Nodes/ComfyUI-SwissArmyKnife/LLM Studio Picture Describe
ComfyUI Node

LLM Studio Picture Describe

Captions a single IMAGE tensor by relaying it to an LM Studio vision-language model, returning the generated paragraph for downstream prompt or logging nodes.

By sammykumar·Created 11 months ago·Updated 7 months ago· 5
LLM Studio Picture Describe
  • image
  • caption
base_urlhttp://192.168.50.41:1234
model_nameqwen/qwen3-vl-30b
api_endpointcompletions
caption_promptPlease analyze the image and provide a detailed description of the the person's hair and facial features. Focus especially on texture, style, length, flow, facial structure, eyes, brows, nose, lips, skin tone, any visible markings (freckles, moles, etc.), and how light or shading affects their face. Use precise, descriptive language. Return description as one paragraph
temperature0.2
verbosefalse
CategorySwiss Army Knife 🔪/Media Caption

Inputs (7)

NameTypeDefaultDescription
base_urlSTRINGhttp://192.168.50.41:1234LM Studio server URL (e.g. http://192.168.50.41:1234)
model_nameSTRINGqwen/qwen3-vl-30bModel name in LM Studio (e.g. qwen/qwen3-vl-30b)
api_endpointCOMBOcompletionsAPI endpoint: 'chat/completions' (may include thinking tags) or 'completions' (cleaner output)
imageIMAGEInput image to analyze
caption_promptSTRINGPlease analyze the image and provide a detailed description of the the person's hair and facial features. Focus especially on texture, style, length, flow, facial structure, eyes, brows, nose, lips, skin tone, any visible markings (freckles, moles, etc.), and how light or shading affects their face. Use precise, descriptive language. Return description as one paragraphPrompt for image caption
temperatureFLOAT0.20–2Temperature for text generation
verboseBOOLEANfalseShow detailed processing information in console

Outputs (1)

NameTypeDescription
captionSTRING