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

Python function

hash_video

`hash_video()`

max.support.image.hash_video(pixel_values, grid_thw)

source

Compute the hash of preprocessed video pixels and grid metadata.

The input must already be the sampled, resized, normalized, model-ready video tensor. This helper does not decode or preprocess video frames.

Parameters:

pixel_values (ndarray[tuple[Any, ...], dtype[Any]]) – The preprocessed, model-ready video pixel tensor.
grid_thw (ndarray[tuple[Any, ...], dtype[integer[Any]]]) – The temporal/height/width grid metadata for the video.

Returns:

The signed 64-bit video hash.

Return type:

int

hash_video()​

`hash_video()`