Fixed large uv wrapping #249
Conversation
|
This is the code that wraps uv´s back into valid bounds |
|
Arthur tested this and confirmed it works as intended btw. Should be ready for one more review and then being merged |
Arthur suggested this a solution.
Lilaa3
force-pushed
the
fixed-large-uv-wrapping
branch
from
72e460c to
117ac42
Compare
|
All known issues should be fixed now |
fast64_internal/f3d/f3d_writer.py
Outdated
| return (SShift * sMirrorScale, TShift * tMirrorScale) | ||
|
|
||
|
|
||
| def getUVInterval(f3dMat) -> tuple[int, int]: |
There was a problem hiding this comment.
I don't think this behaves as expected if e.g. tex0 has shift=0 mirror=True and tex1 has shift=1 mirror=False
There was a problem hiding this comment.
When shift is 0, SShift will be 1, because 2^0 is 1. What´s wrong here?
There was a problem hiding this comment.
The result of your case should be 2.
| 2 if f3dMat.tex0.T.mirror or f3dMat.tex1.T.mirror else 1, | ||
| ] | ||
| # To prevent wrong UVs when wrapping UVs into valid bounds, | ||
| # we need to account for the highest texture shift and if mirroring is active. |
There was a problem hiding this comment.
To reiterate on my previous comments why "highest"? There must be a piece of code elsewhere on exporting that handles this similarly? Where
There was a problem hiding this comment.
There isn´t because this is doing something else, we don´t need to account for texture resolutions and what not, so there is no function which does this behavior specifically.
|
I have some comments on style but the actual code makes sense and if it is tested then I don't feel the need for any functionality changes |
|
|
|
Oops closed on accident, what I was going to say is that fazana had incorrect mask values, the pr should be working correctly. |
getSTUVRepeats couldn´t auto type hint, getUVInverval can. Use 1.0 instead of 1
Could you test then please? This shouldn´t affect oot differently at all. |
Dragorn421
added
bug
Arthur suggested this a solution, he said he could test it later today. Until then this will remain a draft pr.