Skip to main content
Unlisted page
This page is unlisted. Search engines will not index it, and only users having a direct link can access it.

ComfyUI

The ComfyUI integration gives power users access to advanced, node-based AI video and image generation workflows directly within ACT3 AI. ComfyUI is designed for technical users who need fine-grained control over generative parameters and want to chain multiple AI models in a single render pipeline.

Connecting to a Local ComfyUI Instance​

ACT3 AI can connect to a ComfyUI server running on your local machine. This lets you run custom node workflows for image generation, character design, or set concept art without sending data to the cloud.

  1. Install and run ComfyUI locally (see the ComfyUI GitHub project for setup instructions)
  2. In ACT3 AI, go to Account → Integrations → ComfyUI
  3. Enter your ComfyUI endpoint URL (typically http://127.0.0.1:8188)
  4. Click Test Connection to verify the link
  5. Once connected, you can trigger workflows directly from the Shot or Actor editor in ACT3 AI

The local connection allows you to use your own GPU and models without consuming ACT3 AI cloud credits for those generations.

What ComfyUI Does​

ComfyUI provides a visual node editor where you build generative workflows by connecting processing nodes. Each node performs a specific operation — loading a model, applying a prompt, adjusting resolution, adding effects, or routing output. You chain nodes together to create complex multi-step generation pipelines that no single model can produce on its own.

Key Capabilities​

  • Node-Based Workflow — Build and edit generative pipelines visually using drag-and-drop
  • Multi-Model Integration — Chain outputs from Stable Diffusion XL, Flux, Google Veo 3.1, and other supported models
  • Custom Parameters — Set detailed control over resolution, frame rate, motion settings, lighting, and style weights
  • Template System — Save reusable workflow templates for different production types
  • Batch Processing — Render multiple shots or variations in parallel

How to Use​

  1. In the Editor, select a shot or scene and click AI → ComfyUI
  2. The ComfyUI node editor opens
  3. Add nodes from the node library: prompt input, model selection, resolution, style filters, output
  4. Connect nodes in sequence to define the generation pipeline
  5. Set parameters for each node
  6. Click Generate to produce output and review in the Preview panel
  7. Save the node setup as a workflow template for reuse

Building a Basic Pipeline​

A simple text-to-video ComfyUI pipeline looks like:

[Text Prompt] → [Model: Stable Diffusion XL] → [Upscale] → [Style Filter] → [Output: Video Clip]

A more advanced pipeline might be:

[Text Prompt] → [Model: SDXL] → [Image Output]
                                        ↓
[Motion Prompt] → [Model: Flux] → [Animate Image] → [Color Grade] → [Output: Video Clip]

Template Workflows​

ACT3 AI includes built-in ComfyUI templates for common production needs:

  • Concept Art from Script — SDXL image generation from shot descriptions
  • Animate Still Image — Flux-based motion applied to a static image
  • Style Transfer — Apply a visual style from one image to a generated scene
  • Multi-Pass Render — High-quality render using multiple models in sequence
  • Fast Concept Preview — Low-cost quick generation for creative ideation

Start with a template and modify nodes as needed.

Credit Usage​

ComfyUI renders are billed based on the models and resolution used:

  • Multi-model workflows consume credits for each model in the chain
  • Example: SDXL → Flux pipeline at 1080p uses approximately 5 credits per 5 seconds
  • Complex workflows with many nodes cost proportionally more

Monitor credit usage in the Credits Panel before committing to large batch runs.

Best Use Cases​

  • Advanced VFX shots requiring multiple AI passes
  • Complex motion sequences needing layered rendering
  • Experimental or highly stylized sequences
  • Automated generation of multiple visual variations for client review
  • Technical users who want maximum control over every generation parameter

Who Should Use ComfyUI​

ComfyUI is recommended for users with experience in AI generation workflows who need more control than a single prompt allows. For most production needs, Google Veo 3.1, Runway, or Flux provide excellent results with simpler setup.

Tips and Best Practices​

  • Start with a prebuilt template to speed up workflow creation
  • Use Set Design outputs as a base layer and enhance with ComfyUI pipelines
  • Keep track of credit usage when chaining multiple models in one render
  • Save custom workflows as named templates for future use
  • Test with low-resolution outputs before committing to final quality

Troubleshooting​

Node connections not working — Check that output types match between connected nodes (e.g., image output connects to image input, not text input).

Pipeline fails mid-run — Check each node's parameters individually. A single invalid setting stops the whole pipeline.

Credit usage higher than expected — Review the number of models in your pipeline. Each model charges separately.