ComfyUI Node
JoyCaption Beta One (External LLaMA)
A ComfyUI node in Mira/SubPack/JoyCaption with 8 inputs and 3 outputs.
JoyCaption Beta One (External LLaMA)
- image
- caption
- prompt
- extra_options_examples
◄caption_typeDescriptive►
◄caption_lengthlong►
◄extra_options►
◄temperature0.60►
◄top_p0.90►
◄max_tokens512►
◄llama_urlhttp://127.0.0.1:58080/chat/completions►
CategoryMira/SubPack/JoyCaption
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| caption_type | COMBO | Descriptive | 11 options: Descriptive, Descriptive (Casual), Straightforward, Stable Diffusion Prompt, MidJourney, Danbooru tag list, +5 |
| caption_length | COMBO | long | 31 options: any, very short, short, medium-length, long, very long, +25 |
| extra_options | STRING | If there is a person/character in the image you must refer to them as {name}. NOT include information about people/characters that cannot be changed (like ethnicity, gender, etc), but do still include changeable attributes (like hair style). Include information about lighting. Include information about camera angle. Include information about whether there is a watermark or not. Include information about whether there are JPEG artifacts or not. If it is a photo you MUST include information about what camera was likely used and details such as aperture, shutter speed, ISO, etc. Do NOT include anything sexual; keep it PG. Do NOT mention the image's resolution. You MUST include information about the subjective aesthetic quality of the image from low to very high. Include information on the image's composition style, such as leading lines, rule of thirds, or symmetry. Do NOT mention any text that is in the image. Specify the depth of field and whether the background is in focus or blurred. If applicable, mention the likely use of artificial or natural lighting sources. Do NOT use any ambiguous language. Include whether the image is sfw, suggestive, or nsfw. ONLY describe the most important elements of the image. If it is a work of art, do not include the artist's name or the title of the work. Identify the image orientation (portrait, landscape, or square) and aspect ratio if obvious. Use vulgar slang and profanity, such as (but not limited to) "fucking," "slut," "cock," etc. Do NOT use polite euphemisms—lean into blunt, casual phrasing. Include information about the ages of any people/characters when applicable. Mention whether the image depicts an extreme close-up, close-up, medium close-up, medium shot, cowboy shot, medium wide shot, wide shot, or extreme wide shot. Do not mention the mood/feeling/etc of the image. Explicitly specify the vantage height (eye-level, low-angle worm’s-eye, bird’s-eye, drone, rooftop, etc.). If there is a watermark, you must mention it. Your response will be used by a text-to-image model, so avoid useless meta phrases like “This image shows…”, "You are looking at...", etc. | |
| temperature | FLOAT | 0.600–2 | Higher = more random, lower = more deterministic |
| top_p | FLOAT | 0.900–1 | Nucleus sampling parameter |
| max_tokens | INT | 5121–2048 | Maximum number of tokens to generate |
| llama_url | STRING | http://127.0.0.1:58080/chat/completions | External LLaMA service URL |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| caption | STRING | Generated image caption |
| prompt | STRING | Prompt used for caption generation |
| extra_options_examples | STRING | Examples of extra options that can be used |