How to use Deep3dFaceRecon model?

[Feywell]

Thanks for you great work to reimplement hififace!
I find that you use the Deep3dFaceRecon model straightly in size (256, 256). But the offical code of Deep3dFaceRecon is (224, 224).
Did you train the model use size (256, 256)?

[usingcolor]

From my experience, using 256x256 should be fine. Also, the Deep3dFaceRecon repo mentioned that they applied a number of augmentations during training. So I thought Deep3dFaceRecon should be robust enough to different image sizes.