Real time rendering with unreal 4... Is there any point to vray anymore?

[super gnu]

it may be much more computationally intensive, but it doesnt require a perfect model with everything nicely unwrapped, and you dont need to use complex hacks to simulate changing lighting conditions., reflections etc.. as we are finding now with BF in vray artists time > rendertime... if the hardware is powerful enough to do path tracing at 100fps on any scene you can throw at it, why use efficiency tricks?

id love to see a video of RT at 30fps!

[cheerioboy]

Optimizing for Vray RT might be easier than optimizing and exporting into a game engine. Maybe you could even bake everything and view it in Vray RT simply for a few mirror reflections..

At the moment all my experiments are with Unity. I'm going through a process of optimizing and baking textures, only to use the game engine for the player character view height and making a ground plane to walk on. This is enough for reviewing with designers and even for clients if you guide their viewpoint.

[Dariusz Makowski (Dadal)]

Yeah guys but come on. 5 years a go Vray and mental ray (mental ray till today really) were just as hard as game engines are now. You look 2 years in to future and game engines will be stupidly easy and user friendly- well maybe 5 years...

I really do hope Vray will evolve or have a sister program that will be a lot like Game engines so it can compete. Its like looking at Mental Ray and Vray. MR was for what 20+ years? It now died off to Vray. I dont want to see vray dying off to cryengine

Also on a side note game engines might be hard but just how powerfull are they. The fact that they have build in eco system> snow>rain>day>dry>desert> fog>atmosphere all pretty real time and easy to manipulate...

[samuel_bubat]

I think 3D is in pretty interesting times right now. Probably the most interesting times since I started in this business 15 years ago. A lot of technologies suddenly are available for the average user, that either are new or were just something for the gurus in their laboratories. Augmented Reality, Virtual Reality, 3D Printing, scanning respectively photogrammetry, realtime engines. Some of them will be real game changers. After getting into the Unreal engine for the last two weeks in my opinion it could replace a non-realtime renderer in many branches. I also don´t think a realtime scene is that hard to set up. For example animated films. If you look at these dozens of 3d animated children TV series. A realtime engine could easily do this job. For animations in general I can imagine a lot of cases where building a realtime model is more cost efficient than rendering for days. Realtime 3D presentations are a niche that may be something for the highend archvis and product sector. For still images I think they are not really interesting, not now and not in the near future. And that´s what the majority of Vray users is probably into. But I also think there are a lot of things non realtime renderers can learn from realtime engines and visa verse. for example the next thing we will see is real dynamic time GI (some engines allready do it). May it´s worth taking a look at Nvidias VXGI or imperfect shadow mapping. Why not implementing realtime dynamic GI engines into a non-realtime renderer as well ? Or maybe importance volumes, Screen space reflections for glossy reflections that render in realtime, reflection Cubes,
Hardware Tesselation, Parallax Occlusion Mapping. Of course all these things are not that fancy as raytraced non-realtime stuff. But often you don´t need it that correct, You just have a tight deadline and need to get the stuff done. I have no idea if things like this can be included into a raytracer like vray at all, but maybe some features are possible. After all, time is money in our business, no matter how fast computers get. If one can render an animation in days that tooks weeks some years ago, that´s great.. but if there is a dude who can make it in hours than that´s a business advantage. That´s actually why I appreciate Vray so much over other renderers. It´s so flexible and so fast. Those hurff blurff über physically correct renders like maxwell, for me are toys for hobby scientists. But those realtime engines will be a big competition for non-realtime renderes in the not to distant future. I still can imgine a Vray for game engines that does the pure GI calculations and maybe shaders within a game engine I mean realtime raytracing is also something that exists for quite some time, and one day it will make it into those engines that´s for sure. In my opinion it´s time to jump on the bandwagon

[grantwarwick]

Games are so far from looking photoreal it's a joke yet people continue to drop numbers like 5-10 years away. It's not even anything to do with the software, it's the artists limitations. It takes an artist 10+ years of solid experience to understand reality enough to begin approaching photoreal rendering. 99.9% of the work done in all rendering looks fake, there's too much detail in real life and current blockbuster movies are a great example of that. They've been stuck in that 95% photoreal realm for so long now but I haven't seen any massive improvements in the overall industry.
There will always be a few scattered gems but the artists behind that work are a very very small bunch of genius freaks.
That's my opinion anyway.

[samuel_bubat]

Games are so far from looking photoreal it's a joke yet people continue to drop numbers like 5-10 years away. It's not even anything to do with the software, it's the artists limitations. It takes an artist 10+ years of solid experience to understand reality enough to begin approaching photoreal rendering. 99.9% of the work done in all rendering looks fake, there's too much detail in real life and current blockbuster movies are a great example of that. They've been stuck in that 95% photoreal realm for so long now but I haven't seen any massive improvements in the overall industry.
There will always be a few scattered gems but the artists behind that work are a very very small bunch of genius freaks.
That's my opinion anyway.

The term photrealism is completly overrated anyway. Many many of my customers are simply not satisfied with "photorealism". They want a realistic looking image. Realistic looking and photoreal can be miles apart if you look into the detail. They want light where no light every would happen, shadows removed that a light would cast. Soft Shadows from a sun that look like our sun is seconds from imploding. Highlights removed or added where non would be. They want exposure settings that are not to accomplish with a traditional photo. Interiors bright in a shot from the outside that would be black in reallife.. and so on. Not that´s this is my personal sense of a good image, but I need to be able to produce whatever I´m asked for. And that´s what I expect from the tools I use. There is also the factor time. The good old rule over the thumb wich was the first thing I learned in this business... the last 5% quality will cost you 95% of the time. Most customers simply don´t pay you for this last 5%. Also in architecure photography many things get faked, like shadows beeing removed, dummy lights are used for example.
Regarding games and photorealism. A game usually doesn´t demand absolute photorealism. Especially the scale of an AAA title would never justify the amount of work to get the last few percent done. But when you look at the last engine Demos, than you´ll see that in theory you can get very close to.. let´s call it a perfect believable image. Just look at the unreal paris demo. Sure some small things are not perfect, like the curtains, the mirrors. But this quality for an animation and 99% of possible customers are completly satisfied. Once a scene like this is done, you can offer them half an hour animation done within a day. Because you just have to screen capture. That´s a game changer in my opinion. 3D is a wonderfull thing, but once you go from hobbyist to professional you primary motivation must be.. as hard as it sounds, money or your business won´t last long. And the competition is allready hard and it´s getting worse.

[alexnode]

I believe that there is potential in PBR and that it might actually reach critical mass in 6 years time. First problem is hardware because of the consoles all gaming development will be really low end - for at least six years (if not ten). Hardware will evolve in about 15 percent increase in performance per year - sluggish rate. Intel -Nvidia don't have anything to be scared of at the moment. Therefore there will be the need for non realtime render for quite some time in the future. 4K will makes the situation even worse for small studios.
For concept design, which is what i do, there is no need for photoreal renders all the time- so a game engine like unity and unreal could do the trick for some projects. But still it requires more work from the artist side than using max. And artist time is very important. Big studios create all these nice workflow pipelines that eliminate artist fiddling with redoing materials etc but small studios can't afford that.
Mobile technology on the other hand moves really quickly maybe they could challenge the status quo. Imagine having some parallel processing for mobile chips. Intel and Nvidia will collapse - power efficient - heat efficient - fast chips coming in the hundrends of millions.
Pricing is another problem from time to time pricing of softwares falls due to community based efforts - aggresive companies etc
Look the likes of renderware - unreal - they cost millions back in the day and after unity they got it .
Windows go free, Could the Autodesk and Adobe monopolies be challenged ? could CG ?
The time to prepare the unreal demos is of major importance. Secondly when we are talking about real world work my latest job was to design 10K square meters of an upcoming world expo -

basic architectural model size of 600mb video memory needed 8gb - exhibition model size 2 gb. video memory 12 gb. Rendering time for 8 seconds 720p video -brute force light cache (no time to do passes for the animated objects , video and people ) 2 hour over 100 dual 12 core xeons.
Your clients don't tell you let's make it easy and use procedural trees and one instanced deer or rabbit. And you don't have time to optimise anything most of the times.

[Lupaz]

That is a given; we can't really compete with something like Unity or Unreal as they get better, especially if they continue to be free. V-Ray will probably be pushed to more specialized and tricky areas.

Best regards,
Vlado

Wow. This is big coming from you.

This could mean that game engines have the potential of killing maya/max as an all purpose 3d platform in, say, 10 years...
Not to mention vue, realflow, and many other plugins that depend on these platforms.
Am I going too far?

[glorybound]

I have been doing this work for 2 decades and it's always been something, so I don't worry about things like this.

[Lupaz]

Who's worried?

[vlado]

This could mean that game engines have the potential of killing maya/max as an all purpose 3d platform in, say, 10 years...

Not really, I don't think so. Max and Maya are general tools that allow you to do things like modelling, texturing and animation; game engines only put existing content together - i.e. you need to model/paint/animate your models somewhere else first.

I was merely referring to the area of walkthrough animations, where the fact that game engines are interactive and can produce a decent result will definitely make them an attractive choice instead of traditional static pre-rendered walkthroughs.

Best regards,
Vlado

[Neilg]

I was merely referring to the area of walkthrough animations

An area of arch viz which is already dying and clients are moving away from - now a lot want directed films, something more than a standard single cam walk through.

[Syclone1]

An area of arch viz which is already dying and clients are moving away from - now a lot want directed films, something more than a standard single cam walk through.

I have to agree with you on that one, for the record I've never liked the one cam single shot fly through as they bore the hell out of me, even my own I did in the past. I much prefer the directed film approach. Self guided walk throughs are a different matter and I think if they are done well they have a huge potential.

[werticus]

When I started this thread I didn't realise how much work went into a game engine scene. I've had a play with unreal 4 and I think to get the best results out of it you will still need 3dsmax and vray anyway, just to bake textures. At the end of the day producing a VR experience with unreal 4 is a whole extra layer of work on top of what you would do for just regular arch vis.

I still see VR as the future but we will probably need vray anyway, unless we drop the 'regular' still images and flythough animations off completely, which is not likely, as clients will still want those even if VR is the future.

[khoross]

while we can't predict the future. i still see V-ray having a big role to play. 10 years ago chaosgroup were trying to find solutions for hardware limitation and as of results we had the vray-proxy , instancing . irradiance map . light cache ....... and most of competitive render engine followed in their footsteps .
nowadays hardware are not the issue, its normal to see different branches of 3D visualization taking advantage of that but that doesn't mean one would replace the other . just like having a good camera on your iphone doesnt mean that you wont be needing a Dslr anymore .