Emm...
Better set 0.6 it's recomend
Noted : dataset sample By. Vienna Art for clothes before trained
Outfit :
Armor
Shirt
Skirt
Trousers
Jacket
Robe
etc.
This is just a concept as if the clothes are branded, depending on the fit, please try, here I learned to make LoRA clothes, so please help me review, if it's not good I admit it because I'm having trouble and maybe I will unpublish it if you don't like it because this is BETA.
Thanks Enjoy to Generate.
Description
FAQ
Comments (17)
Wow! Result are amazing👌, Can you share parameters this trained on, I trained some clothes couple times and i had very bad results, and you've made such good clothing training, would be amazing using some good parameters
but for sure results on clothes, I always emphasize the sample clothes in the picture and duplicate them a little and hopefully reinforce the training of each shirt model, e.g fot trashhold learn just 0.70
and always croscheck caption and triger, it's better to have almost nothing in common in each trigger to keep it as simple as possible, but emphasize what you want to display, hope it helps a little.
@Random_ChaRAREsearch i see i'm new too, i mean the setting this clothes trained on, or maybe you can send me the tutorial you followed maybe there i could find parameter
@ThexSam which one do you use Web or Kohya Colab, for Colab you can use the following references: https://www.youtube.com/watch?v=rozKiRk16gg
i just using colab, very lazy to install directly on the PC :D
@Random_ChaRAREsearch Thank u so much
@ThexSam welcome dude
I hope you intend to add more clothing brands to this because I love making characters wearing streetwear or urbanwear outfits so this'll definitely come in handy! Uniqlo comes to mind ;)
wahaaa ><, thanks, great input, I'll add it and look for different styles from similar models for more variety.
Thank you so much :) I can't wait to see what you come up with :D
Yo this LORA looks sick but does this works well with realistic or semi realistic models too?
not sure, because my dataset and trained using Anything v3.1, so don't expect much, maybe you're the one who can give it a try, btw thanks for question :).
Yes it does !!!! is sick !!!! very well programmed/trained
@Random_ChaRAREsearch Yes it does is sick very well programmed/trained !!!!
@Random_ChaRAREsearch I wish I can try the new update but unfortunately the latest Auto111 version broke my Stable Diffusion...again. 😭
its working great for me!!! ill post some stuff when i get some more results
Thanks for review :D
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.