Heatmaps

[affromero]

Hello, nice work.
I have a couple of doubts regarding the heatmaps.

1. Could you please elaborate on these values? Why resizing and shifting heatmaps and why those numbers for different regions of the face (x and x2)? In the main paper, there is nothing about heatmaps or keypoints, so I am trying to understand the intuition.

2. Are wing.ckpt pre-trained weights the same as in this work, or do they differ in some way?

3. Does CelebA_HQ work without heatmaps?

Thanks :)

[youngjung]

Hi, thanks for your interest.

1. Those heatmaps indicate the regions where we want to keep the source image stronger.
   x is for a coarse layer and x2 is for finer layer. (similar definition of coarse and fine in StyleGAN)

2. Yes, the same.

3. Yes it still works. Just eyes, nose and mouth of the source image will be less-reflected.

Do these answer your question?

[affromero]

Yes, pretty much.
The last remaining thing is about the resizing and shifting of the heatmaps.

[youngjung]

Ah, sorry I missed that part.
Those are our customization to fit region to eye-nose-mouth-chin part.
The region looks like below.
source
x * source
x2 * source

Empty part around mouth is not to keep mustache of the source.