ComfyUI Node
LM Studio Generate Text
A ComfyUI node in LM Studio with 11 inputs and 1 output.
LM Studio Generate Text
- connection
- image
- generated_text
◄promptDescribe this image.►
◄max_tokens500►
◄temperature0.70►
◄top_p0.95►
◄seed0►
◄system_prompt►
◄max_image_size1024►
◄image_formatjpeg►
◄batch_separator
►
CategoryLM Studio
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| connection | LM_STUDIO_CONNECTION | — | |
| prompt | STRING | Describe this image. | — |
| max_tokens | INT | 5001–8192 | — |
| temperature | FLOAT | 0.700–2 | — |
| top_p | FLOAT | 0.950–1 | — |
| seed | INT | 00–4294967295 | — |
| imageopt | IMAGE | — | |
| system_promptopt | STRING | — | |
| max_image_sizeopt | INT | 1024256–4096 | Images are downscaled so their longest side is at most this many pixels before sending. Most VLMs resize internally anyway, so larger values mostly cost latency, not quality. |
| image_formatopt | COMBO | jpeg | jpeg: ~10x faster encode, ~10x smaller payload. png: lossless, better for pixel-exact or text-heavy images. |
| batch_separatoropt | STRING | When `image` is a batch of more than one image, each image gets its own request and the resulting captions are joined with this string. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| generated_text | STRING | — |