Nodes/civitai-comfy-nodes/openai / dall-e-2 / createImage
ComfyUI Node

openai / dall-e-2 / createImage

The museum piece of the Civitai pack

By civitai·Created 2 months ago·Updated about a month ago· 42
openai / dall-e-2 / createImage
  • api_config
  • images
  • errors
  • workflow_id
  • raw_json
prompt
size
output_format
image_metadata
quantity1
backgroundauto

Let's be blunt: in 2026, DALL-E 2 is a museum piece. It launched in 2022, it's been superseded twice by DALL-E 3 and then by the GPT Image line, and its generation quality is visibly a couple of model generations behind everything else in this pack. It exists here because the civitai/civitai-comfy-nodes pack is generated from Civitai's Orchestration API spec - one node per engine - so the old OpenAI model gets a node the same way the new ones do. That's the whole reason to know this node's shape: it's the completeness artifact, and understanding it tells you how the pack handles any engine.

What it actually does. CivitaiImageGenOpenaiDallE2CreateImage (engine openai, model dall-e-2, createImage) sends your prompt to Civitai's cloud, which runs OpenAI's DALL-E 2, and returns an IMAGE to your graph. No local model, Buzz-metered. The required inputs are prompt and size - and the size list (256x256, 512x512, 1024x1024) is a time capsule; nobody's shipped 256-square generation as a flagship option since 2022. Optional inputs: quantity (1-10 - the highest ceiling in the image pack), background (auto/transparent/opaque), output_format, and image_metadata.

Where it's actually useful. Two niches, both narrow. One: backgroundtransparent. DALL-E 2 can emit an image with an alpha channel, and if you need a quick asset with transparency, this node is one of the few in the pack that offers it directly. Two: nostalgia / intentional style. DALL-E 2 has a softer, painterly default look that some people genuinely prefer for certain aesthetics - it's not "worse," it's "different," and it's cheap in Buzz terms because it's old. If neither niche applies to you, skip it: DALL-E 3 in this same pack is strictly better for generic generation, and GPT Image is better still.

Install and auth. Same as every node here - you install the pack once:

cd ComfyUI/custom_nodes
git clone https://github.com/civitai/civitai-comfy-nodes.git
pip install -r civitai-comfy-nodes/requirements.txt   # requests

Or ComfyUI Manager → Custom Nodes Manager → search Civitai Comfy Nodes → Install, restart, and look under Civitai/Image/openai. Credentials resolve from a wired Civitai Auth node, then CIVITAI_API_TOKEN, then a stored key or sidebar OAuth. 402 = out of Buzz, 401 = token expired, 429 = rate limited, 30-minute timeout, early-preview pack.

The take. Use this node for the transparency trick or the retro look; otherwise treat it as evidence that Civitai ships every engine in its catalog, not just the ones you'd choose. The pack's real DALL-E story starts with the DALL-E 3 node - but if you ever wonder why the menu has four OpenAI image entries, this is why. Generated, not hand-picked.

CategoryCivitai/Image/openai

Inputs (7)

NameTypeDefaultDescription
promptSTRING
sizeCOMBO3 options: 256x256, 512x512, 1024x1024
output_formatoptCOMBO4 options: , jpeg, png, webP
image_metadataoptSTRINGExternal metadata that will be stored with the image
quantityoptINT11–10
backgroundoptCOMBOauto3 options: auto, transparent, opaque
api_configoptCIVITAI_CONFIGOptional Civitai Auth connection; defaults to CIVITAI_API_TOKEN or stored OAuth login.

Outputs (4)

NameTypeDescription
imagesIMAGE
errorsSTRING
workflow_idSTRING
raw_jsonSTRING