How do I control which part of the image moves?

Use the mask stage. Preview the animated region with MaskPreview and refine the mask before it’s converted by MaskToImage. Keep edges soft for cleaner blends and ensure the moving region doesn’t intersect high-contrast edges unless that’s intentional.

My loop has visible pops at the seam—how can I fix it?

Reduce motion intensity in ByteDance2ReferenceNode, shorten the duration slightly, and soften the mask edges. Also check fps/duration in GetVideoComponents/CreateVideo so the start and end frames align more naturally. Subtle motion yields smoother loops.

What image types and resolutions work best?

High-quality, well-lit portraits or product shots with a relatively clean background produce the best cinemagraphs. Square and vertical formats are social-friendly; set your desired output size/fps in GetVideoComponents/CreateVideo to match your target platform.

Do I need to change the custom utility nodes with ID-like names?

No. The nodes f4cfd806-6db2-425e-9f50-3eb8fe546a77 and e42b05e2-13fc-4ad5-822d-b89a8a297d35 are preconfigured for this workflow. Leave them as-is unless you know their specific role in your local build.

Seedance 2.0 - Viral Cinemagraph Effect

Seedance 2.0 - Viral Cinemagraph Effect turns a single character image into a short, looping cinemagraph with subtle motion. The workflow starts by loading your still using LoadImage (or BatchImagesNode for multiples), then uses a lightweight text pipeline (PrimitiveStringMultiline → RegexReplace → GeminiNode/GeminiNanoBanana2) to interpret your motion description into concise, model-friendly motion hints. A mask-driven pass defines where movement should happen: MaskPreview helps you visualize the region, and MaskToImage converts that mask for downstream use so the subject or selected area animates while the rest of the frame stays perfectly still.

At the core, ByteDance2ReferenceNode applies Seedance 2.0’s reference-based image-to-video motion on just the masked region, preserving identity and composition from your source photo. Frames are assembled with GetVideoComponents and CreateVideo, then you can inspect results with PreviewAny and export via SaveVideo (and SaveImage for any keyframe/mask snapshots). Two custom utility nodes (f4cfd806-6db2-425e-9f50-3eb8fe546a77 and e42b05e2-13fc-4ad5-822d-b89a8a297d35) are included in the graph to support the pre/post steps the workflow needs; they’re prewired and require no manual changes for a one‑click run. The result is a clean, seamless loop ideal for socials, hero headers, and character reveals—precise, controlled motion without rebuilding the entire scene.

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.

ed1c4eade3a3.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

Часто Задаваемые Вопросы