OK! Finally dropping my new #DreamBooth fork--you can now use BLIP to caption your dataset, then extract image captions to txt files. This lets you train on the text/image pairs instead of providing general instance prompts.
#aiart #stablediffusion
github.com/KaliYuga-ai/Dr…
#aiart #stablediffusion
github.com/KaliYuga-ai/Dr…
You can ALSO specify instance and class prompts to train on *alongside* the image captions. they get appended to the image caption in the text file.
You can train multiple concepts at once, too! I've been getting some next-level results out of it so far.
LMK how it works for y'all!
LMK how it works for y'all!
as a fun note, I did the coding mostly with chatGPT and it's phenomenal how well it works, even if it took a lonnnnnng time to get close to right and then debug 🪲. I'm sure a coder could have done it faster, but I can't believe I could do it at all!!!
• • •
Missing some Tweet in this thread? You can try to
force a refresh
