Can I use a local PC to build a generation environment?

[QTHunYuan]

Can I use a local PC to build a generation environment?
For example, running the qwen3 model on a local PC to generate script related settings, storyboard scripts, and dialogue, and then running comfyui to execute image and video workflows within comfyui.

[xvrrr]

Can I use a local PC to build a generation environment? For example, running the qwen3 model on a local PC to generate script related settings, storyboard scripts, and dialogue, and then running comfyui to execute image and video workflows within comfyui.

Sure. That said, there's a bit of a mismatch with the current agentic workflow. To make it work, probably need to tailor the workflow a bit (e.g., modify the storyboard frames generation and reference logic).

[Sh1d0w]

I am also interested in this, would be nice to add out of the box support for comfyui for complete local generation.

[jmanhype]

Hey @Sh1d0w @QTHunYuan @xvrrr 👋

I've been working on exactly this! Built a Python library called vibe-aigc that does local generation via ComfyUI.

What it does:

• Uses local LLM (or cloud like z.ai/GLM) for script/planning
• Connects to ComfyUI for actual image/video generation
• Full async pipeline: Vibe → Plan → Generate → Real Content

Quick demo:

from vibe_aigc.comfyui import ComfyUIBackend

backend = ComfyUIBackend()  # connects to localhost:8188
result = await backend.generate_image(
    prompt="cyberpunk android, neon city",
    width=512, height=512, steps=25
)
print(result.images)  # actual generated image URLs

Repo: https://github.com/jmanhype/vibe-aigc
Install: pip install vibe-aigc

Also ported Comfy-Pilot to Windows if anyone needs that: ConstantineB6/comfy-pilot#2

Happy to collaborate if there's interest in integrating this with ViMax's approach!