Fix #108229: Wider tolerance shuffling NLA strips
In `transdata_get_time_shuffle_offset_side`, the tolerance for terminating the loop is too tight, sometimes it will lead to a infinite loop due to non-integral start/end values. Now fixed. Pull Request: https://projects.blender.org/blender/blender/pulls/108257
This commit is contained in:
@@ -119,7 +119,9 @@ static float transdata_get_time_shuffle_offset_side(ListBase *trans_datas, const
|
||||
}
|
||||
|
||||
total_offset += offset;
|
||||
} while (!IS_EQF(offset, 0.0f));
|
||||
} while (!IS_EQT(offset, 0.0f, 1e-4));
|
||||
/* Needs a eps greater than FLT_EPS because strip->start/end could be non-integral, and after
|
||||
* those calculations, `offset` could fall outside of FLT_EPS. */
|
||||
|
||||
return total_offset;
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user