TextGenerateLTX2Prompt - ComfyUI Built-in Node Documentation

The TextGenerateLTX2Prompt node is a specialized version of a text generation node. It takes a user’s text prompt and automatically formats it with specific system instructions before sending it to a language model for enhancement or completion. The node can operate in two modes: text-only or with an image reference, using different system prompts for each case.

Inputs

Parameter	Description	Data Type	Required	Range
`clip`	The CLIP model used for text encoding.	CLIP	Yes
`prompt`	The raw text input from the user that will be enhanced or completed.	STRING	Yes
`max_length`	The maximum number of tokens the language model is allowed to generate.	INT	Yes
`sampling_mode`	The sampling strategy used to select the next token during text generation.	COMBO	Yes	`"greedy"` `"top_k"` `"top_p"` `"temperature"`
`image`	An optional input image. When provided, the node uses a different system prompt that includes a placeholder for image context.	IMAGE	No
`thinking`	When enabled, the model will output its reasoning process before the final answer.	BOOLEAN	No
`use_default_template`	When enabled, the node will use the default chat template for formatting.	BOOLEAN	No
`video`	An optional video input that can be used as additional context for generation.	VIDEO	No
`audio`	An optional audio input that can be used as additional context for generation.	AUDIO	No

Note: The behavior of the node changes based on the presence of the image input. If an image is provided, the generated prompt will be formatted for an image-to-video task using a system prompt that describes how to expand the prompt based on the image’s content. If no image is provided, the formatting will be for a text-to-video task using a system prompt that expands the prompt into a detailed video generation description.

Outputs

Output Name	Description	Data Type
`output`	The enhanced or completed text string generated by the language model.	STRING

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): c037e526efab4b4e8eeeb84b4374d33a295996632b9897259806502ca32f48e1

​Inputs

​Outputs

Inputs

Outputs