Which garment image type typically produces the best try-on results?

There’s no universal winner—each has trade-offs. Flat-lay shots are clean and consistent, aiding segmentation but offering fewer 3D shape cues. Mannequin shots add structure without pose conflicts. On-model shots provide natural drape and folds but can introduce pose or shadow mismatches with the target person. The workflow’s ImageCompare output helps you decide which is most reliable for your catalog.

How does the workflow preserve the person’s face and pose?

The KlingTryOnNode is designed to keep the subject’s identity and pose intact while substituting the clothing region. For best results, use a clear person photo with a visible face and minimal occlusions (e.g., avoid heavy scarves covering the jawline) so the preserved areas remain stable.

How can I ensure a fair comparison between garment types?

Keep the person image constant, use similar framing and scale for each garment input, and run at the same resolution. Avoid changing lighting or background between tests. Review the same criteria in ImageCompare—neckline alignment, sleeve fit, hem placement, and pattern fidelity—so differences reflect garment input style rather than setup variations.

What are common failure modes, and how can I reduce them?

Artifacts often arise with extreme poses, crossed arms covering the torso, heavy accessories, or partial garment visibility. Use front or slight 3/4 person shots, select garment images with full visibility (especially neckline and sleeves), and prefer uncluttered backgrounds. If on-model garment shots cause pose conflicts, try the mannequin or flat-lay version of the same item.

Kling: Virtual Try On - ComfyUI Workflow

This workflow compares how different garment image types—flat-lay, mannequin, and on-model—perform in a virtual try-on using the Kling model. Two LoadImage nodes bring in a person photo and a garment image; the KlingTryOnNode then generates a realistic try-on while preserving the subject’s face and pose. The ImageCompare node assembles a side-by-side view of the original person photo and the generated try-on so you can visually judge fit alignment, drape plausibility, color/pattern fidelity, and identity preservation. SaveImageAdvanced stores both the try-on and the comparison output, and a MarkdownNote provides in-graph guidance.

To make this a fair comparison, keep the same person image while swapping the garment source across flat-lay, mannequin, and on-model product shots. Normalize garment framing and scale before each run so the KlingTryOnNode sees similar inputs. Expect trade-offs: flat-lay images are clean and easy to segment but offer fewer 3D cues; mannequin shots add structure without pose conflict; on-model shots carry natural drape cues but may introduce pose/shadow mismatches. The side-by-side ImageCompare output helps you quantify which input style yields the fewest artifacts for your catalog or campaign.

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.

9038d2d953b6.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