n̶e̶ ̶t̶u̶n̶e̶ ̶o̶n̶ ̶t̶h̶e̶ ̶b̶a̶s̶e̶ ̶r̶e̶s̶o̶l̶u̶t̶i̶o̶n̶ ̶S̶t̶a̶b̶l̶e̶ ̶D̶i̶f̶f̶u̶s̶i̶o̶n̶ ̶w̶a̶s̶ ̶t̶r̶a̶i̶n̶e̶d̶ ̶o̶n̶,̶...

Other than pre-processing your inputs and removing the background, you can try messing with the `coarse _class_text` and `embedding_reg_weight` parameters, both together or one at a time. - `embedding_reg_weight` will...

> each of which could be used to judge where in the training process the overfitting occurs. In the DreamBooth paper, one of the key features to prevent overfitting is...

You can use `coarse_class_text` to better condition the model. I'm currently experimenting with this by using two generated images of a class, then using the `mixing_prob` to randomly choose a...

> > Hello, thanks for the implementation! It works very well. > > As a suggestion, it may be helpful to provide code that works with the broader Github community....

Hello. Try this out. In any of the swap scripts that your using, find the line that looks like this. https://github.com/neuralchen/SimSwap/blob/dd1ecdd2a718636d33977ab3097a69a0ecf080d8/test_video_swapsingle.py#L58 Then add this parameter at the end. ```python app...