We train the network with a self-supervised super-resolution task. As the image representation is continuous, we can visualize and zoom in the image in an arbitrary resolution.
arxiv:
(3/n)
New work with Yinbo Chen, one of my first PhD students: Learning Continuous Image Representation
with Local Implicit Image Function. Check our video showing images in arbitrary resolutions.
proj:
code:
@YinboChen
@SifeiL
(1/n)
Inspired by recent progress in implicit function in the 3D community, we propose to encode 2D images using the implicit function as a continuous representation. The approach takes a local image feature and the coordinate as inputs, and predicts the RGB value as the output.
(2/n)
@xiaolonw
I wonder what would happen if you queried outside the original image frame? I know it's not trained to do outpainting, but sometimes continuous representations generate cool patterns/textures as you go out to infinity. See for example this expanded image I created with SIREN.