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

[Mokiki]

Vlado lets assume there are people waiting to buy vray the moment it has some build in lensflares. How many new adopters would you need in order to make developing such a feature worth it?

[vlado]

I would tell them to simply get photoshop or something. I really don't want to spend time on that - regardless of how many people are waiting for it.

Best regards,
Vlado

[super gnu]

i was being sarcastic by the way.. i dont want lensflares in vray. however, if we could get it to play a selection of your favourite music while it renders......

[cheerioboy]

I'm interested in this topic. Although I don't see UE4 replacing Vray, there is definitely going to be a merging of pre-rendered vs real-time 3d. I'm hoping more for Vray integration within UE4 or Unity

Either as a baking solution or somehow optimizing Vray RT to work in a game engine. I just want to use Vray for VR!

[super gnu]

well assuming a very approximate rendertime of 1 hr a frame on a modern gpu for a noise free image, we just need to wait for gpus to be 3600 times faster and we can have RT at 60fps.


or of course buy 3600 gpus...

[Crush]

I can imagine a niche market for the kind of quality and possibilities that realtime offers, but at least for most of our clients "good enough" is not good enough.
I quite like to adjust things like exposure and white balance in realtime tough, and VRay already took steps in this direction with the latest VFB features.

[Deflaminis]

No question Vray does far more, but I understand what some posters mean about ease of use for things like SSS or bloom/glare. Feels like we're comparing apples to oranges though here. Isn't the goal of Vray to be a physically accurate renderer? We're sure not getting that out of unreal, though it's faked enough to look pretty decent no doubt. Seems to me the post fx are what makes the unreal render look this good.

Can't help but think that people are expecting too much from a render software, it's not an FX software. Might be better to ask if Chaosgroup will ever release a plugin for photoshop that reads things like zdepth and pixel brightness and create some of these fake fx. Asking for these things in vray cheapens the point of vray, but asking for a PS plugin as a separate product might not be too out of the question perhaps? Even photoshop kinda sucks at dealing with 32bit images when it comes to postfx and I think that's where the real problem is. Not everyone can afford Nuke. There's a weakness in the pipeline that isn't the problem of Chaosgroup, but Chaosgroup might be able to fill it in. You see some progress here in Arion to fill this same need in their PS plugin.

That being said, vray is amazing and I really appreciate all the support and work that Chaosgroup has done to get us here.

[Companioncube]

there is an article today on arch daily about the pros and cons http://www.archdaily.com/607849/unre...o-game-engine/

[Lupaz]

They are all post-effects. If they work for you, this is great, but I'm not really interested in going there. The goal of V-Ray is to simulate DOF and atmospheric effects accurately, even if this is (much) slower. Bloom you can already do in the VFB, but there are no plans for lens flares.

Best regards,
Vlado

That wasn't the case of vray when it started. Isn't (or wasn't) vray a biased renderer? IMHO it was Vray's speed what got it to where it is now, at the expense of accuracy (physically).

Add to vray the possibility of having post effects like unreal, and you get an amazing engine. No after effects or photoshop. Just render and you're good to go.

[Neilg]

I dont feel like many people commenting on this subject have actually tried to build something and convert it to unreal engine. the idea of fast revisions just because the viewport has shadows and can run at 60fps is laughable.

even if the image quality does compare shot for shot, getting a decent looking image out of ue4 is the result of a phenomenal amount of work. it took us weeks to convert one of our most basic scenes to ue4 at a comparable quality, and the other 90% of work we do I wouldnt even know where to begin, or if it was possible.
it goes against everything i've ever known about architecture - if you're not expected to bang out a model revision in 24 hours (which with ue4 is pretty much starting over completely), then you're working with billions of polygons.

vlado, keep doing what you're doing.

[vlado]

That wasn't the case of vray when it started. Isn't (or wasn't) vray a biased renderer? IMHO it was Vray's speed what got it to where it is now, at the expense of accuracy (physically).

There are limits to everything

Best regards,
Vlado

[Dariusz Makowski (Dadal)]

Its tricky question. I think in 2 years a lot of people will move from render engines to game engines as they will simply become fast and good enough to replace them. At that time render engines will be strictly used for Movies and high end production. In 10 years I'm pretty sure everything will be instant and a new bread of engines will eat us alive. Unless vray will do dramatic change and evolve he will lost some users...

[vlado]

Its tricky question. I think in 2 years a lot of people will move from render engines to game engines as they will simply become fast and good enough to replace them. At that time render engines will be strictly used for Movies and high end production. In 10 years I'm pretty sure everything will be instant and a new bread of engines will eat us alive. Unless vray will do dramatic change and evolve he will lost some users...

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

[super gnu]

you could also flip that argument round completely, a more realistic scenario imho.

vray rt is getting faster as it is optimised, and hardware is improved.. its a long way off, but why not RT replacing game engines?

game engines are horribly difficult to use, precisely because they have to cut every corner and use every trick in the book to get an image out fast enough. rather like renderers (more than) a few years ago. as computing power became available, less corners had to be cut, realism could be simulated more directly,and things became simpler and easier to use.

imagine if you didnt have to bother completely optimising your models and textures in order to get a reasonable image quality, but could just drop them into the game engine. any setup that offfered this, would likely have more in common with vray RT than unreal or unity.

[vlado]

vray rt is getting faster as it is optimised, and hardware is improved.. its a long way off, but why not RT replacing game engines?

Path tracing is inherently a lot more computationally expensive than precomputing stuff, so V-Ray RT will always be at a disadvantage. And not everyone needs raytracing precision for walk-throughs anyways. Btw we can actually get 20-30 fps on a VCA right now for a 1280x720 image, but it's terribly expensive hardware.

Best regards,
Vlado