Video workflow types for WAN, LTX 2.5/2.3, Seedance, HappyHorse, and MiniMax H3
models.
r2v (reference-to-video) is the multi-reference workflow shared by
HappyHorse (1-9 reference images fetched from referenceImageUrls) and
MiniMax H3 (minimax-h3-ref2va-fp8_r2v, uploaded images and/or videos, plus
optional audio). Because the two families do not take the same assets over
the same transport, resolve requirements with
getVideoAssetRequirements(modelId) instead of reading
VIDEO_WORKFLOW_ASSETS.r2v directly.
flf2v (first-and-last-frame-to-video) is the MiniMax H3 workflow that
interpolates between two required anchor images.
Video workflow types for WAN, LTX 2.5/2.3, Seedance, HappyHorse, and MiniMax H3 models.
r2v(reference-to-video) is the multi-reference workflow shared by HappyHorse (1-9 reference images fetched fromreferenceImageUrls) and MiniMax H3 (minimax-h3-ref2va-fp8_r2v, uploaded images and/or videos, plus optional audio). Because the two families do not take the same assets over the same transport, resolve requirements withgetVideoAssetRequirements(modelId)instead of readingVIDEO_WORKFLOW_ASSETS.r2vdirectly.flf2v(first-and-last-frame-to-video) is the MiniMax H3 workflow that interpolates between two required anchor images.