🎥 Watch the workflow demo on YouTube:

🚀 ComfyUI Direct Integration — BFL Flux3 Video V2V Workflow

Continue and extend existing videos directly inside ComfyUI using the official Black Forest Labs FLUX 3 Video API.

This workflow uses my BFL Flux3 Video custom node to provide direct API integration between ComfyUI and Black Forest Labs, allowing you to submit an existing video as the starting point for a new FLUX 3 generation without leaving your ComfyUI workflow.

🎬 What This Workflow Does

This is a V2V — Video-to-Video / Video Continuation workflow.

Load or generate a source video, describe how you want the scene to continue, configure your FLUX 3 settings, and queue the workflow.

The BFL Flux3 Video node handles:

• 🎥 Video-to-video / video continuation

• ▶️ Existing video input via start_video

• 🌐 Direct communication with the official Black Forest Labs API

• 🔄 Automatic generation-status polling

• 🎞️ Automatic MP4 download

• 🔊 Optional synchronized audio generation

• 📐 Configurable aspect ratio

• ⏱️ Configurable output duration

• 🖥️ HD / FHD generation options

• 🛡️ Configurable safety tolerance

• ⚡ Draft generation

• ✨ draft_enhance support

• 🎥 Native ComfyUI VIDEO output

The resulting video can then be passed directly into other compatible ComfyUI video nodes.

▶️ How the V2V Workflow Works

Connect an existing ComfyUI VIDEO to:

start_video

Then set:

mode = v2v

Use the prompt to describe what should happen next in the video.

This makes V2V useful for:

• extending an existing shot

• continuing camera movement

• continuing character action

• carrying a scene into a new moment

• building longer sequences from shorter generated clips

• chaining multiple FLUX 3 generations together

• experimenting with iterative video continuation

The BFL Flux3 Video node automatically prepares the connected video for submission to the Black Forest Labs API.

✨ Prompt Enhancement

This workflow also uses my LM Studio Prompt Enhancer custom node, allowing you to refine or expand your continuation prompt using a locally running LLM through LM Studio before submitting it to FLUX 3.

You can use your own continuation prompt directly, or use the prompt-enhancement stage to create a richer cinematic description of how the source video should evolve.

📦 Required Custom Nodes

You will need:

🎬 BFL Flux 3 Video

By Puppet-Vision

Provides the direct ComfyUI integration with the Black Forest Labs FLUX 3 Video API.

✨ LM Studio Prompt Enhancer

By Puppet-Vision

Provides optional local LLM-powered prompt enhancement through LM Studio.

🔑 Black Forest Labs API Key Required

[!IMPORTANT]
This workflow uses the official hosted Black Forest Labs API.

FLUX 3 Video is not running locally in this workflow.

You need:

• ✅ A Black Forest Labs account

• ✅ A valid BFL_API_KEY

• ✅ Sufficient Black Forest Labs API credits

Create an API key and purchase credits here:

https://dashboard.bfl.ai/

The BFL Flux3 Video node can use either:

BFL_API_KEY

from your environment, or an API key entered directly into the node.

Using the environment variable is recommended when sharing workflows.

💳 API Credits & Generation Costs

Every FLUX 3 generation is processed by Black Forest Labs and consumes BFL API credits according to their current pricing.

[!WARNING]

💸 draft_enhance can consume significantly more credits

Draft generation can be useful for previewing a result before committing to a higher-quality render, but enhancing a draft requires additional API credits.

Check your available BFL credit balance before using draft_enhance.

Billing, API pricing, credit balances, rate limits, and API availability are controlled by Black Forest Labs.

🧠 Basic Workflow

Source VIDEO
     │
     ▼
Continuation Prompt
     │
     ▼
LM Studio Prompt Enhancer
     │
     ▼
Enhanced Prompt
     │
     ▼
BFL Flux3 Video
mode = V2V
     │
     ▼
Black Forest Labs FLUX 3 API
     │
     ▼
Continued Video
     │
     ▼
Native ComfyUI VIDEO

⚙️ Typical Settings

Example:

mode             = v2v
duration         = 5
resolution       = hd
generate_audio   = true
draft            = false

Connect your source video:

VIDEO ─────► start_video

Then use the prompt to describe the continuation you want FLUX 3 to generate.

For example:

The camera continues moving forward through the forest while the fog thickens and warm sunlight begins breaking through the trees.

Adjust the settings according to the length, quality, and type of continuation you want to generate.

🔁 Build Longer Video Sequences

One of the most useful applications of this workflow is chaining generations.

Original VIDEO
      │
      ▼
BFL Flux3 Video V2V
      │
      ▼
Continuation 1
      │
      ▼
BFL Flux3 Video V2V
      │
      ▼
Continuation 2
      │
      ▼
Longer Sequence

This allows you to experiment with progressively extending a scene while remaining entirely inside your ComfyUI graph.

Each continuation is a separate Black Forest Labs API generation and may consume additional credits.

🔌 Direct ComfyUI Integration

The main purpose of this workflow is to make the official BFL Flux3 Video API feel like a native part of ComfyUI.

Instead of manually encoding a video, creating API requests, polling jobs, downloading results, and importing videos back into ComfyUI, the custom node handles the API workflow automatically.

ComfyUI VIDEO
      ↓
BFL Flux3 Video Node
      ↓
Official Black Forest Labs API
      ↓
FLUX 3 Video Continuation
      ↓
ComfyUI VIDEO Output

This makes it much easier to incorporate FLUX 3 video continuation into larger automated ComfyUI pipelines.

⚠️ Disclaimer

This workflow and the associated custom nodes are independent community projects.

I am not affiliated with, endorsed by, sponsored by, or associated with Black Forest Labs or Comfy Org / the ComfyUI organization.

Black Forest Labs, FLUX, FLUX 3, ComfyUI, and related trademarks belong to their respective owners.

API access, billing, pricing, credits, availability, and usage policies are controlled by Black Forest Labs.

🔗 Links & Contact

🔗 ComfyUI BFL Flux 3 Video – GitHub
https://github.com/AllenCraigBarnard/ComfyUI-BFL-Flux-3-video

🔗 ComfyUI BFL Flux 3 Video Workflows
https://github.com/AllenCraigBarnard/ComfyUI-BFL-Flux-3-video/tree/main/workflows

🔗 PuppetVision
https://puppetvision.nl