Trained on Illustrious-XL(v0.1), generated on NoobAI-XL (V-pred-0.75S)
Recommended weights: 1.0.
Recommended to use highres fix.
For better eyes highly recommended to use:
pink pupils, blue eyesDescription
reworked tags
expanded dataset
FAQ
Comments (5)
Your Citlali lora has probably the best recreation of her I've seen, but in exchange the flexibility in giving her different outfits seems to take a hit. I tried putting her outfit tags in the negatives but it still applies the textures of her default outfit to the outfit that you are trying to prompt. Settings and dataset wise you seem to have enough variety to allow for flexibility, the only thing that I could identify that might be causing this issue is that you're training the dataset on 8 repeats per epoch. I think adjusting it to 1 or at most 2 repeats per epoch should fix the flexibility.
Apart from that, I think you should add 'colored eyelashes, facial mark' to the trigger words section. I don't think there's any need to redo tagging on the dataset since prompting them seems to work well enough.
Thanks for your feedback! I lack experience, so your advice is very valuable to me!
Maybe someday i'll make v5.0 based on this changes.
(P.S. maybe i misunderstood you, but i tried using different outfit (school uniform) and it goes well for me (you can see in examples))
Yeah that's probably just because the tag is that strong (a ridiculous amount of school uniform images on danbooru), I uploaded some images to show what I mean too
@5e5 yep i see now, you are right.
Maybe i'll try to fix that in future.
where has this lora been my whole life?😭 like the others been driving me crazy not looking good til I met this lora great job thank you!!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.