Same here.

@lokithorodin - that's disappointing. I made an SDXL one that's okay but I really want one for Flux. Guess I need to start training Flux Loras now.

@leisure_suit_larry I use CivitAIs system. Works surprisingly well. Just take the images you've collected and let them write the description. It's almost difficult to create a bad one.

@lokithorodin - I've considered checking it out, by I just genuinely prefer to do it at home on my PC. I just couldn't add much heat to my space over the summer and took a long break from training, during which time Flux emerged and I just haven't gotten around to upgrading Kohya or installing Fluxgym or AItoolkit etc. etc. I've never given a thought to buzz and I've never generated or trained anything on site, but I've made dozens of loras and fine-tuned lots of 1.5 models, too. I'll check out what the site has to offer. Thanks. In the mean time you should go train a Raquel Welch Lora. If you need some images let me know. :) Happy Halloween.

@leisure_suit_larry Sure! If you want to zip up your collection of Raquel Welch images I'll run it through CivitAIs trainer. My buddy loves Raquel Welch and Wonder Woman so I wanted to surprise him with some great images. We'll just have to find a way to transfer the file. I'll make it public unlike all my other work.

@lokithorodin - Ok, these are the images I used to train my SDXL Lora. It's a lot of images so you'll probably just need to pick out the best for Flux, since it takes so much less data to train Flux. I hope this link works. I also hope the civitai trainer doesn't require images to be square. https://c.mail.com/@1177777928738449179/ZbmI3u3aTZq3l6AqlOMrew

@leisure_suit_larry Great! Downloading it now. No, square images aren't required. Thanks.

@leisure_suit_larry Got it. You can remove that zip file if you'd like.

@lokithorodin - Awesome. Look forward to seeing your work. Thanks.