Accurate Human Rendering in Game [2018]

Status
Not open for further replies.
My irritation is still lack of proper breathing [that affects everything from nose down to belly]. Detailed [or any] belly/chest/neck breathing animations are still quite rare since they are not mocapped [mocapping is tracking bulky suit].

Wait the Last of us 2
 
detroit_becomehuman_26cun5.jpg

detroit_becomehuman_2gau77.jpg
 
Good assets + good shaders + good post process pipeline + excellent, consistent lighting.
I was very annoyed by the low resolution DOF commonly used in the game. One, who is responsible for this shader integration in recent Sony games must be fired.
Why has somebody decided that it would be a great idea to use quarter or even lower res DOF almost everywhere in the game? Because of this, the game looks like a 1080p rather than 4K even with checkerboarding, I even checked options several times to be sure that it really works in 4K.
The same low res DOF was very distracting in recent GOW and here it's again, thanks Sony:mad:
 
Last edited:
Blurs are invariably less than native res because you can't perceive the difference with a bilinear upscale from a blurred source. Here's a blurred source, a quarter res+blurred+upscaled version, and the deviation between the two.
Image1.jpg
I'm guessing you're talking about fringing on silhouettes. The overhead of native res for post effects is too high in most cases so lower res needs to be used, and a different masking solution is needed to avoid edge artefacts.

Edit : Another example, downscaled to 1/16th (quarter res both dimensions), blurred at a quarter of the kernel size, and upscaled to original res. Saves 93.75% of the blur processing (bandwidth and shader workload) versus processing at native res. For quality you could improve the upscale process, but the artefacting is extreme in this test case and not noticeable in real scenes.

Image1.png
 
Last edited:
The problem with half res DOF is often parts of the scene the were supposed to be burred over a small, maybe pixel-sezed radius, also get downsampled, and then the low res becomes very noticeable. Ideally they'd only downscale to 1/3 res the parts that are gonna be burred enough for it not to be noticeable.
 
this is fucking NPC gameplay animation

IMO they have been the leaders in animation blending for a while now. Its always been a wonder to me as to why other studios can't seem to match let alone exceed them in this area other than that it must be very very difficult. But what they showed for TLoU2 is something else.
 
What is the graphical difference between ingame and cinematic in the last of us 2? Some told me that detail decreases in actual gameplay. The line seems blurred for some whats actual gameplay and whats not.
 
What is the graphical difference between ingame and cinematic in the last of us 2? Some told me that detail decreases in actual gameplay. The line seems blurred for some whats actual gameplay and whats not.

No idea for TLOU2, I don't see the detail decrease maybe less muscle and skin deformation.

giphy.gif
 
Status
Not open for further replies.
Back
Top