Mu Hu

We have not especially considered the deployment on Apple devices. I think DepthPro from Apple would help.

> Hello! Thank you for your amazing work! I already read the paper and I might miss some information, I would appreciate your answers, thus I have several questions: >...

“相同image_size，相同焦距的情况下”，我认为你说的是实际物理量而非pixel单位。此时像素单位的焦距并不相同。 这种情况下，图像会先被resize到相同大小，这时焦距会成比例变换，resize后的图像像素焦距相同

I am not sure whether it will work well. The scaling factor upon different loss items should be properly adjusted.

I am not quite sure about the torch.hub parts. Could Zach @ZachL1 introduce some experience here?

> @YvanYin, so I flipped the height and width, > > in the **2m GT** image I got 2m at intrinsic = [500, 500, 1512, 2016]  > > in...

Thank you for your interest and there are two ways: (1) As a temporary solution, you could directly change (like a hard coding) the intrinsic [here](https://github.com/YvanYin/Metric3D/blob/f3833ba284767925c85098c1960d06f99ac3e466/mono/utils/custom_data.py#L33) and apply the in-the-wild...

> Hi! I have 640x360 images. I wonder if I can use the model directly on these lower resolution images without resizing to 616x1064. We cannot because the original ViT...

Have you checked the keys in the model.name_parameters() and the checkpoint? I remember there are some name issue when loading the DINO checkpoint. But for our weights, this should not...