Nodes/VRGameDevGirl Video Enhancement Nodes/VRGDG ChatGPT Images Browser
ComfyUI Node

VRGDG ChatGPT Images Browser

ChatGPT image generation inside ComfyUI — no API key, just a real browser

By vrgamegirl19·Created about a year ago·Updated 2 days ago· 717
VRGDG ChatGPT Images Browser
  • image1
  • image2
  • image3
  • image4
  • image5
  • image6
  • image7
  • image8
  • image9
  • image10
  • image11
  • image12
  • image13
  • image14
  • image15
  • image16
  • image17
  • image18
  • image19
  • image20
  • image21
  • image22
  • image23
  • image24
  • image25
  • image26
  • image27
  • image28
  • image29
  • image30
  • image31
  • image32
  • image33
  • image34
  • image35
  • image36
  • image37
  • image38
  • image39
  • image40
  • image41
  • image42
  • image43
  • image44
  • image45
  • image46
  • image47
  • image48
  • image49
  • image50
  • image
promptcreate a cinematic image
image_count0
debug_port9223
timeout_seconds600
reuse_open_projecttrue

Let's get the obvious thing out of the way first: this node does not call an OpenAI API and you don't need a key. VRGDG ChatGPT Images Browser is browser automation. It opens a real Chrome window, signs into chatgpt.com, pastes your prompt into ChatGPT's image tool, and downloads the result back into ComfyUI as a normal IMAGE tensor. It's part of VRGameDevGirl's bigger video-production pack (the one whose flagship is the LTX 2.3 AI Video Builder), and it exists so you can feed OpenAI-quality generated or edited images into a ComfyUI graph without ever leaving the canvas.

Why bother? Because for certain jobs - strong prompt adherence, editing a reference into something specific, a quick style pass on a character shot - ChatGPT's image model is genuinely good, and none of that is available locally. This node hands the image straight to your workflow, which is exactly what you want when you're assembling scene images for a video project and don't want to tab out, generate, download, and drag the file back in.

How it works

The mechanism is a small Node.js/Playwright script (same machinery as the Flow browser nodes) talking to a Chrome instance launched with remote debugging enabled. It uses a separate port, 9223, so it never collides with the Flow node's 9222. It also uses its own Chrome profile, chrome-chatgpt-profile, which is why you sign into ChatGPT once and stay signed in.

  • prompt - the text pasted into ChatGPT Images. Multiline.
  • image_count - how many of the image1image50 sockets this run reads, starting at image1. 0 means text-only generation. If you connect images, raise this to match.
  • debug_port - the remote-debugging port. Leave 9223 unless something else took it.
  • timeout_seconds - how long ComfyUI waits before giving up. Default 600; raise it for slow generations or big uploads.
  • reuse_open_project - on by default, so automation reuses the current ChatGPT tab instead of navigating back to the images start page for every image.

The optional image1image50 inputs attach reference images through ChatGPT's paperclip before the prompt is submitted. The single output is the downloaded image as an IMAGE, which plugs straight into whatever comes next.

Installing it

This is a two-step setup, and the second step catches people out. First install the pack itself - Manager, search vrgamedev, or:

cd ComfyUI/custom_nodes
git clone https://github.com/vrgamegirl19/comfyui-vrgamedevgirl.git

Restart ComfyUI and hard refresh the browser. The pack's requirements.txt is heavy (transformers, opencv, and the notorious voxcpm/llama-cpp-python builds); on the Windows portable build, run the pip tooling commands from the README first. Python 3.12 is the safer environment.

Then the part people forget: run the VRGDG Flow Browser Setup node once (or npm install in the flow_automation folder manually). That node installs Node.js and the Playwright dependency this automation needs. Chrome must also be installed - automation drives a real browser, it can't conjure one.

Common issues

  • It does nothing / errors on launch. Almost always the setup node hasn't been run, so Playwright isn't installed. Run it once and let it finish.
  • "Not signed in." The automation drives its own chrome-chatgpt-profile. You have to sign into chatgpt.com in that profile - the launcher .bat files in flow_automation/ open it for you (Windows).
  • Socket inputs ignored. image_count controls how many image sockets exist. If you connect an image but leave it at 0, it's not read.
  • Slow generations timing out. Bump timeout_seconds; 600 is plenty for a normal gen, not for a multi-image edit on a bad day.

One honest caveat: this is a nice-to-have utility, not a daily driver for most people. It's fragile in the way all browser automation is - ChatGPT changes its UI and the script has to catch up. Use it for the occasional image you want in your graph, and don't build your whole pipeline on it.

CategoryVRGDG/ChatGPT

Inputs (55)

NameTypeDefaultDescription
promptSTRINGcreate a cinematic imagePrompt to paste into ChatGPT Images. Connected images are attached first, then this prompt is submitted.
image_countINT00–50How many image sockets this run should attach, starting at image1. Use 0 for text-only generation.
debug_portINT92231–65535Local Chrome remote-debugging port for ChatGPT Images automation. Default 9223 keeps it separate from the Flow node's 9222 port.
timeout_secondsINT60060–2400Maximum time to let ChatGPT create and download the image. Increase for slow generations.
image1optIMAGEOptional image attachment #1. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image2optIMAGEOptional image attachment #2. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image3optIMAGEOptional image attachment #3. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image4optIMAGEOptional image attachment #4. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image5optIMAGEOptional image attachment #5. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image6optIMAGEOptional image attachment #6. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image7optIMAGEOptional image attachment #7. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image8optIMAGEOptional image attachment #8. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image9optIMAGEOptional image attachment #9. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image10optIMAGEOptional image attachment #10. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image11optIMAGEOptional image attachment #11. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image12optIMAGEOptional image attachment #12. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image13optIMAGEOptional image attachment #13. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image14optIMAGEOptional image attachment #14. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image15optIMAGEOptional image attachment #15. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image16optIMAGEOptional image attachment #16. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image17optIMAGEOptional image attachment #17. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image18optIMAGEOptional image attachment #18. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image19optIMAGEOptional image attachment #19. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image20optIMAGEOptional image attachment #20. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image21optIMAGEOptional image attachment #21. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image22optIMAGEOptional image attachment #22. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image23optIMAGEOptional image attachment #23. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image24optIMAGEOptional image attachment #24. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image25optIMAGEOptional image attachment #25. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image26optIMAGEOptional image attachment #26. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image27optIMAGEOptional image attachment #27. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image28optIMAGEOptional image attachment #28. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image29optIMAGEOptional image attachment #29. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image30optIMAGEOptional image attachment #30. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image31optIMAGEOptional image attachment #31. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image32optIMAGEOptional image attachment #32. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image33optIMAGEOptional image attachment #33. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image34optIMAGEOptional image attachment #34. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image35optIMAGEOptional image attachment #35. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image36optIMAGEOptional image attachment #36. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image37optIMAGEOptional image attachment #37. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image38optIMAGEOptional image attachment #38. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image39optIMAGEOptional image attachment #39. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image40optIMAGEOptional image attachment #40. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image41optIMAGEOptional image attachment #41. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image42optIMAGEOptional image attachment #42. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image43optIMAGEOptional image attachment #43. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image44optIMAGEOptional image attachment #44. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image45optIMAGEOptional image attachment #45. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image46optIMAGEOptional image attachment #46. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image47optIMAGEOptional image attachment #47. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image48optIMAGEOptional image attachment #48. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image49optIMAGEOptional image attachment #49. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
image50optIMAGEOptional image attachment #50. Set image_count high enough to include this socket. Images are attached to ChatGPT through the paperclip before the prompt is submitted.
reuse_open_projectoptBOOLEANtrueWhen enabled, non-manual automation reuses the current ChatGPT Images tab/conversation instead of navigating back to the images start page for every image.

Outputs (1)

NameTypeDescription
imageIMAGE