Open ray0809 opened 2 years ago
Hi! All LaMa models in the paper were trained using 256x256 crops from Places. Original resolution of images in Places is approximately 512
Feel free to reopen the issue if you have further questions
Hi, @windj007. When using Places for training, why doesn't lama scale the image to 256 before cropping, is it more meaningful than directly making 256x256 crops ?
Due to the nature of convolutions, the networks adapt to the scale of objects/textures - and perform best on exactly that scale. Inpainting in 256 is not very interesting in practice - so why optimize methods for such low resolution? Original resolution of images in Places is 512 - so we decided to keep that scale (i.e. average size of objects in pixels), but training directly in 512 is very expensive - so we used crops.
Hi, Firstly, Thank you for making such a great project open source. I found the out_size in released big-lama config.yaml is 256, was the big-lama model trained with images' size 256?