Skip to content

Conversation

bigcat88
Copy link
Contributor

Currently, the KlingImageGenerationNode node always passes image_type to the Kling API, which forces users to provide a reference image, even though the Kling API supports generation from prompts alone and the reference image is optional.

Without this PR, the following error can occur when trying to generate an image:

image

Tested with this PR that fix are fine and does not break anything:

Screenshot from 2025-08-10 08-44-39

Since the Kling-v1 model does not support image references, tests with it were skipped in some parts, as shown in the screenshot.

@bigcat88 bigcat88 force-pushed the fix/api-nodes/kling-image branch from 248e6c5 to 14ec234 Compare August 11, 2025 05:21
Copy link
Collaborator

@christian-byrne christian-byrne left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@comfyanonymous comfyanonymous merged commit f4231a8 into comfyanonymous:master Aug 11, 2025
6 checks passed
Thor-ATX pushed a commit to asteriafilmco/ComfyUI that referenced this pull request Aug 11, 2025
…ams-update

* commit 'f4231a80b1b904b45ade0def9b37320c4adfe71b':
  fix(Kling Image API Node): do not pass "image_type" when no image (comfyanonymous#9271)
  Support SimpleTuner lycoris lora for Qwen-Image (comfyanonymous#9280)
  Update template & embedded docs (comfyanonymous#9283)
  remove creation of non-used asyncio_loop (comfyanonymous#9284)
  Only show feature flags log when verbose. (comfyanonymous#9281)
  Bump pytorch cuda and rocm versions in readme instructions. (comfyanonymous#9273)
  Not sure if AMD actually support fp16 acc but it doesn't crash. (comfyanonymous#9258)
  Users report gfx1201 is buggy on flux with pytorch attention. (comfyanonymous#9244)
  async API nodes  (comfyanonymous#9129)
  _ui.py import torchaudio safety check (comfyanonymous#9234)
  Fix RepeatLatentBatch not working on multi dim latents. (comfyanonymous#9227)
  Update frontend to v1.24.3 (comfyanonymous#9175)
  Update template to 0.1.52 (comfyanonymous#9206)
  Fixed an issue where qwenLora could not be loaded properly. (comfyanonymous#9208)
@bigcat88 bigcat88 deleted the fix/api-nodes/kling-image branch August 12, 2025 05:22
@hasanodomseries
Copy link

I'm getting a similar image now with the Open ai image gen node. When using a reference image, I get the error that it cant see my reference image and its asking for a png/jpeg. Without a reference image, it works fine. Bug happens only on 3.50. I use the desktop version so i had to install a portable to revert back to 3.48/49 (i forget which one). No bug there. I've put in a support ticket, hope it gets resolved soon, all of our workflows are using GPT API nodes with reference images.

@bigcat88
Copy link
Contributor Author

bigcat88 commented Aug 16, 2025

Maybe this is fixed in #9348

Can you check on that PR or on master branch?

Vander-Bilt pushed a commit to Vander-Bilt/ComfyUI that referenced this pull request Aug 26, 2025
…mfyanonymous#9271)

* fix(Kling Image API Node): do not pass "image_type" when no image

* fix(Kling Image API Node): raise client-side error when kling_v1 is used with reference image
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants