the input_size and out_size of big-lama about lama HOT 4 OPEN

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.

from lama.

Hi! All LaMa models in the paper were trained using 256x256 crops from Places. Original resolution of images in Places is approximately 512

from lama.

Feel free to reopen the issue if you have further questions

from lama.

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 ?

from lama.