If you want to support me, bidding buzz on the weekly auctions page for the model helps me keep it up for on-site generation.
Anima Base v1
Text encoder and VAE can be found on this HuggingFace page.
Recommended settings :
Positive prompt : masterpiece, best quality
Negative prompt : worst quality, low quality, jpeg artifacts, signature, sepia, fewer digits, extra digits, bad hands, bad anatomy, watermark, censored, score_1, score_2, score_3
Sampler : ER SDE / Euler a - 24 steps
CFG Scale : 4-5
I highly recommend using the ADetailer extension for better faces and eyes.
The Anima iteration of this checkpoint is a bit of a departure from the Illustrious v2 version, going for a slightly less flat style overall, but keeping a colorful, sharp 2D anime style for faces while getting a little painterly side to bodies and backgrounds, as well as shading.
Just like with other Anima based checkpoints, artist styles work, and natural language to enhance your prompts or add text also works.
Thanks to @ComradeAnanas, @Manityro, and @justTNP for helping create this checkpoint.
IllustriousXL/NoobAI EPS1.1
Recommended settings :
Positive prompt : masterpiece, best quality
Negative prompt : worst quality, low quality, bad anatomy, watermark, censored
Sampler : Euler a
CFG Scale : 5-7
I highly recommend using the ADetailer extension for better faces and eyes.
For v2, I merged v1 with some NoobAI models, @GarejiSama's AdAstra and @Manityro's Vermilion, making v2 much more compatible with NoobAI based LoRAs.
The baked in VAE is Anzhc's, which you can get here, gives a much sharper look to images, reduced noise, and makes colors less bland overall.
LustrousMix v2 generations are more colorful than v1's, with more contrast and sharpness. There is a slight art style change as well, with more mature faces.

Hope you enjoy!
Description
FAQ
Comments (19)
sorry man, i have never tried anima merge checkpoints before, and i am kinda lost because it doesnt work with the typical workflow for anima. am i missing something? probably right?
What do you mean? What's your workflow, and what's not working?
Have you installed the required components as per the original Anima?
Both Anima and other modified Anima Base have required Text_Encoder ( qwen_3_06b_base.safetensors ) and Vae ( qwen_image_vae.safetensors ).
@Hoseki never mind bruh, i had to update comfy and install all the requirements xd
lets fucking go Hoseki Anima😍
Do you know MorpBase.com ?
We did it reddit, we Made Anima Great Again!
My first impression is that the biggest advantage of Anima over Illustrious is its ability to understand natural language. I'm trying to use fewer Booru tags, or even none at all, and the results still turn out surprisingly well. With Illustrious, I often need to use a LoRA if it's existing.
Many people say that Anima has far more built-in styles, although I'm already comfortable with the styles in Illustrious. Still, let's see where this new journey takes us.
Better backgrounds, more booru tags as well due to more recent data, more variety in image composition amongst other things.
My favorite model so far.
Very Good Model!
good!
This is my first time using ComfyUI. I can successfully use the pre-built Anima workflow in ComfyUI, but I am unable to use this model. The workflow returns the following error:
AttributeError: 'NoneType' object has no attribute 'shape'
How can I resolve this issue?
Your most likely missing a specific item in the workflow that causes this, see if you can click the comfyUI manager and 'download missing', If that doesn't work, try making a new window, and dragging and dropping the preview image(s) from this checkpoint and it should give you a proper workflow. Other things it could be is that something isn't connected, something is outdated, etc etc. OR my biggest guess is that you need to click the diffusion model, and then click the model a second time (its just a weird bug where comfyUi says your using something but in reality its just an empty line, meaning you have to enter it.), If your still having issues would you mind telling me if you're on an older version of comfyUi, as well as what node in particular is circled red and or pink?
Pumpkin__man You don't get any errors until its in runtime. It happens on the Ksampler. I have noticed this when making a flux image and trying to upscale it using sdxl <- just an example of one time I have also gotten this error.
I've tried many anima models, and I'm calling this one the best!!!
👍👍👍👍👍
Trying Anima for the first time, through trial and error and my Adetailer fucking up my images with black boxes and having to get another fork. I can say Hoseki you've done it again, your stuff never disappoints. From a Forge NEO user


