reve / editImage
Edit the closed model that won't ever ship weights
- images
- api_config
- images
- errors
- workflow_id
- raw_json
The edit half of the Reve story: same closed model, but now it takes an image and an instruction and returns the edited image. Since Reve has no open weights - it's one of the hosted frontier generators people were already name-dropping back when they said "OAI, Reve and Ideogram" in the same sentence - this node is the only realistic way to run Reve edits inside a Comfy graph. It goes through fal's engine on Civitai's orchestration layer, billed in Buzz.
This is worth reaching for when you want a closed-model edit that local Qwen-Edit or Kontext chains can't quite match, and you don't want to leave ComfyUI to do it. It's also the most hands-off editor in this pack: no seed, no strength slider, no sampler. You give it an image and a sentence, it gives you an image.
How it works
Standard civitai-comfy-nodes mechanics: upload your source image and prompt to the Civitai Orchestration API as a reve / editImage job on the fal engine, long-poll until done, and convert the result back to a native IMAGE tensor. Because the model is hosted, the node only exposes what fal's API exposes - which is deliberately sparse.
The inputs that matter
- prompt and images - both required. The instruction goes in the prompt, the source goes in
images. - aspect_ratio - the same wide menu as the create node (
auto,4:1through1:4, default1:1). - quantity - 1 to 10 edited outputs per job.
- output_format / image_metadata - format choice and optional metadata.
That's everything. No strength, no seed, no negative_prompt - the API doesn't take them, so don't go looking for a knob to control how hard the edit lands. Adjust via the prompt itself.
Outputs
images → Preview/Save Image. errors, workflow_id, raw_json are diagnostics; raw_json holds the full job result for debugging and cost checks.
Installing and authenticating
Pack-wide install - Reve nodes ship in civitai-comfy-nodes:
cd ComfyUI/custom_nodes
git clone https://github.com/civitai/civitai-comfy-nodes.git
pip install -r civitai-comfy-nodes/requirements.txt
or ComfyUI Manager → "Civitai Comfy Nodes" → Install → restart. Dependencies: requests + python-socketio[client].
Auth: Civitai Auth node → CIVITAI_API_TOKEN → stored key/OAuth in ~/.civitai/ → browser login. Headless:
export CIVITAI_API_TOKEN=your_token
Where people get burned
- No strength control. On a local editor you'd reach for a denoise/strength slider to keep edits subtle. Here you can't - the only lever is prompt wording ("slightly", "subtly"). Plan prompts accordingly.
- Your source image leaves the machine, along with the instruction, and it goes through hosted moderation. With no weights, there's no community bypass for anything the filter refuses.
- Per-image billing through fal. Each edited output is a metered job;
quantitymultiplies the cost. - No reproducibility. No seed means a lucky edit can't be recreated exactly - if you need a result again, save the
IMAGEimmediately. - Don't wire
imagesinto the createImage variant by mistake - that node ignores it and just generates from text.
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | — | |
| images | IMAGE | Address these from the prompt as <frame>N</frame>, 0-based. | |
| output_formatopt | COMBO | 4 options: , jpeg, png, webP | |
| image_metadataopt | STRING | External metadata that will be stored with the image | |
| aspect_ratioopt | COMBO | auto | 18 options: auto, 4:1, 3:1, 21:9, 2:1, 17:9, +12 |
| quantityopt | INT | 11–10 | — |
| 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 | — |