>>2870025Because it's a privatter list-only upload, so either it's only a part of the password or an entirely fake one.
Also, holy autism:
>> Facial close-ups with a FOV of 15 or more are prohibited.>>2870066What you show here looks natural enough to pass for detailed shadows, unlike picrel.
Anyway, it happens because of the principles outlined here:
https://sites.google.com/site/moggproject/%E3%82%BB%E3%83%AB%E3%83%95%E3%82%B7%E3%83%A3%E3%83%89%E3%82%A6 (oversimplified but good to get the gist)
https://learn.microsoft.com/en-us/windows/win32/dxtecharts/cascaded-shadow-mapsI assume you're using RayCast. What I had in picrel was mostly caused by using a large camera distance (>60) with a small FOV (5) for a close-up, which also relates to PSSM, or, in this case of a rectangle light, the absence of it. I thought it could be solved by increasing SHADOW_MAP_SIZE (which depends on quality, but the code stops at #elif SHADOW_MAP_QUALITY >= 2), similarly to
https://www.bilibili.com/read/cv13504451/, but, ironically, only reducing it alleviated the aliasing, because even the maximum setting of 4096 that my 6GB RTX 2060 could take only made it worse.
You can also reduce the artifact size by massively refining (high-polying & smoothing) the geometry, changing the light direction to hit your model at an angle closer to 90, reducing light hardness & the limits of shadowHardness (for additional lights), limiting camera distance, only using PSSM-based lighting at high quality settings, or maybe some other ways. Maybe you can get antialiasing to resolve it, but playing with its settings didn't seem to have any effect in my case.
>>2870087Take YYB Miku and Tda Append Miku, do a headswap, and presto!