With the current system, converting UVs back and forth causes precision issues, e.g. 2047→converted to float→converted back to short→2046.
The image below shows the original source NJA file (left) compared to the NJA output by SAMDL (right)
This PR fixes the issue by making UVs use double instead of float to improve precision.
With the current system, converting UVs back and forth causes precision issues, e.g. 2047→converted to float→converted back to short→2046. The image below shows the original source NJA file (left) compared to the NJA output by SAMDL (right)
This PR fixes the issue by making UVs use double instead of float to improve precision.