Skip to main content
Kling Virtual Try-On is a partner node in ComfyUI that puts a piece of clothing onto a person in a photo. You connect two images, a photo of a person and a photo of the garment, and the node returns the same person wearing the item with the face and body preserved. It accepts product photos, flat lays, mannequin shots, and on-model photos. In ComfyUI the model runs through the Kling Virtual Try-On partner node. It is a cloud API node: it needs a Comfy account with credits and downloads no local model. You take two images in and get one image out, alongside a side-by-side comparison. See Partner Node pricing for the per-run rate. The result matches the size of the person image, capped at 2048 pixels on the longest side. Images larger than that are downscaled before the run.

​
Available workflows

​
Virtual Try-On

Dress a person in a garment and compare the result against the original person image. The template loads a person photo and a garment photo, runs the try-on at full 2048px fidelity, and shows an ImageCompare view of the input person and the output. Kling Virtual Try-On workflow preview

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “Kling: Virtual Try On” in Template Library
Input materials

model_colorful_outfit.png

Load into the person LoadImage node that feeds the Kling Virtual Try-On node

green_floral_dress.png

Load into the garment LoadImage node that feeds the Kling Virtual Try-On node

​
Workflow overview

The template uses five functional nodes, plus a MarkdownNote that lists the input assets:
  • LoadImage (person): provides the photo of the person
  • LoadImage (garment): provides the clothing to put on
  • KlingTryOnNode: the core node that combines the two images
  • SaveImageAdvanced and ImageCompare: save the result and show it next to the person image

​
Steps to run

  1. Load the person photo: upload or select a photo of one person in the person LoadImage node
  2. Load the garment: upload or select the clothing in the garment LoadImage node
  3. Set keep_pose: leave it on to hold the pose, or turn it off to let the pose adapt for a better presentation
  4. Click Queue or press Ctrl+Enter to generate

​
Node controls

​
Tips

  • Use a clean, front-facing person photo. The node works best when the subject faces the camera or is turned about three-quarters toward it.
  • Clothing only. Shoes, bags, and accessories are not supported; connect the garment itself.
  • An on-model garment photo carries more than the item. When the garment image is an on-model shot, the rest of that outfit can come along, which helps when you want a complete look.
  • The result keeps the person image size. The output is the same shape as the person photo, capped at 2048 pixels on the longest side.
  • Expect variation between runs. The seed does not lock the result; the same settings can still produce a different take.

​
Get started

  1. Update ComfyUI to the latest version
  2. Go to Template Library, search for Kling: Virtual Try On