Function smartResizeVideo

smartResizeVideo(
    numFrames: number,
    height: number,
    width: number,
    temporalFactor?: number,
    factor?: number,
    minPixels?: number,
    maxPixels?: number,
): [number, number]
HF smart_resize, VIDEO variant: the token budget counts the (padded) temporal extent — t_bar·h_bar·w_bar against min/max pixels.
Parameters
- numFrames: number
- height: number
- width: number
- temporalFactor: number = QWEN3VL_TEMPORAL_PATCH_SIZE
- factor: number = ...
- minPixels: number = QWEN3VL_VIDEO_MIN_PIXELS
- maxPixels: number = QWEN3VL_VIDEO_MAX_PIXELS
Returns [number, number]
- Defined in packages/inference/src/input/vision/qwen3vl-preprocess.ts:298