Except that modern off-the-shelf engines like UE5 and Unity already support raytracing... And Metro devs also already have an engine working with RT, among other devs. It's not that hard.

The majority are gaming on consoles BECAUSE its a known spec for many years. Dont think a 2x more expensive gpu than a single console is going to change the overall landscape. And you seem to forgetting that consoles launch weaker than top end pc to hit the 400-500€ "MSRP". If you had a new console generation next year you would see so minor improvements that it wouldnt make sense to abandon the current consoles. That matrix demo ran at 24-30fps while dropping frames, you shouldnt get swept up by cross-marketing gimmicks.

dude im not getting swept up at all. you seem to forget UE5 its self is still beta and the matrix thing was a tech demo. running on like you said midrange £500 area. so next gen in about 5 years. like i said in my first comment it was only 6 years ago brigade engine 3.0 took 4 GTX titans to get a still image to render at about 1fps when raytraced

i was replying as to why developers are pushing for it. this is just one of a few i have come across myself in the few months of playing around in EU4/5. and unfortunately as humans are like magpies and want the shiny shiny storytelling is going to continue its downward trend in favour for fidelity. which raytracing would almost certainly free up more development time in the cycle for just by the nature of how it works. no more build times every time you add/remove/move a light source. no more programming light movements for moveable objects.

And how do you "upgrade" ps5/xboners that only raytrace lighting (while looking better than baked lighting) with underpowerd hardware and aiming for 1080p+ 60fps (as most games this generation are aiming). Or alienate majority of pc gpus that lack raytracing hardware(i doubt that is going to change too much 4+years from now).

i never said anything about upgrading consoles. what's a generation these days? 5 years? next years AMD 7900xt is supposedly going to be around 2.7x the performance of the 6900xt which i can almost get 60 in any raytraced game at 1440p ultra today. so 3 times that would be more than fine for 1080p 60fps IF a next gen console came NEXT YEAR. imagine what there will be in 5 years time? next gen of consoles is going to be fully ray traced and possibly near photo realism. check out the matrix EU5 demo running on a ps5. YeongYea done a review of it . incredible

Thanks for mentioning the UE5 demo. It blew my mind!

That being said, the thing that still bugs the crap out of me are that screen-space reflections are still terrible... I had to watch someone play the demo city (because the demo isn't on PC - thanks for that, marketing!) and as soon as the character walks past any window panels - the reflections just fall apart for me. Other than that it's pretty...unreal :)

Next gen? Probably not, at least not full-on raytracing in most titles. Small titles like Quake II RTX already ran very well on last gen GPUs, but they obviously don't match up in quality to the likes of GTA V, bar lighting/reflection quality (it was too fun to play with fully dynamic lighting in Q2 for me). You're right, the consoles and their price point will keep being a setback for such tech advancements - no doubt about it. Thinking how even the latest XBONE SX can't do proper 4K at ultra in FH5 when a PC with a 3090 can do it no problem and even go for 5K - the price obviously matters here.

That being said - at some point it will happen. Maybe not the next gen - perhaps the one after, or the one after that, but it IS where the industry is going and has gone already for us in the offline world. When I do my CG - I don't have to worry about baking normal maps, optimizing geometry, saving memory, baking lighting or anything like that - I use the highest quality assets I can muster, I apply physically-based shaders, lighting, etc, I set up my camera as I would for my photography and hit render, letting the GPUs bruteforce through trillions of rays to brings me the final image. I'm already impressed by the quality and speed of offline rendering - what used to take hours now takes mere minutes! Similarly - where raytracing used to be only possible by letting the PC churn out the math for hours - we can now do it real-time on the GPU (as a side note - Nanite in UE5 also tries to break away from having to bake normal maps and optimize geometry by letting people use full quality assets. Not RT related, but also a step in the right direction). If anything - this transitional stage happened faster than I thought it would - I genuinely thought it needed another 5-10 years, but then Nvidia came out with RTX and blew my mind. What comes next is exciting! Well, for me, anyway... I see a lot of people complain about it, I get why, but it also sucks to see a lot of uneducated comments like "gimmick" when it comes to RT. If anything - raster is the fake one :D

Dear, Mr Lead Software Engineer,

GPUs cannot possibly start taking "little to no performance hit" in raytracing vs raster the same way as Maxwell's equations can't be computationally less intense than 2+2. Shooting out trillions and billions of rays into a virtual simulation of a physical world is always going to be more resource-intensive than any form of fancy triangle shading. Your silly little AA and AO comparisons are not even close to the same level of performance difference. You should know that - you're a lead software engineer!

What you describe here is not that - you describe the *practicality* of raytracing, where a raytraced game runs "acceptably well" for the public to not be warranted a nuisance anymore. Fair point - that is what matters in the real world, even if you think I went on a tangent. But then where is that line of acceptance? Since RT is just so much more computationally intensive, where a pure raster game can run at 3000fps - a pure RT game may run at, say 60. Is that worth the downgrade? What about potential resolutions, 16K vs 1-2K? These comparisons may sound ridiculous, but if we think where we started (240p or less on digital, 10fps video, etc) and where we are now (8-16K content and displays, high framerates for movies, games and displays, HDR, etc) - where will we draw the line where raster's speed is no longer desired by most people, so we can now move on to RT as a standard? Every tech upgrade we get is to gives us MORE, not less or same - faster GPU for moar fps, more Ks on the display, maybe an OLED for better contrast, etc. Where will people be ok with the massive tradeoff and/or stagnation? They're certainly not happy about trading 4K rasters for 1080p with limited RT or an FPS impact on the latest games, so why would they be happy trading the same later? Just by mere exposure if we wait long-enough?

The changeover to RT and full-res assets will have to happen, I already said it and it's undeniable. But that will happen to make game development easier, to make the visuals marginally better - all that good stuff. That performance impact will still be there, though, and it will impact the industry's decisions - either the 1440p-4K era will last way longer than it would have on raster, maybe the game visuals will stagnate for a while while more layers of RT replace raster, maybe we'll have to accept lower framerates. We are already seeing this in action with RT games spawning things like extra ray accelerator hardware, DLSS and FSR to combat these common downgrade annoyances and bring RT performance more in line with raster. Problem is - as these current RT implementations are very limited. As hardware evolves - we'll be able to add more and more layers of RT, at better qualities, etc - but raster will be miles faster than that more complex RT on the now-faster hardware! Whatever the tradeoffs remain - the performance hit is there and it will be felt somewhere down the line, perhaps even invisible to the end consumer unless prompted (though costs most people can definitely feel - the extra hardware doesn't come free).

All the above was a reply to YOU, Mr Lead Software Engineer. The reason for that - the OP clearly started he prefers more raw (raster) performance over RT performance - and his sentiments are far from uncommon. My reply to him explains that there cannot possibly be no hit, but RT will have to become standard some day anyway due to its other benefits. If this is how badly you follow a simple topic online - perhaps this explains why software always comes out buggy. Or perhaps it is on me for not explaining the effects of those innate performance negatives better like I did above (resolution/framerate stagnation while having to buy better hardware, lower fps, upscaling, costs, etc). Either way - next time rather than wanting to flex on somebody, actually think about what they're saying and whether or not to spend 3 paragraphs moaning about the wrong point to them and with bad examples.

Sincerely,

Please at least be more attentive than this in your software engineering endeavors

You want a slow gimmick for you instead of a clap?