Open lvZic opened 1 month ago
There's no improvements on single image fitting. It is just fast for multiview or monocular video fitting.
There's no improvements on single image fitting. It is just fast for multiview or monocular video fitting.
so what's your opinion about model fitting by multi-view images, i mean the fitting by single image has low similarity with origin image, especially for asian people.
Compared to a single image, multi-view images provide a lot of additional stereo matching information, so in theory the fitting results will be much more accurate.
thans for you work. I wonder the improvement compared with 3d-reconstruction on single image (3DMM-pytorch-fitting) ? And if there is any material or paper about the effect difference between multiview and single image ?