search

pbaylies / stylegan-encoder

StyleGAN Encoder - converts real images to latent space
Other
740 stars 182 forks source link

Question about non human face data #27

Closed hbarovertwo closed 4 years ago

hbarovertwo commented 4 years ago

Hi @pbaylies

I know that the faces dataset model is the highest quality among the several models that exist, but I was interested in the cars model. It generates good images, and I was hoping to encode some existing images into the car model's latent space and interpolate between them. However the results, using the existing pre-trained cars model and training resnet a bit did not do well at embedding images. Any advice? Thanks! acura1

pbaylies commented 4 years ago

Hi @hbarovertwo

I haven't tried this personally with cars, but the ResNet can take a while to train well. As for encoding, your results may vary, especially if the images you are trying to encode aren't similar to the training data. Good luck!

cssddnnc9527 commented 2 years ago

Hi @pbaylies

I know that the faces dataset model is the highest quality among the several models that exist, but I was interested in the cars model. It generates good images, and I was hoping to encode some existing images into the car model's latent space and interpolate between them. However the results, using the existing pre-trained cars model and training resnet a bit did not do well at embedding images. Any advice? Thanks! acura1

Hi @hbarovertwo

Have you resolve your work to generate high quality car image? Currently I'm also trying to encoder a car image, not very clear as follows: baoma

hbarovertwo commented 2 years ago

@cssddnnc9527 I just used https://github.com/NVlabs/stylegan2 , they have pretrained car networks. Used those to encode car images and interpolate between vectors for some good looking synthetic cars.