ComfyUI Node

Recaption

Caption an image by sending it to an OpenAI-compatible vision chat endpoint. The system prompt is selected from YAML files in <user>/default/coras_textgen/prompts/recaption/. The model dropdown is filtered to vision-capable ids by a name heuristic; disable the filter in Settings -> Cora's Textgen -> Recaption to show every model.

By cora-clanker·Created 3 months ago·Updated 3 months ago· 0
Recaption
  • image
  • text
style
model
Categorytextgen

Inputs (3)

NameTypeDefaultDescription
imageIMAGE
styleCOMBOPrompt style. Edit or add YAML files under <user>/default/coras_textgen/prompts/recaption/.
modelCOMBOVision-capable models advertised by the configured endpoint.

Outputs (1)

NameTypeDescription
textSTRING