For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Python module

max.pipelines.lib.generate_images

Image generation from text prompts and input images, in-process.

Each prompt becomes the same request max serve builds for POST /v1/responses, so the model’s defaults and the request’s validation are the same as in serve. The images come back as PNG, so they match serve’s images for output_format="png". Serve defaults to JPEG. An input image goes to the tokenizer directly, the same place serve’s data URI image ends up. Prompts run one at a time, since an image pipeline takes one request per execute.

Classes

GenerateImagesResult	The result of generate_images().
ImagePipeline	The pipeline surface generate_images() drives.
ImageTokenizer	The tokenizer surface generate_images() drives.

Functions

generate_images	Generates images for each prompt, one prompt per pipeline run.