![](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="96" height="96"><rect fill-opacity="0"/></svg>)
#NovelAI Image Generation Progress Showcase:
When you include tags in your prompts, you may produce more of the same character with greater consistency!
![画像](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="680" height="453"><rect fill-opacity="0"/></svg>)
20
566
1,320
会話
返信先: さん
That's so cool! At first i thought you've meant like hashtag of a character for example # Rem and you've used a Textual Inversion to train an anime module on a thousands of copyrighted characters, which would be cool, but I don't know how legal that is. But that's even better!
2
2
11
返信先: さん
Since we are training on Danbooru, it also learns character names and their visuals. You can prompt for "masterpiece portrait of smiling rem, re zero, caustics, textile shading, high resolution illustration" and get this:
![画像](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="384" height="640"><rect fill-opacity="0"/></svg>)
返信先: さん
Oh my god! This is truly a masterpiece! Which Danbooru dataset do you use? Judging by some of the results of testing the Discord bot, it should be the Danbooru 2019 dataset or something like that. But maybe you're training on a 2021 dataset? If you have 5 TB of storage of course
1
3
返信先: さん, さん
There's literally a blurry signature at the bottom-right of the left picture. This is just theft.
3
4
141
返信先: さん, さん
Hey so how were the artists you trained this on compensated for all the Labor they put in?
1
284
返信を表示
返信をさらに表示
その他のツイート
![画像](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="623" height="680"><rect fill-opacity="0"/></svg>)
SPY×FAMILY
ヨル・フォージャー 
![画像](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="680" height="680"><rect fill-opacity="0"/></svg>)
502
8,178
14万
このスレッドを表示
![画像](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="680" height="592"><rect fill-opacity="0"/></svg>)
141
8,761
8.2万