Download a subset of the data and place them in data
Download the checkpoints and place them in checkpoints
Place any images of class 'classname' in 'data/png/classname'. For example, interpolating two images of broccoli should be placed in 'data/png/broccoli'. Images should be 256 x 256 and grayscale.
Make sure to set the batch size equal to the number of images. For example, if interpolating between 5 images, run:
python3 test.py --batch-size 5 --img-class broccoli
When running interpolation, outputs will be saved to 'results/test'.
On lines 92-93 in 'test.py'
y = linear(encoded, ibf=20, ease=sigmoid)
y = catmullRom(encoded, ibf=20, ease=sigmoid)
You can configure the spatial interpolation (linear/catmulRom), the number of inbetween frames (ibf) and the easing amount (linear/sigmoid). For the linear call, you may also pass in a set of interpolation points p. For example,
y = linear(encoded, ibf=20, ease=sigmoid, p=[(0.2,-1), (0.8,2), (1,1)])
Will run linear spatial interpolation with sigmoid easing with anticipation and follow-through.