Jump to content


This topic is now archived and is closed to further replies.

Reflection bug (aka snow instead wet)

Recommended Posts

I think everyone saw this 'phenomenon' in DR2, presenting wet surface to be looked like covered by snow.

This one I spotted on recent video posted by @PJTierney. Here is a screenshot from this video modified in order to explain the issue. What I've done (besides desaturation) is I stretched pixels found inside a puddle upward to compare their brightness with a sky.

(click for original size)


As you can see, reflection if water is way brighter than image being reflected (sky). At the same time puddle is brighter than a background (in case some want to argue about transparency).
This makes water puddles brighter than they actually should be. It looks unrealistic if image being reflected doesn't justify such brightness.

Similar issue we can see on car liveries. I already saw on the forum several complaints that livery reflections are too strong. And here is a proof. The image has been taken from the same video.


There is no doubt that the result is too bright. There might be more reasons contributing to this issue. Let's devs find and solve it.
But... where is PBR? PBR doesn't allow to create more energy (brighter reflection) from something emitting less of it (original image). And tbh gfx artists should spot this inconsistency without even knowing physics.

Share this post

Link to post
Share on other sites

Your methodology is flawed. There's 2 things at play in these scenarios, 1 is the direct reflection of materials, and another is the Fresnel effect. The Fresnel effect will often give this highly reflective/whiteout appearance. It's not made any better when it's a real time game, so there's lots of little cheats and hacks going on instead of a full path or ray traced render.

The PBR is clearly noticeable, so... I'm questioning if you really know what PBR is and how it works.

Share this post

Link to post
Share on other sites

there are no two separate reflections in real world (aka direct and Fresnel). there is a light bouncing from objects. Fresnel equation says how much light is refracted and how much is left to be reflected in function of angle. 

Yes, for majority of materials reflection is most intensive at very narrow angles (it's what you are referring to). But the result of reflection cannot be brighter than image being reflected - never. there is no adding light in a process of reflection. 

In other thread you are using "too reflective" and "over-reflective" words. I understand your intentions , but for me "too reflective" must still fit rules of physics. if reflection is brighter than a source - it's an obvious bug. 

Main goal of PBR is to maintain proper amount of light (and its proportion across all objects). At the end light is the only one thing we see. In this case it obviously doesn't work.

Share this post

Link to post
Share on other sites