Can I use a pinhole image for the inference input?

[byran-wang]

Hi @flamehaze1115 and @xxlong0,

Thanks for your great work.

From the inference code, the camera model for the condition image is orthogonal. However, the camera model of most images obtained from a telephone or the internet is pinhole.

How would the performance be affected if I use a pinhole image for the inference input? Can I use a pinhole image for the inference input?

Thanks for your kind reply.