Capability
Upsampling And Downsampling With Spatial Temporal Awareness
4 artifacts provide this capability.
Want a personalized recommendation?
Find the best match →Top Matches
1
make-a-video-pytorchFramework41/100
via “upsampling and downsampling with spatial-temporal awareness”
Implementation of Make-A-Video, new SOTA text to video generator from Meta AI, in Pytorch
Unique: Implements sampling operations that explicitly preserve temporal dimensions (frame count) while modifying spatial resolution, rather than treating video as 3D volume where all dimensions are sampled uniformly
vs others: More efficient than naive 3D sampling (which would reduce frame count) while maintaining temporal information, enabling practical multi-scale video processing