openai / dall-e-3 / createImage
The last-gen classic with style and quality knobs
- api_config
- images
- errors
- workflow_id
- raw_json
DALL-E 3 occupies a funny spot in 2026: superseded by GPT Image, sure, but still the model a generation of prompters learned on, still genuinely good at following detailed natural-language prompts, and - unusually for a modern image model - it ships with a style control. natural vs vivid is a real, visible aesthetic lever, and it's one of the few places in this entire Civitai pack where you get to flip between "photographic" and "illustrated" without touching the prompt. That alone makes CivitaiImageGenOpenaiDallE3CreateImage worth a look even though OpenAI's newer models exist.
What it is. Engine openai, model dall-e-3, operation createImage, one of the ~160 nodes in the official civitai/civitai-comfy-nodes pack. Required inputs: prompt and size - and the size list (1024x1024, 1792x1024, 1024x1792) is where DALL-E 3's landscape/portrait options live. Optional: style (natural/vivid, default vivid), quality (auto/hd/standard, default auto), background (auto/transparent/opaque), output_format, and image_metadata (external metadata stored with the image).
Inputs that matter. style is the star: vivid (default) gives you the saturated, stylized DALL-E look; natural dials it toward photorealism. quality → hd buys detail at more Buzz. And background → transparent gets you an alpha channel, the same trick the DALL-E 2 node offers but at a much higher quality ceiling. size picks your canvas - go 1792x1024 for wide compositions, 1024x1792 for vertical ones.
How it works. Same machinery as everything in this pack: prompt goes into a workflow payload, POSTed to orchestration.civitai.com, long-polled until the cloud job finishes, and the result comes back as an IMAGE on the images output alongside errors, workflow_id, and raw_json (the full response, handy for cost checks). No local model, no VRAM, Buzz-metered.
Install. The whole pack installs at 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, then find it under Civitai/Image/openai.
Auth and gotchas. Credentials resolve from a wired Civitai Auth node, then CIVITAI_API_TOKEN, then stored OAuth/key. Error hints: 402 = insufficient Buzz, 401 = token expired, 429 = rate limited; 30-minute job cap, early-preview pack. DALL-E 3's reputation for aggressive filtering is earned - its refusal rates are part of why people moved on - and Civitai's own moderation sits on top. And as always with cloud nodes, your prompt leaves the machine.
The take. DALL-E 3 is the "if you want the classic look" node. The natural/vivid switch and the transparent-background option are genuinely useful, and its prompt comprehension is still better than several newer hosted options. If you're picking one OpenAI create node for a serious project, GPT Image 1.5 is the better default - but when you specifically want DALL-E's house style, this is the node, and it's a pleasant reminder that "old" and "useless" aren't the same thing.
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | — | |
| size | COMBO | 3 options: 1024x1024, 1792x1024, 1024x1792 | |
| output_formatopt | COMBO | 4 options: , jpeg, png, webP | |
| image_metadataopt | STRING | External metadata that will be stored with the image | |
| styleopt | COMBO | vivid | 2 options: natural, vivid |
| qualityopt | COMBO | auto | 3 options: auto, hd, standard |
| backgroundopt | COMBO | auto | 3 options: auto, transparent, opaque |
| api_configopt | CIVITAI_CONFIG | Optional Civitai Auth connection; defaults to CIVITAI_API_TOKEN or stored OAuth login. |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| images | IMAGE | — |
| errors | STRING | — |
| workflow_id | STRING | — |
| raw_json | STRING | — |