Been a while since any updates - Is this project now done and dusted?
I love the QR Code models, both v1 and v2 for SD 1.5 and the SDXL Model too! Was the whole reason I got into AI/Generative art as i believe it was the first thing to provide what I considered to be the perfect balance of novelty and purpose, moreso than any other AI tool had done at that time.
I think after the 2 years since its release - there is still a lot of mystery around HOW this model was trained, what techniques were used to create the dataset, what parameters were used, how any images had to be made prior to the model being trained etc
Myself and many others would still love to see an answer to that question, and I know that it might help to further development of different types of models i.e a Flux or Hi-Dream compatible versions of the ControlNet - to keep inline with the trends of gen art.
Are we able to get some words to describe the process used, or at the very least, would you make the dataset public now? no reason not to after all this time i'd think.
Is such thing as a Flux compat ControlNet model even a possibility or am i being rather silly in thinking that?