Video - https://youtu.be/gmnv6CYCQrI

Overview

Z-Anime is a Z-Image based model you can using in ComfyUI with a little as 8GB VRAM. If you like Z-Image but prefer a tilt towards the anime style, then this one is for you!

Huggingface page - https://huggingface.co/SeeSee21/Z-Anime

They say their key features include:

✅ Full fine-tune on Z-Image Base — not a LoRA merge

✅ Rich anime aesthetics with strong style diversity

✅ Natural language prompting — works best with descriptive prompts, not tag lists

✅ High diversity across characters, poses, compositions, and layouts

✅ LoRA training ready — strong base for further fine-tuning

✅ 8GB VRAM compatible

✅ GGUF variants available

✅ AIO variants available (Base, 4-Step, 8-Step)

Also included are some different CLIP models -

Engineer V4 (qwen_3_4b-engineer-v4-*) — an alternative full fine-tune of the Z-Image text encoder by BennyDaBall, drop-in compatible. Often produces more varied outputs from the same seed.

Workflow Information

There are 2 workflows attached to this post. One is for just a single model Z-Anime (or Z-Image) and the other for using 2 models, such as Z-Anime and ERNIE-Image.

Model Downloads

The model download links are provided in-workflow, so simply download the ones you prefer. There are plenty to choose from as well as a variety of CLIP models.

Custom Nodes

If you don't have them already, you'll also need to install a few custom nodes using ComfyUI Manager, or the usual git clone method:

Model Loader Groups

The default model loader uses the split files:

If you'd prefer to use the AIO checkpoint or GGUF model files, then you can instead connect those to the Z-Image SET nodes - or simply delete those groups if you'll never use them!

Size & Steps

The 4 & 8 step models are super fast and need a CFG of just 1, though you can go a little higher if you want your negative prompt to come into effect:

For the base model you'll want a CFG above 3, and steps above 20.

Prompting

Z-Anime prefers fairly long, descriptive, natural language prompting. Typing long prompts can be hard on the paws, so I've also included the ERNIE prompt enhancer.

It is slightly more lightweight than using ollama, and needs only native ComfyUI nodes, though isn't quite as powerful. Feel free to change the system prompt!

For more variety you may also wish to change the random number from fixed, as well as the other values in the TextGenerate node itself:

Generation Options

Select your desired generation options by flipping the switches to enable/disable the names groups:

If using the Image Compare options, remember you can change which of your generated images you'd like to compare by picking values from the GET nodes:

If you're using the 2-model workflow, you can select which stage you want to use the second model (ERNIE-Image by default, but you could use the original Z-Image Turbo, or other models if you liked as well):

The default is the first model (Z-Anime). Personally I think ERNIE-Image is a great upscale option!

Bookmarks

Each column also has numbered bookmarks available, so you can just press 1 to jump straight to the loading area, or 5 to just straight to output.

Example Images

"Create a British rodent gentleman character with a black moustache, red cape, top-hat, and Victorian shirt and trousers, and place him in a typical London City scene using a classic anime style. Surround his with an aura of greater power, and use a majestic pose for him."

"Generate a gothic looking human female character with long, black hair and a cool tattoo on one arm. She should be wearing a ripped, purple t-shirt which a rodent-face logo on it and torn blue jeans. She is working behind the counter of Game Stop. She has a scowling expression, and is looking straight at the camera. The counter in front of her is piled high with used X-Box games. The scene is a chibi anime style."

"A Mecha (Sci-Fi), sharp-edged, anime style image which features a towering, hyper-muscular man with a singular, round, red, ocular (eye) implant, brown hair, and an intricate, metallic black and red chest-piece. His arms are made of polished chrome. He is carrying a huge yellow, American-style bus above his head, the front of which is just visible. The front and back bus tyres are round, with a deep tread. He is in a New York street from the 1960s."

Enjoy! :)