Anima Version is Available!!
Nova Anime XL
Nova Anime XL is Nova Anime: Anime/2.5D/3D checkpoint model for SDXL
Pony version aims to get the similar look as both Nova Anime and Nova Domain
All images on example were created with diffusers with custom png tags
PixAI - IL v4.0 Happy Valentine
PixAI - IL v3.0 Happy New Year
PixAI - IL v2.5 Merry Christmas
Rules
You cannot use the generated images for commercial use if it's not edited (or just turning it to black and white)
You can share images without any restriction if you don't monetize it
Advertising this model to outside is always welcome
Recommend Settings
Sampler: Euler a
Steps: 20~30
Clip Skip: 1-2
Denoising Strength: 0.65 - 0.8
(Pony)
CFG Scale: 5~7
Prompt: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, source_anime, BREAK
Negative Prompts: score_4, score_5, 3d, jpeg artifacts, username, watermark, signature, normal quality, worst quality, large head, low quality, text, error, missing fingers, extra digits, fewer digits, bad eye
(Illustrious)
CFG Scale: 4~6
Prompt: masterpiece, best quality, amazing quality, 4k, very aesthetic, high resolution, ultra-detailed, absurdres, newest, scenery, {Prompt}, BREAK, depth of field, volumetric lighting
Negative Prompts: modern, recent, old, oldest, cartoon, graphic, text, painting, crayon, graphite, abstract, glitch, deformed, mutated, ugly, disfigured, long body, lowres, bad anatomy, bad hands, missing fingers, extra digits, fewer digits, cropped, very displeasing, (worst quality, bad quality:1.2), bad anatomy, sketch, jpeg artifacts, signature, watermark, username, signature, simple background, conjoined,bad ai-generated
Description
Better posing and hands
NoobAI EPS v1.1 + Illustrious v2.0 stable DARE applied
FAQ
Comments (67)
Can this model be used to train lora?
Yes
Is the new version could recognize newer characters? (e.g. Cantarella from Wuthering Waves or Castorice from HSR)
Probably but I cannot guarantee
Спасибо за обновление. Обновление обнаружил случайно, почему-то не приходят уведомления об обновлениях, хотя подписан.
В правом углу есть колокольчик, я всегда смотрю уведомления на подписанных авторов.
@gausssidorov928 Я знаю. Но почему-то уведомление об обновлении не пришло
LESSS FUCKING GOOOOOOOOOOOO
Better hands! Common massive W for this model! Now, I'll be off trying to fix my prompt, the hair isn't listening to me
So far, this is the most accurate model I have used for generate poses.
Found a new favorite model!
This checkpoint is the very best here
Твой проект самый лучший, все также на высоте, развивай его дальше
I Love This Model <3
Не способна генерировать адекватных мужских персонажей. Как и большая часть прочих нейронок, создана для генерации девочек-писичек. Знает всех женских персонажей тайтла, при этом не знает ни одного мужского.
Я знаю, что Nova Anime XL не очень хорош для создания мужских персонажей. Попробуйте использовать Nova Orange XL.
Женские киски спасут мир ))))
Датасет содержит больше половины женских персонажей. Конечно модель будет лучше их знать. Тут уже вопросы к художникам) Спрос на тянок больше
@Togovikss Модель создана путём слияния нескольких моделей. Большинство моделей обучаются на наборе данных Danbooru. В Danbooru больше изображений женских персонажей, чем мужских, поэтому результаты для женщин лучше, чем для мужчин.
A really decent model with the only unsolved problem is "floating coloring". Between normal generations, you may get high contrast or high saturation or warm color. It may be possible to fix it completely with the help of VAE or/and negative, but I would like stability "out of the box"
Great model. To make the best of it, be sure you use booru tags in your prompts. Makes all the difference.
why it dosnt work well in comfyUI? the colors is messed up and the images so bad. anyone know why?
Can I see the parameters like sampler, cfg scale and prompt?
@Crody I encountered the same issue with the following parameters: CFG scale at 5.0, Euler sampling method, 20 steps, and denoising strength set to 1.0.
@TTeslaThunder Denoising at 1.0?
Set it around 0.4 for both i2i and hires fix
Also, use Euler A (Euler Ancestral) instead
My parameters:
Steps: 20
CFG: 4.5
Denoise: 0.4
@Crody Switching to the Euler_a sampler resolved the issue perfectly - my sincere gratitude for your assistance.
@Crody i got same problem, the images is just blank green color image, i used the parameters you used, and this is my prompt: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, source_anime, BREAK, beautiful scenery nature glass bottle landscape, , purple galaxy bottle,. maybe thats a vae problem idk tbh
@TTeslaThunder can you tell me the nodes you using?
@hakuu00 For Illustrious, don't use score tags
Score tags would work to improve the image but while doing it, it'll add text 'score' in image
Also, some words would make image worse because of calculation for weights in sdxl sometimes goes negative
Find the word and replace it with others
@Crody still dosnt work, i still get a green image, i use the comfy build in basic image generation workflow btw,
@hakuu00 Try using this workflow
https://civitai.com/models/1386234/comfyui-image-workflows?modelVersionId=2009563
If this didn't work as well, try redownloading the model using aria2
@Crody is there a more simple workflow?, bc this one is really overwhelming tbh :/
@Tofu080 Use this one
https://github.com/Faildes/comfy-hires-color-adjust/blob/main/t2i.json
Always generate 4 fingers
Try using negative prompts like fewer digit, missing fingers, bad hands
why cant you download novaAnimeXL_ilV50 anymore?
the person in the image always has their back to the viewer and is looking back. why?
also the image is heavily tinted a color
Great model for vanilla and older anime style - especially given the orientation taken by other models.
Please tell me about commercial use.
It says, “You cannot use the generated images for commercial use if it's not edited (or just turning it to black and white).”
However, would adding a little gradation be considered editing?
Editing something like cropping, channel editing etc.
Adding gradation is considered to be editing
Cropping and some fixes in details additionally would be great
Thank you for your prompt comment. I can't do any technically complicated editing, but I think I can do that much without any problems. It's like a dream to be able to use this wonderful model!
With all respect to the author — and honestly, this is a great model!
The model uses the Illustrious License (FAIPL-1.0-SD).
This license clearly says:
“The output of this software is not covered by this license, and no contributor claims any rights to it.”
This means: after you generate an image, it fully belongs to you. You can use it commercially, unless you break some very basic rules (like hate, doxxing, etc).
The rules in the model’s description (like “edit before commercial use”) are the author’s personal wishes, but they are not legally required.
They actually go against what’s written in the official license on this model’s page.
So you can use the pictures as you like. If you want, you can credit or thank the author, but you don’t have to.
Just always check the license field on the page — not just the description.
Thanks again to the author for this awesome model!
You can find the license in the “Details” section of the model page, at the bottom right.
It’s shown as License: Illustrious License — you can also click the link there to read the full license text.
Additionally, there’s a box at the bottom right (titled “This model permits users to:”) that summarizes what the license allows.
For this model, you can see it permits commercial use and selling images you generate, according to the Illustrious License.
Waiting for v10.0. Return colorful saturation from v8.0, please. In version v9.0 saturation is much lower than in version v8.0.
good
Does this model has cyberfix?
Probably not but still, the model is capable of maintaining finger posing while doing extreme structures which you can see on most of sample images
Crody How did you improve the hands?
Shinwoh By merging several models just for IN00: the one that makes finger details
Stable model (NTR Mix for example) as base, merging several models with different posing using artist LoRAs with 0.1-0.25 IN00 alphas results in stable and unique hands' posing
Crody oohh, if you apply cyberfix to this model it would be dope
I use raehoshi and it includes cyberfix, it improves the hands and feet a lot
this good
v9 is better than v8 at following prompts, well done
Edit: v10 is even better
Still waiting for v10. v9 is more detailed than v8, but the colors feel a bit too dull and inconsistent. I still really like how v8 draws — the skin tones come out with such a smooth gradient. v9 feels a lot sharper. You can see a similar difference between v6 and v4 — one looks soft and smooth, the other more harsh and defined.
Currently I have some issues with GPU
v10 is coming soon
Sample images:
https://ibb.co/9mNKGSgg
Crody I cant wait!! You're the only creator im going to use a model for in the future, nothing else compares to your work! V9 is so easy to work with and always incredible results. Do you have an ETA on v10?
heyoai I cannot guarantee about the date but I'm sure it'll be in this month
Crody Releasing limited time sample images? damn I missed it, bet it's good tho
Will you do an anime finetune of Qwen Image?
Love it!
VERSION 10 AT LAST :OOOOOOOOOOOOOO
I like NovaAnimeXL, what is the changes on v10?
As requested by others, coloring is similar to pre-IL v8.0 while posing gets better
v10 is fun.
Thank you so much, I love your models.
V10 is the best yet. Vibrant colors, great overall composition and image structure, great with hands, and amazing prompt adherence. Thank you for creating this!!
Details
Files
novaAnimeXL_ilV90.safetensors
Mirrors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
NovaAnimeILV92.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
novaAnimeXL_ilV90.safetensors
n0v4_9.0.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.



