What’s the difference between the Image-based and VLM-based methods?

Image-based feeds the style image directly into OpenAIGPTImageNodeV2, typically giving more faithful style transfer. VLM-based uses AILab_QwenVL_Advanced to describe the style first and then prompts the image node, which is more flexible for variations but can be less strictly faithful to the reference.

How is the stop-motion effect created?

VHS_SelectEveryNthImage samples only every Nth frame from the video, and ImageBatchRepeatInterleaving holds each selected frame for multiple steps. Together, these nodes reduce motion fluidity and create the characteristic choppy cadence of stop motion.

Why do my outputs have mismatched sizing or artifacts?

Ensure the style/image processing path matches the video resolution using GetImageSize and ResizeImageMaskNode. If your style image has transparency, enable ImageRemoveAlpha+ to prevent edge artifacts. Preview intermediate frames to catch issues before rendering.

Does the final video include audio?

No. CreateVideo/SaveVideo in this workflow output a silent video. If you need audio, add it later in a video editor or with a separate audio muxing step.

How can I speed up or reduce costs when using remote model nodes?

Increase the N in VHS_SelectEveryNthImage to process fewer frames, and raise the repeat count to maintain duration. Trim with Video Slice and lower the output resolution with ResizeImageMaskNode.

Motion Collages - ComfyUI Workflow

Motion Collages turns any source video into a stylized stop‑motion sequence guided by a single reference image. The workflow samples frames with VHS_SelectEveryNthImage, holds each selected frame using ImageBatchRepeatInterleaving, and then reassembles the sequence with CreateVideo and SaveVideo. Two alternative style-transfer paths are provided as grouped node blocks you can toggle: an Image‑based path that feeds your style image directly into OpenAIGPTImageNodeV2, and a VLM‑based path that first has AILab_QwenVL_Advanced describe your style image and then uses that text to prompt OpenAIGPTImageNodeV2.

Under the hood, LoadVideo and GetVideoComponents extract frames, resolution, and FPS, with Video Slice available for trimming. ImageListToImageBatch and RebatchImages handle batching so frames flow efficiently through the style stage. ResizeImageMaskNode and GetImageSize keep your style and video dimensions in sync, while ImageRemoveAlpha+ cleans transparency if your style reference has an alpha channel. StringConcatenate assembles clean prompts or filenames, and PreviewImage/PreviewAny help you inspect intermediate outputs. The result is a controllable stop‑motion look: select every Nth frame for the cadence you want, repeat frames to accentuate the hold, and choose either faithful image‑guided transfer (Image‑based) or flexible, text‑described transfer (VLM‑based).

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.

c27ee1ed3d54.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