v4.0
(The NEW label is misleading, it's an update.)
Trained with the goal of enabling multi-subject capability but didn't turn out well. At least there's a positive probability of multi-char working. Still need to figure out why it fails.
In addition to the character tags, custom style tags were added:
fanart: fan arts from Danbooru (optional, neutral style)u149ani: screenshots from the airing U149 animeapuri: Cinderella Girls and Starlight Stage mobile game art (somehow dominated by early CG style, will probably retag in the future)
As the outfits were not manually tagged, it should not be expected they can be replicated.
Training data can be found in my data repo. Training details will be properly documented when I have time. Would appreciate suggestions or any sort of feedback.
Description
Conv + Linear weights. More bytes but not necessarily higher quality
FAQ
Comments (8)
uoh!!!
finally!!!
Very excited to see. We DO need SOMEONE to post pics of each with DATA in them. None so far.
Civitai doesn't show txt2img params containing non-ASCII characters (a fact I did not know when making this). The info is still there; you just need to download the images and use PNG info.
説明不足過ぎて分かりません。どうやったらこのLORAを使えるのでしょうか？
With the required extension installed, just prompt with the trigger words and use related tags as necessary. The Hadamard Product LoRA should work similarly to a normal LoRA. I made a bad decision by using Japanese names as trigger words; they don't show up in Civit's UI but you can download the preview images and use PNG info to see what the example prompts are like.
Well the data was there last time I checked. Guess the site changed something. I'll update the description.
拡張機能のダウンロードがよく分かりません。
@jiangpingzhaiteng708gmailcom976 Found a guide. Just use the extension url and restart.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.