Error while running on ICON results

[nghorbani]

Hi,
Thanks for sharing your great work! I tried running scanimate on the icon results the way by dumping ICON meshes as ply and creating pkl files in an exact format as the scanimate example. Yet when training a new scnamiate first two steps finish successfully, i.e. pretraining skinning nets, and trainig skinning nets, but outputing samples of canonicalization gets into this error:
Traceback (most recent call last): File "/home/nghorbani/miniconda3/envs/scanimate/lib/python3.9/runpy.py", line 197, in _run_module_as_main return _run_code(code, main_globals, None, File "/home/nghorbani/miniconda3/envs/scanimate/lib/python3.9/runpy.py", line 87, in _run_code exec(code, run_globals) File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 749, in <module> trainWrapper() File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 746, in trainWrapper train(opt) File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 727, in train train_igr(opt, ckpt_dir, result_dir, igr_net, fwd_skin_net, inv_skin_net, File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 474, in train_igr smpl_cano_n = compute_normal_v(smpl_cano.permute(0,2,1), smpl_face.expand(smpl_posed.shape[0],-1,-1)).permute(0,2,1) File "/mnt/d/code-repos/SCANimate/lib/geo_util.py", line 143, in compute_normal_v Ns = Ns.scatter_add_(1, Fs.long().reshape(Fs.size(0),-1,3).cpu(), nf.reshape(Fs.size(0),-1,3).cpu()).type_as(Vs) RuntimeError: index 4592573275984363520 is out of bounds for dimension 1 with size 6890
do you have a clue how to go past this error?

[Bill-WangJiLong]

Hello, how do you write the SMPL parameters of ICON into pkl format or npz format?

Hi, Thanks for sharing your great work! I tried running scanimate on the icon results the way by dumping ICON meshes as ply and creating pkl files in an exact format as the scanimate example. Yet when training a new scnamiate first two steps finish successfully, i.e. pretraining skinning nets, and trainig skinning nets, but outputing samples of canonicalization gets into this error: Traceback (most recent call last): File "/home/nghorbani/miniconda3/envs/scanimate/lib/python3.9/runpy.py", line 197, in _run_module_as_main return _run_code(code, main_globals, None, File "/home/nghorbani/miniconda3/envs/scanimate/lib/python3.9/runpy.py", line 87, in _run_code exec(code, run_globals) File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 749, in <module> trainWrapper() File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 746, in trainWrapper train(opt) File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 727, in train train_igr(opt, ckpt_dir, result_dir, igr_net, fwd_skin_net, inv_skin_net, File "/mnt/d/code-repos/SCANimate/apps/train_scanimate.py", line 474, in train_igr smpl_cano_n = compute_normal_v(smpl_cano.permute(0,2,1), smpl_face.expand(smpl_posed.shape[0],-1,-1)).permute(0,2,1) File "/mnt/d/code-repos/SCANimate/lib/geo_util.py", line 143, in compute_normal_v Ns = Ns.scatter_add_(1, Fs.long().reshape(Fs.size(0),-1,3).cpu(), nf.reshape(Fs.size(0),-1,3).cpu()).type_as(Vs) RuntimeError: index 4592573275984363520 is out of bounds for dimension 1 with size 6890 do you have a clue how to go past this error?