4 ms·
It's a pity that you cannot upload your own image(s) as starting points. My theory is that because this is built be using pre-trained BigGAN models, they do no
by fxtentacle 7y ago
It's a pity that you cannot upload your own image(s) as starting points.
My theory is that because this is built be using pre-trained BigGAN models, they do not have access to the encoder to convert an image to the latent space, but instead only to the generator that converts the latent space to an image.
Accordingly, they can only remix parameters in latent space, but not locate the latent space that a real-world photo would reside at.
- gspr 7y ago> It's a pity that you cannot upload your own image(s) as starting points. What do you mean? Click for example "create" -> "portrait" -> "upload".
- fxtentacle 7y agoApparently it works for portraits. I tried the general case, which doesn't offer an upload.
- fxtentacle 7y agoIt sure is quite slow at ~1 hour per uploaded portrait. That looks more like an exhaustive parameter search to me, than a trained encoder network.