Nodes/zhangp365/ComfyUI-utils-nodes/volcano ark image edit
ComfyUI Node

volcano ark image edit

Image editing via ByteDance's Volcano Ark API

By zhangp365·Created 2 years ago·Updated 5 months ago· 135
volcano ark image edit
  • image
  • image2
  • image3
  • edited_image
prompt
level2k
aspect_ratio9:16
max_images1
timeout120
api_key
api_url
model

This pack ships three separate Volcano-branded nodes, and it's worth being precise about which one this is: VolcanoArkImageEditNode is the multi-image version, specifically built to take several input images at once and edit them together at a chosen resolution and aspect ratio through ByteDance's Volcano Ark API - distinct from the pack's simpler VolcanoImageEdit and VolcanoOutpainting nodes.

What it's for

If you're already using Volcano Engine (ByteDance's cloud platform) for other generation work, or you specifically want an image-editing backend that natively takes up to three reference images in one call rather than one at a time, this is the node for that. It's a cloud API wrapper, not a local model - the actual editing happens on Volcano's servers.

How it works

You supply a primary image, an editing prompt, and up to two additional reference images, along with a target output resolution tier and aspect ratio. The node sends the request to the Ark API and returns a single edited result.

The inputs and outputs that matter

  • image (IMAGE, required) - your primary input image.
  • prompt - describes the edit you want, multiline.
  • level (1k / 2k / 4k, default 2k) - output resolution tier. Higher tiers cost more in both API billing and request time.
  • aspect_ratio (8 choices including 1:1, 16:9, 9:16, 21:9, default 9:16) - the output shape.
  • max_images (default 1, range 1–10) - how many result images to generate per request.
  • timeout (default 120, range 10–600) - how long the node waits before giving up on the request.
  • image2 / image3 (optional) - additional reference images beyond the primary one, for edits that need to combine multiple sources.
  • api_key (optional) - your Volcano Engine key.
  • api_url (optional, blank by default) - lets you point the request at a specific endpoint instead of whatever the node defaults to internally. Worth knowing this exists if the default endpoint doesn't reach you reliably, since Volcano Engine is a mainland-China cloud platform and network routing to it isn't uniformly fast worldwide.
  • model (optional, blank by default) - override which Ark model handles the request, if you have access to more than the default.

Output: a single edited_image (IMAGE).

How to install it

Via ComfyUI Manager: search "ComfyUI-utils-nodes," install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/zhangp365/ComfyUI-utils-nodes

Restart ComfyUI. No local model download - you need a Volcano Engine account and an API key. The pack's README walks through getting one for the related Volcano nodes: visit the Volcano Engine console, log in, go to the access-control / key-management settings, and create a key. That same key works for this node's api_key field.

Common issues & troubleshooting

The three-Volcano-node situation in this pack is genuinely easy to mix up - VolcanoOutpainting, VolcanoImageEdit, and this one (VolcanoArkImageEditNode) all hit related but distinct API surfaces. If you're getting errors that don't make sense for what you're trying to do, double-check you picked the multi-image Ark variant specifically rather than one of its siblings.

An empty or invalid api_key will fail every request, same as any billed API node. If requests are failing to connect at all rather than failing with an API-level error, that's when api_url is worth investigating - the default endpoint may not be the fastest or most reliable path depending on where your ComfyUI instance is actually running, and this field exists specifically to let you override it. And keep timeout in mind against level: a 4k request with multiple reference images and a higher max_images count is doing meaningfully more work than the 120-second default was likely tuned around - raise it before assuming a slow request has actually failed.

Categoryimage/ark

Inputs (11)

NameTypeDefaultDescription
imageIMAGE
promptSTRING
levelCOMBO2k3 options: 1k, 2k, 4k
aspect_ratioCOMBO9:168 options: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, +2
max_imagesINT11–10
timeoutINT12010–600
image2optIMAGE
image3optIMAGE
api_keyoptSTRING
api_urloptSTRING
modeloptSTRING

Outputs (1)

NameTypeDescription
edited_imageIMAGE