V3-V4 info: Two more versions trained with more recent features. Both was trained with T5 and clip-l encoders. V3 was trained with long prose from joycaption. V4 was trained with trimmed captions from torii captioner. Between those two I personally prefer V3.
V1-V2 info: Two versions, first one is trained with 512 res and booru captions for 2000 steps and second is 768 res trained for 5000 steps with nlp captions. Can't say which one of them is strictly better, probably second one, since it's transferring style more harder even with another char lora. Haven't tested it with anything else than base checkpoint in fp16/fp8 mode and afaik rn it will not work with some quantization types in popular ui's.
Workflows with prompts are right inside the images, you can just drag them into your comfyui browser window, probably could work better with "anime screencap" in pos and "realistic, 3d" in negs", negs unavail with 1 cfg, but you could hack it with this extension, settings for which is also into the flow already.
Description
NLP tagged 768px version
FAQ
Comments (13)
How do you even use it? You left no promt or instructions. For me this lora doesn't change the style in any way, maybe some words should be used?
Had to use a third party service to see your workflow because I don't use comfy. Lora excellent quality, thank you for your efforts but please be more clear on how to use your product next time
You can copy the nodes here on Civitai, paste it in a text editor and search for "prompt".
Well, it's just a basic thing with comfy, when you want to see the workflow you could just drag image to your comfy window in browser, I didn't think that it should be mentioned, but ok, glad you liked that eventually.
yes please share actual prompts for examples, I don't use comfy either
anime screencap, 1girl, animal ear fluff, animal ears, black choker, blue skirt, holding book, brown hair, cat ears, cat girl, choker, foot out of frame, frilled skirt, frills, hair between eyes, hair ornament, hairclip, holding book, indoors, knee up, legs, long hair, open mouth, purple eyes, shirt, skirt, solo, two side up, white shirt, book title is "lora bake for dummies", sitting on chair near pc, dummy face, reading book
One of the promts
Looks like "anime screencap" is the only tag needed for lora activation
I've tried different prompts, probably yeah "anime screencap" in pos and "realistic, 3d" in negs should work better, negs will only work when you hacking it with dynamic thresholding.
Excellent cook 👍You mentioned that you used nlp captions in the second version in which case I would expect long sentences describing the character and scene however in the nodes of the example images there are still booru tags. Did you train with long sentences but used booru tags for the example images?
Edit: Never mind, I guess I was on the wrong page when I copied the nodes 😅
Yes, some of them still prompted with booru tags, some with long sentences.
Tags are working so well, this is like prompting on animagineXL/pony, I love it, please make more <3
Will try to do some more sovlfull artists.
This looks very promising! 👌💕