How do I know which generated frame will cut most seamlessly?

Use the Preview Images Difference node. It compares each generated frame to the last input frame; the darkest preview indicates the smallest color difference. Set that index in ImageFromBatch+ so the continuation begins at that frame.

How can I make the continuation feel more consistent with my source footage?

Write your prompt to closely describe the immediate next moment (lighting, motion direction, subject pose, background). Keep style cues consistent with the source. Generate more frames than you need, then trim at the best-matching index using the difference preview.

What about audio—will it continue automatically?

GetVideoComponents extracts the original audio, and AudioConcat can stitch it with the extended segment. Since generated frames don’t include new audio, you’ll typically preserve the original track up to the cut or add new audio in post.

How long can I extend the video?

It depends on your hardware and the settings in ByteDance2ReferenceNode (e.g., number of frames to generate). Quality and scene coherence generally decrease the farther you extend, so it’s best to generate a modest overrun and pick the optimal trim point.

Seedance 2.0 - Extend Video - ComfyUI Workflow

Seedance 2.0 - Extend Video is a ComfyUI workflow that continues an input clip by generating follow-on frames and trimming to the most seamless handoff point. It uses LoadVideo and GetVideoComponents to ingest your source, extract frames, FPS, and audio, and then feeds a follow-up text prompt into the Seedance 2.0 generator (ByteDance2ReferenceNode). The prompt describes what happens immediately after the last frame of your input, and the model predicts a continuation as a batch of new frames. PrimitiveStringMultiline and StringConcatenate are used to assemble the final conditioning text you send into the generator.

Because model-generated continuations won’t perfectly match your last real frame, the workflow includes an automatic cut-point aid. The Preview Images Difference node (f1f7c973-e50b-4ec7-84c5-0ef2eef1fc84) compares each generated frame to the last input frame and visualizes color difference; the darkest preview indicates the closest visual match. You select that index and trim the generated batch starting from that frame with ImageFromBatch+, then concatenate original frames with the trimmed continuation via BatchImagesNode. CreateVideo rebuilds a new clip at the source FPS, AudioConcat preserves or stitches audio as needed, and SaveVideo writes the final extended video. PreviewImage nodes are included to quickly inspect the last source frame and candidate match frames before committing to render.

API

Use this workflow from code

Every Comfy workflow is a JSON graph. The payload below is this workflow, exactly as ComfyUI runs it — fetch it from the URL, keep it in version control, or load it in ComfyUI and run it node by node.

fe773d7d7377.json

Fetching workflow JSON…

Run it from TypeScript or Python with the Comfy SDK. The same code targets Comfy Cloud or a ComfyUI you host yourself — only the base URL changes.

// Install (beta)
npm i @comfyorg/sdk

// Run this workflow (TypeScript)
import { Comfy } from "@comfyorg/sdk";

const client = new Comfy({ apiKey: "comfyui-..." });
const wf = await client.workflows.fromFile("workflow_api.json");
const job = await client.run(wf);
await job.getOutputs("<output-node-id>")[0].toFile("output.png");

The SDK takes a workflow in API format: open this workflow in ComfyUI and use File → Export Workflow (API).

Comfy Cloud API access requires a plan with an API key.

SDK docs Get an API key

FAQ

Frequently Asked Questions