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
Color correction and better posing
NoobAI EPS v1.1 + Illustrious v2.0 stable DARE applied
FAQ
Comments (27)
You really cooked with 10.0, this might be your best one yet
The anticipated version 10!
I have to admit, ever since trying this model from version 3 to now 10, huge congratulates to Crody the legend that is still releasing models that are actually decent AND consistent (you don't see that every day)
I never really thought that this model would be my main driver when I tried it out for the first time
thank you and hope you're doing well, Crody!
I've tested this V10 model, and the results are excellent. Limb generation is stable, prompt responsiveness is positive, and the composition sense/aesthetics have significantly improved over V9. At a time when IL-series models have largely stopped development, this update gave me some pleasant surprises. Indeed, it's currently the strongest 2D flat anime-style IL-series model
Is 10.0 ver still based on illustrious 2.0, or you it changed based model?
Base model is same as usual: NoobAI EPS v1.1 base, Illustrious v2.0 stable DARE applied
This results in great adherence and posing
For LoRA, you can use both ones for NoobAI EPS and ones for Illustrious
Crody Thanks for answer and for you checkpoint!
Crody Btw I have another one question: Is there any chance we ll get an FP32 (13 gb) version of Nova Anime in the future?
Jenneth Unfortunately no
There's not much purpose with fp32 since everything can be done with either fp16 or bf16
fp8 however is promising if only PyTorch fully support it
Crody To be honest, as a user who generates a LOT of images and often experiments with different settings, I see a HUGE difference in results between the 7GB and 13GB models.
But I believe what you say, and I dont rule out my skill issue
So dont care about my speech
I'm wondering if v10.0 includes the new and updated danbooru dataset?
I merged some newer models so probably?
I cannot say much about datasets but I would say that without OnomaAI (the ones created Illustrious) open source the latter versions like 3.0, there's no significant updates with datasets
Very good model!
The latest v10.0 turned out really solid.
Pros? Best detailing and prompt adherence out of all the model versions so far.
Cons? Just a few:
1. Sometimes it spits out really bad low-detail art. Chance is <5%.
2. Anatomy is still a problem. Both hands and legs can look unnatural. I try to reduce the chance of bad limbs with negative prompts, but of course it’s impossible to fully eliminate anatomy issues. Chance is >20%.
3. Color saturation is unstable. Happens in about 3 out of 10 gens. What does it look like? The model outputs an illustration with almost white skin tones, even though I set the prompts for warmer colors. I assume this might be a LORA issue with the character, but I’ve noticed it with more than one LORA. I’ll try some tests without LORA and update the feedback. Chance is >20%.
4. Struggles with backgrounds that include lots of people. No matter how hard I tried, it’s very rare to get multiple people properly generated in the background.
Despite the cons, I’d still say the model remains super easy to handle and adjust.
Novinha
amazing model.
How to get dark skin girls, normal skin guys, V10 version is hard to get
How to fix blurry eyes?
If you didn't specify the eye color, do it
Other than that, adding detailed eyes helped me a lot when I got blurry eyes
It is a common thing in both anime models and even non-AI fanart (which is undoubtedly why some models are doing that).
By using a face detailer ?
@Encolpe724 got better results with ADetailer. The only problem that it changing eyes color.
amazing model!
❤❤❤
this model is my waifu ♥
Is there a list of the new characters?
I don't
The models I used with DARE merge had all of those additional datas thus I actually don't know what characters added in (probably around late 2024 - early 2025 characters are added)
But I'm certain that newer posing like 'facing to the side, looking at viewer' works