top | item 16905042

(no title)

potlee | 7 years ago

Can someone who understands this tell us what is the catch here?

discuss

order

johndough|7 years ago

You'll probably hit memory limits if you go much beyond 512x512-sized holes.

Additionally, computation times grow quickly with higher resolution and you already need a high end GPU for this resolution to get a reasonably interactive response time.

You'll also need a favorably licensed pretrained model or a few 10000 training images and masks.

So all in all, I can't see any deal breakers, but I'd probably still use PatchMatch instead.

rhcom2|7 years ago

For reference the GPU they're using for this paper is the NVIDIA V100 GPU, a datacenter GPU costing $8,000.

hanrelan|7 years ago

Since they programmatically generate the masks you wouldn't need those, just the set of training images. So it wouldn't be too hard to find since you're not looking for paired images, just a bunch of images of faces/landscapes/whatever you're trying to inpaint.

taeric|7 years ago

The catch is it is more like an artists interpretation. Anyone expecting that it will truly fix old damaged pictures to be like they were is going to be disappointed. Anyone just wanting to fill in some gaps will be excited. (Same for just procedural generation of some images, I suppose. Would be neat to see what it can do from a very low fidelity outline of a house/forest/etc.)

taneq|7 years ago

As opposed to it being truly magical, I guess?