Sign In

KuroTao Director - Creative Prompt Generator

Updated: Jun 28, 2026

toolanimepromptgeneratoraicomfyui

Download

1 variant available

Archive Other

16.79 KB

Verified:

Type

Workflows

Stats

30

Reviews

Published

Jun 27, 2026

Base Model

Anima

Hash

AutoV2
58ABFB0E01
default creator card background decoration
KU

KuroTao

License:

Anima

The Anima Model is licensed by CircleStone Labs LLC. Copyright CircleStone Labs LLC. IN NO EVENT SHALL CIRCLESTONE LABS LLC BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

Built on NVIDIA Cosmos

KuroTao Director

Ever opened ComfyUI and stared at an empty prompt box with no idea what to create?

That's exactly why I built KuroTao Director.

KuroTao Director is a local AI prompt generator for ComfyUI powered by Ollama and Qwen. Instead of randomly combining keywords, it creates complete image concepts with a clear scene, mood and visual storytelling.

The goal isn't to generate longer prompts.

The goal is to generate better ideas.


Features

✔ Runs completely locally using Ollama

✔ No API keys or subscriptions

✔ Generates creative scene-based prompts instead of simple tag lists

✔ Adjustable creative direction through easy dropdown menus

✔ Optimized for Anima's natural language prompting.

✔ Designed for rapid idea generation and creative exploration


How it works

Choose only a few creative directions such as:

• Genre

• Setting

• Characters

• Mood

• Visual Hook

Everything else can stay on Random.

KuroTao Director lets Qwen fill in the missing pieces and create a unique image concept every generation.

Sometimes you'll get something beautiful.

Sometimes something funny.

Sometimes something completely unexpected.

That's part of the fun.


Included

• KuroTao Director custom ComfyUI node

• Ready-to-use workflow

• Ollama Modelfile

• Installation guide


Extra Instructions

The Extra Instructions field allows you to give direct instructions to Qwen.

Unlike traditional prompt fields, you don't need to write Stable Diffusion tags or keywords.

Simply write your instructions in natural language.

Examples:

• The main character should have white hair and red eyes.

• Use Rem from Re:Zero as the main character.

• Use Pikachu as the main character.

• Always include a cute black cat somewhere in the scene.

• Create a peaceful autumn atmosphere.

• Avoid modern technology.

• The character should always wear a witch hat.

Qwen will use these instructions together with your selected dropdown options to create a unique prompt.

Tip: Simple keywords work well for individual attributes, while complete sentences are useful for more specific creative instructions.

__________________________________________________________________________________________________

Recommended Workflow

Leave most dropdowns on Random.

Only choose one to three options you really care about.

Examples:

Genre → Horror

Setting → Library

or

Genre → Cozy

Setting → Café

Generate multiple prompts and keep your favorites.


Hardware Requirements

KuroTao Director runs a local language model while ComfyUI generates images.

Recommended:

• 16 GB RAM or more

• Enough available VRAM to run both your image model and your selected Qwen model at the same time.

The included Modelfile is configured for:

Qwen 2.5 14B

If your hardware is powerful enough, you can also use other Qwen models.

Simply install your preferred model with Ollama, change the first line of the Modelfile, and recreate the model.

Example:

FROM qwen3:14b

or

FROM qwen3:32b

Larger models may produce more detailed prompts but require significantly more RAM and VRAM.


Safety

KuroTao Director relies on Qwen running locally through Ollama.

Qwen has its own built-in safety guidelines and will refuse certain requests.

This behavior comes from the language model itself and is expected.


Credits

This is my very first public ComfyUI project.

I'm not a professional Python developer.

KuroTao Director was created through a lot of experimentation and with the help of ChatGPT for coding, brainstorming and documentation.

If you find a bug, have suggestions, or something doesn't work correctly, please leave a comment.

I may not know the answer immediately, but I'll gladly investigate it and continue improving the project.

Have fun creating! ❤️