It's necessary for certain things, not necessary for a transformative improvement over the previous gen.I'm catapulted back to debate before this gen launched on whether RT hardware was necessary. Software GI isn't perfect, but it has come a long way since those early comparisons where it exhibited more atefacts. I feel could go either way at this point - would be nice if this gen had better RTHW as what we got don't do a lot, but it'd also have survived without any I think. Although HWRT results should improve similarly as devs find best use workarounds.
INteresting how it is going straight to recreate the Prey 2 vibe and even gameplay mechanic. Ledge hanging while shooting was shown off in prey 2.
It's necessary for certain things, not necessary for a transformative improvement over the previous gen.
HWRT is of course better, just not necessary to advance visuals far above the prior gen.Right. We were promised dynamic probe based GI last generation and it never materialized. Lumen took it one step further and actually delivered. It’s a huge leap from the baked GI or flat lighting of previous games.
It’s therefore easy to look at Lumen and say wow that’s great we don’t need HWRT yet. That’s a very short-sighted view IMO. Firstly Lumen has very clearly documented limitations. We shouldn’t pretend that it does everything well. It is not a general purpose GI solution. More importantly the availability of RT hardware is a catalyst for ray/path tracing innovation and optimization that wouldn’t be possible without it. We should be ecstatic that we’re getting to see both paths evolve simultaneously and offer a bit of variety and competition.
HWRT is important in the consoles if nothing else as training wheels for when it sees full use next-gen. But side-by-side, if you had one platform with HWRT and one without, the leap in visuals wouldn't be so constrained on the no-HWRT device to look dated or inferior.Right. We were promised dynamic probe based GI last generation and it never materialized. Lumen took it one step further and actually delivered. It’s a huge leap from the baked GI or flat lighting of previous games.
It’s therefore easy to look at Lumen and say wow that’s great we don’t need HWRT yet. That’s a very short-sighted view IMO. Firstly Lumen has very clearly documented limitations. We shouldn’t pretend that it does everything well. It is not a general purpose GI solution. More importantly the availability of RT hardware is a catalyst for ray/path tracing innovation and optimization that wouldn’t be possible without it. We should be ecstatic that we’re getting to see both paths evolve simultaneously and offer a bit of variety and competition.
Didn't driveclub use dynamic GI ?Right. We were promised dynamic probe based GI last generation and it never materialized. Lumen took it one step further and actually delivered. It’s a huge leap from the baked GI or flat lighting of previous games.
Didn't driveclub use dynamic GI ?
One of the best looking games of last gen imo. Still looks great to this day. A shame it is stuck in hell nowDidn't driveclub use dynamic GI ?
That's what i generally think as well. Clearly the consoles are not there yet for full path traced games. Hell many high end PCs aren't. And certain RT features and their quality despite being hw accelerated are still a trade off that needs to be considered based on the game being made and everything.HWRT is important in the consoles if nothing else as training wheels for when it sees full use next-gen. But side-by-side, if you had one platform with HWRT and one without, the leap in visuals wouldn't be so constrained on the no-HWRT device to look dated or inferior.
This is nothing more than a testament to the brilliance of software developers in finding novel solutions, and I'm glad it's got the attention it needed to develop instead of going straight to HWRT and missing the compute-based software alternatives. I think the continued investment in developing software-solutions will be valuable in informing better practice of the hardware-assisted options, including designing better hardware.
No, software based solutions are not worth it. Epic has created new problems and now they are trying to solve them instead of going the obvious way with HWRT. Now UE5 is so slow that reflections and direct lightning get shoved away. There isnt even any new advancements here. It gets more and more bloated with every new fancy software solution while efficiency gets worse.This is nothing more than a testament to the brilliance of software developers in finding novel solutions, and I'm glad it's got the attention it needed to develop instead of going straight to HWRT and missing the compute-based software alternatives. I think the continued investment in developing software-solutions will be valuable in informing better practice of the hardware-assisted options, including designing better hardware.
They incorporated it into software global illumination and basic reflections for Crysis 1 Remastered and Crysis 2 Remastered, but also hardware accelerated them using hardware RT for increased quality. (Can it run Crysis settings + Expermintal RT Boost mode).What happened to crytek neon noir ?
No, software based solutions are not worth it. Epic has created new problems and now they are trying to solve them instead of going the obvious way with HWRT. Now UE5 is so slow that reflections and direct lightning get shoved away. There isnt even any new advancements here. It gets more and more bloated with every new fancy software solution while efficiency gets worse.
It seems the main drawback from software lumen in my eyes is that it's something unique to ue5. How likely is it that other devs will implement software GI based systems into their own engines this gen as opposed to going the traditional route with 1 or two RT features put on top?[
In regards to software lumen, can individual raytraced effects that sw lumen may not be best at(reflective surfaces with models in them) be used in conjunction with sw lumen in ue5?
I'm catapulted back to debate before this gen launched on whether RT hardware was necessary. Software GI isn't perfect, but it has come a long way since those early comparisons where it exhibited more atefacts. I feel could go either way at this point - would be nice if this gen had better RTHW as what we got don't do a lot, but it'd also have survived without any I think. Although HWRT results should improve similarly as devs find best use workarounds.
I haven’t seen any other recent examples of purely compute based GI. Frostbite devs presented a similar technique to Lumen using screen space surfel caching but the sampling was done via hardware ray tracing into a BVH. Lumen is unique in sampling from an SDF but that comes with its own issues.