In photography post-production techniques, some parts of black and white photos have color. This technique is called selectivecolor
Description
first commit
FAQ
Comments (4)
And you decided that needs a dedicated checkpoint with more than 6 GB. A LoRA wouldn't be good enough?
In hindsight I think this might have been a bit too hard for a comment. This is a topic I'm actually interested in, so excuse my disappointment. I think this would be a good topic for a LoRA, which allows more flexibility without beeing limited to a certain checkpoint. It also would save a lot of disk space if you have already a few models downloaded.
I don't mind the checkpoint much, but it's a very restrictive use for a checkpoint. So please provide some more example, so people can see if it's worth it.
i dont think it really matters, if you want the model u will still download it, ive run the model on its own and it does everything quite well. So maybe because the Uploader i think is from china or korea, perhaps they did dreambooth training and then uploaded the full checkpoint (perhaps they dont know what a LORA is and again used dreambooth to train the model further) so perhaps instead of firing ones mouth off first we try and understand why it is first ? Online politeness is something that many internet users just dont possess.
@Stablediffusionlover Yes, I already expected that, and I wouldn't mind much if there were some good example pictures shown at least. If you upload such a huge amount of data, is it too much to ask for some more pictures so that people can judge if it is worth the time and bandwidth?
I see so much of these low effort, one JPG example stuff recently that it's getting frustrating. This one has at least some description posted to it, I'll give it that.
was it dreambooth training or LORA merged ?