https://discord.gg/TajBR2Dqdy is the link to discord, there are two channels, Chinese and English
QQ群：451450868（主群）
929721518备用群
v7 tensor onsite use linkT-illunai3 - v7 | Stable Diffusion Model - Checkpoint | Tensor.Art
v7国内线上出图平台T-illunai3 - v7 | Stable Diffusion Model - Checkpoint | Tensor.Art (tusiart.com)
This is a donation channel Thank you very much for your support！ All the funds will be used to purchase 5090d！Buy Tonade a Coffee. ko-fi.com/tonade - Ko-fi ❤️ Where creators get support from fans through donations, memberships, shop sales and more! The original 'Buy Me a Coffee' Page.
Tonade正在创作T-ponynai3模型作者，c站id：Tonade，现在是T-illunai3作者 | 爱发电 (ifdian.net) 这是国内爱发电的链接，感谢捐赠支持！
NOOB 1.0版本 1000 单画师测试不可修改版 (qq.com)
可直接对照参考的画师提示词文档，可在illu系模型上完美使用，包括illunai-v7
NOOB 1.0版本 1000 单画师测试不可修改版 (qq.com)It can be directly compared with the reference artist's prompt word document and can be perfectly used on Illu models, including illunai-v7
v6推荐使用的提示词（懒人版）
Recommended prompt words for v6（Lazy version）
POSITIVE：best quality,amazing quality,very
aesthetic,absurdres,masterpiece,
NEGATIVE：bad feet,missing toes,extra toes,lowres,bad anatomy,bad hands,text,error,missing fingers,extra digit,fewer digits,cropped,worst quality,low quality,normal quality,jpeg artifacts,signature,watermark,username,blurry,4-koma,sketch,(bad:1.05),error,fewer,extra,missing,worst quality,jpeg artifacts,bad quality,watermark,unfinished,displeasing,chromatic aberration,signature,extra digits,artistic error,username,scan,abstract,(blurry, lowres, low quality:1.1),ugly,(big head, bad anatomy:1.05),artist name,(watermark:1.1),
v1只是一个初步进行微调的版本，但是效果好的惊人。在动漫绘画的领域比pony的微调效果要好上不少，肢体的正确率也十分可观。希望各位能多多测试一下这个微调。
基于illustrious-xl Illustrious-XL - v0.1 | Stable Diffusion XL Checkpoint | Civitai
V1 just a preliminary version with fine-tuning, but the results are surprisingly good. In the field of anime painting, the fine-tuning effect is much better than Pony's, and the accuracy of body movements is also considerable. I hope everyone can test this fine-tuning and give some comments about it.
base on illustrious-xl Illustrious-XL - v0.1 | Stable Diffusion XL Checkpoint | Civitai
训练须知：
v1版本使用了100张经过筛选的高质量的使用nai3生成的图片，以及画风相近的一些公开素材作为姿势的补充，训练使用了prodigy算法进行训练。illustrious模型的女性素材很多，导致了模型会比较难以生成男性的图像，而且某些时候需要一些比较复杂的提示词才能生成目标图像，这将在之后的更多训练集的版本尝试进行一些改善。
Training instructions:
The v1 version used 100 high-quality images generated using Nai3 that had been filtered, as well as some publicly available materials with similar art styles as pose supplements. The training was conducted using the "prodigy" to train. The illustrious model has a lot of female materials, which makes it difficult for the model to generate male images, and sometimes requires some complex prompt words to generate target images. This will be improved in more training set versions in the future.
在v4中，我使用了600张由nai3生成的风格图片作为默认画风的训练，这一步是基于illustrious模型的，然后我用模型融合的方法在这一fine-tuning后的模型上加入了一些noob的模型成分，以来平衡illu的一些缺失的语义和识别prompt的能力。
In v4, I used 600 style images generated by nai3 as the default art style for training. This step was based on the illustrious model, and then I added some noob model components to this fine tuned model using model fusion to balance some of Illu's missing semantics and prompt recognition ability.
v6这个版本的模型进行了更好的画风调教，改善了大量的基于ntrv4的错误肢体和过拟合的画风，并调整了过细的线条，让整体的线条和效果更好，甚至配合画师提示词使用已经足以媲美nai3，在某些画面构图上已经超越了nai3
推荐使用的提示词都在例图里面，总体来说ntr模型系列的提示词都可以直接使用，这将是一个新的开始，往后会不断强化这个版本
This version v6 of the model has undergone better art style tuning, improving a large number of erroneous limbs and overfitting styles based on ntrv4, and adjusting overly fine lines to make the overall lines and effects better. Even when used in conjunction with artist prompts, it is already comparable to nai3 and has surpassed nai3 in some picture compositions、
The recommended prompt words are all in the example image. Overall, the prompt words of the ntr model series can be directly used, which will be a new beginning and will be continuously strengthened in the future
推荐使用质量词Recommend quality words
正面提示词
best quality,amazing quality,very aesthetic,absurdres,beast quality,masterpiece,artist:xxx
负面提示词
lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry,4-koma, sketch, (bad:1.05), error, fewer, extra, missing, worst quality, jpeg artifacts, bad quality, watermark, unfinished, displeasing, chromatic aberration, signature, extra digits, artistic error, username, scan, abstract, (blurry, lowres, low quality:1.1), ugly, (big head, bad anatomy:1.05), artist name, (watermark:1.1)
Recommend using quality words
Positive prompt words
best quality,amazing quality,very aesthetic,absurdres,masterpiece,artist:xxx
Negative prompt words
lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry,4-koma, sketch, (bad:1.05), error, fewer, extra, missing, worst quality, jpeg artifacts, bad quality, watermark, unfinished, displeasing, chromatic aberration, signature, extra digits, artistic error, username, scan, abstract, (blurry, lowres, low quality:1.1), ugly, (big head, bad anatomy:1.05), artist name, (watermark:1.1)
在线生成网址 online generation link: (340) T-illunai3 - v4 | Stable Diffusion Model - Checkpoint | Tensor.Art (tusiart.com)
ps: For some unknown reasons, my model cannot run online on Civitai, even after multiple applications, it is useless. Therefore, I uploaded it to this platform synchronously, and online generation of the model can also be done on it
出于一些未知的原因，我的模型无法在civitai上在线运行，即使多次申请都无济于事，所以我同步上传了这个平台，在上面同样可以进行该模型的在线生成
Description
更好的手指，更好的足部，更好的画风，更好的多人，更好的图像细节，更好的稳定性，更好的基础画风
Better fingers, better feet, better art style, better multiplayer, better image details, better stability, better basic art style
FAQ
Comments (13)
好模型😋就是推荐用哪个采样器和调度器效果会比较好？
一般来说用euler a就行，高分算法用r-esrgan 4x
Thanks for making great model. can you make it runnable by civitai generator?
Already applied. civitai officials said they can't accommodate any more base models being generated online
I love your models in general, you always put great work on them. ive notice how much variants and different details can be achieved by using artists names. Being honest i dont know Artist in general, and If i can i would like to ask, is there a way to learn or know what artist names affect the images, of course i know i cant ask you for a sample image to see each style xD. but is there any list were i can see the Artist tags?.
@XDDDAI Thank u so much :3
v7 did improve a lot but I feel like the dynamic aspect and creativity/variety of the composition has became less with each version. v3 has the best on this part, could you perhaps look into this aspect if you're updating a new version?
I will train some multiplayer images in the next version to improve the semantic understanding ability of generating multiplayer images. As for the composition issue, it should be because the impact of random composition in the default art style has been reduced, but the model's ability to run better and compose should have been improved because ntrmix is used as a supplement to composition, and the understanding of some perspectives will be better than in previous versions
@Tonade Yes I should specify the composition issue is mostly with multiplayer image or larger image that shows full body with background. For example if you're prompting 2 people standing, v3 can make image with 2 people standing with more layered composition (like one near and one far to the camera), while v7 the two people will mostly stand at the same distance. Sorry for nitpicking, I think other than this part v7 is a very high quality model!
@niniconcon Yes, the parts you mentioned are indeed troubling me. I am currently trying different methods to improve the problem of poor composition in multi person images
越來越強大了，做得好!
At first I was concerned about v7. The base art style of v7 is a little bit different from v4 which had a strong resemblance to the Pony based checkpoint that I like. v7 has a bias that is a little bit glossier than v4 by default, there is an emphasis on bounce lighting.
Still, it's not something I dislike. It's also a little bit easier to get small chested mature looking characters with v7 which I really appreciate.