After experimenting a lot. I found that current SSR in Raymarched mode is more convincing than Raytraced or mixed. RTR seems not to take into account the post processing nor other RT effects such as RTGI. Is there something I’m doing wrong ?
Hey, I navigate with very small information but, to me, it’s most likely a setup issue indeed.
RTR is known to give different results with high intensity objects because the intensity of the lighting in the ray traced reflection is clamped to facilitate denoising.
The default “Clamp Value” is rather low, I suggest increasing this value to recover some of that intensity.
Also, depending on your setup, you may have lights missing in your RTR lighting because the light cluster (see light cluster override) have a default distance of 10m.
Finally, SSR and RTR are not supposed to give exact same results, they are two very different approach to fix the same problem and they is pros and cons in both.
Having a second look on the first gif in particular, it almost looks like the raytraced reflection is not working at, in particular because the reflection on the ground “has a vertical offset”. Look more like a reflection probe.
Have you switched the editor to DX12 ?
1 Like
Yeah I’m already in DX12 and my light cluster distance is set at 50m. I have a better result when I bump the clamp value. Nevertheless, it still seems like the post process are not taking into account in the RT reflections.
Thanks for the capture, there is another thing you could do to improve here.
Basically, for the “last reflection” by default RTR will fallback on reflection probe and then the sky.
As you can see on the right there is blue on the upper side of the beam, which means you don’t have any reflection probe and the RTR is most likely sampling the ambient probe (sky) here.
What you can do is to either make a reflection probe for it to have a proper fallback or remove the sky from the reflection hierarchy like this.
