Possible future...

[Lupaz]

Hi,

 

After reading a lot about Unreal, Unity, etc, and experimenting with UE4 a little bit, many of us probably have been thinking that some things may rapidly change in the next 2 to 4 years. Especially considering that the level of realism in these engines may get to more than just decent, and very much accepted among architects, designers, advertising agencies, and even film.

 

Let's say Autodesk combines UE4's real time GI, RT post-production and real time physics into 3ds Max's viewport, and brings a huge pack of pre-made materials. 3D models are available to buy anywhere (and more is coming from the gaming community), so no need to model anything else other than the actual original design. All this is, in theory, possible TODAY.

 

1. Vray and other brute force render engines will be useless. (Edit: this is because it's too expensive to do photorealism to the smallest detail)

2. Anyone doing CAD today, will be able to do visualization.

3. There won't be a need to outsource the task, more than it is today to outsource CAD.

4. We'll see in-house or outsourced render farms and vfx and post production firms closing doors. Actually many of the biggest VFX companies did close in the last 5 years.

 

I'm interested to hear your opinion about these points. Do you see other future for the next 5 to 10 years?

Edited December 8, 2014 by Lupaz

[Corey Beaulieu]

Gaming engines still, and will until computers are incredibly powerful at low prices, require an enormous amount of setup work. Setup work that is not useful to the CAD drawings or Revit models in current architectural models.

 

Vray seems to be, very intelligently, combining the usefulness of RT's Progressive scan with the powerful and accurate CPU engine and I think this is what we will see more of. I doubt that their tools will go away.

 

Materials are a personal thing. If Autodesk created some vast library, I would still likely prefer to create my own. And if Architects begin to accept the lower quality of default as a "good enough" then the ones who will suffer most are the freelancers. Major Studios more often create large scale marketing packages than simple renders.

 

My last thought here is that I do believe the power of cloud computing will equalize the big and the small. Computing power is not a business divide in my opinion, but it will provide a lot of normalcy to the lives of the small. Perhaps many of us will begin to get some sleep again. Not to mention, there are plenty of people not rendering their work because of cost and time estimates that could suddenly enter our Viz market should they suddenly be able to afford a better quality, fast render because of innovation in the rendering engines.

 

I hardly see our future as bleak. We only need to be smart and nimble with our response to change.

[CliveG]

I think I misquote Mark Twain when I suggest that reports of the impending death of the Arch-viz industry have been greatly exaggerated.

 

I've been in or around this industry for around 30 years and I think for about 28 of those our demise was just around the corner. "You'll just be able to push a few buttons and boom..... there's your finished product" well it hasn't happened yet.

 

There will always be change and advancements, things we have to adapt to and embrace, but at the heart of it there will also always be those wanting to differentiate themselves, to punch above their weight, to out-do their competitors, so the idea that some vanilla flavoured render is going to suit everyone is a bit naive.

 

I especially like the idea that CAD jocks will be able to do visualisation.... no disrespect intended, but it's a bit chalk and cheese, heck spend enough time on this forum and you'll realise that most of us don't even know where to put the damn camera to best advantage, so what your CAD technician is going to make of it all I dread to think.

 

There's a lot more to visualisation than hardware and software, if there wasn't we'd all be producing work superior to The Third and The Seventh seeing as that was produced over 5 years ago on what we would now consider old technology.

[Chris MacDonald]

I'd agree with the above comments.

 

That engines such as UE4 and Unity can produce good (not excellent) quality results is great if you can afford the time UV Unwrapping everything to bake light maps; a process that is not conducive to changes in the design process, which almost all of us have to deal with on a daily basis.

 

The past few years I've seen lots of things come and go, all of which are touted as being the "next big thing", or "push a button and it looks good" solution yet none of them have become anywhere near as ubiquitous as the likes of VRay and Mental ray because of either the work you have to do up-front to get them to work (game engines) OR because they have been oversimplified for the layman and thus unable to achieve spectacular/unique/niche results that are often required (lumion, twinmotion, etc).

 

There's constant talk and rumour of the industry collapsing because of these new advances in software, and I wouldn't ever be quick to dismiss a piece of software but the reality is that it isn't the tools that make the image. It's the artist behind them. Take a look through the finished work section on this website and you'll see what I mean, from stunning million-dollar looking images through to the absolute trash, they all list the same software used. Vray, Mental Ray, Photoshop.

[philippelamoureux]

Hanging out on Ue4 forums quite a lot I came to the conclusion that's it's far from ready to be used in arch viz. Some people make super gorgeous stuff but most people do real crap. I guess the amount of work to achieve good result is too high for this industry. The best thing is to keep an eye on it, play with it so that when it's actually viable, we can jump in!

 

For now I don't see a huge menace for offline renderers. I hope to see more people using GPUs in this field tho. There are almost no communities or tutorials/documentation for renderers like Octane...while there are a METRIC TON of them for vray. mental ray, etc.

 

I can say that if the ue4 workflow (mostly lighting) was more adapted for us, I would make the jump tomorrow. It's so much more fun to pre-viz everything in real-time, navigate your scene, etc Ue4 is such an amazing software, especially for 20$. One day one day!

[Scott Schroeder]

Wasn't TwinMotion the supposed "render killer" 4-5 years ago? How's that working out? Here's the thing about real time that a lot of people fail to remember. Your scene ultimately has to fit on the GPU. So to fit your unoptimized and poorly organized files, because hey real time is the easy button of rendering, onto your GPU you have to go out and buy a K6000 for $5,000+. Then you have to update your PC to power and cool that card, so there's a few more bucks into the mix. How is this cheaper than Vray again?

 

With Autodesk acquiring BitSquid, they did sort of hint hint wink wink that there would be some 3ds Max direct viewport integration during their talks at AU. Even if they do, you still need an artist to do the work.

 

I fully agree with Corey. Cloud computing is going to be the great equalizer. I now have at my finger tips, and relatively inexpensive on-demand render farm with say something like EC2. For larger firms, EC2 now means they can take that one extra cool job on and still be able to render their in the pipeline jobs.

 

This is not Ratatouille. Not everyone can cook. I mean render.

[philippelamoureux]

Hey Scott,

 

On the bright side, at least we have very good GPUs now...Eventually memory will not be a big issue (gtx 980 with 8gb coming soon?), especially if we can get a damn cloud solution ASAP. I'm trying X.io at the moment, with blender, sketchup...pretty decent...looking foward to it in the future! Currently blender with X.io is running on a single gridk520 but I guess it will be elastic...Otoy showed 120 gpu in 2013 running together in the cloud.

 

Have you tested EC2 Scott? my next thing to-do is Renegatt gpubox (running on EC2) for octane render.

Edited December 9, 2014 by philippelamoureux

[Devin Johnston]

Do you see other future for the next 5 to 10 years?

 

This question gets asked at least once a year and I think all the reply's are spot on.

 

Our industry isn't going away until someone creates a AI that can do what we can do with artistic flare. Maybe that sounds a little too Sci-fi but a lot of articles I'm reading are talking about how robots and AI will eventually eliminate most blue & white color jobs. I think this future is still many decades if not a century away but I see no reason why this can't eventually happen unless you believe that a computer can't be artistically creative. That's mostly true today but there are some programs that can paint pretty nice abstract art and even one that can compose music that's indistinguishable from the man made stuff. I know that's a far cry from being able to create a beautiful rendering without human input but I think it will one day be possible but most of us will be long retired by then.

[numerobis]

Let's say Autodesk combines UE4's real time GI, RT post-production and real time physics into 3ds Max's viewport, and brings a huge pack of pre-made materials.

...

1. Vray and other brute force render engines will be useless.

 

UE4 does not have real time GI, it has to be baked. Even in combination with Enlighten as real time "GI" solution i haven't seen any full GI scene (for now).

 

Brute force will not be useless, i think in the not to distant future real time GI engines will be based on brute force calculation. And i think this will be the future.

http://brigade3.com/

 

When softwares like Lumion integrate these better lighting methods and complex physically based materials then i think that it will be a game changer and combined with BIM and bigger and better libraries i think more visuals will be done inhouse.

Sure, its not only the tool that makes the image, and there will always be the high end work but in many cases its the fast and cheap image that counts and i think the quality level and usability in this area will be greatly improved in the near future.

[philippelamoureux]

UE4 does not have real time GI, it has to be baked. Even in combination with Enlighten as real time "GI" solution i haven't seen any full GI scene (for now).

 

Brute force will not be useless, i think in the not to distant future real time GI engines will be based on brute force calculation. And i think this will be the future.

http://brigade3.com/

 

When softwares like Lumion integrate these better lighting methods and complex physically based materials then i think that it will be a game changer and combined with BIM and bigger and better libraries i think more visuals will be done inhouse.

Sure, its not only the tool that makes the image, and there will always be the high end work but in many cases its the fast and cheap image that counts and i think the quality level and usability in this area will be greatly improved in the near future.

 

In fact, UE4 has global illumination, it's called Light propagation volume (LPV) but it's low quality. Good for games maybe, but not architectural visualization.

 

I agree, that brigade 3.0 stuff is very impressive!

Edited December 9, 2014 by philippelamoureux

[Lupaz]

Hi,

 

I didn't say it's the end of arch-viz, and I don't think it's a bleak future. On the contrary, I'm very excited about it.

 

But I feel that the workflow to produce arch-viz will completely change the same way photography did with digital cameras. Remember the quality of the first digital photos? crappy.

 

Today, we do everything manually. This is what I believe will change in 2 to 4 years.

I cannot see how vray will adjust itself to give us the automation that game engines provide.

 

In fact, UE4 has global illumination, it's called Light propagation volume (LPV) but it's low quality. Good for games maybe, but not architectural visualization.

 

To me this is huge. In 2 or 3 years the quality will be more than acceptable for the untrained eye.

And what about real time post production, including RT DOF, RT lens flares, RT bloom, RT atmosphere, exposure auto-adjustment, etc.? Amazing!! I don't want to place lens flares manually again.

 

Materials are a personal thing. If Autodesk created some vast library, I would still likely prefer to create my own.

 

How materials are a personal thing? If the materials that provide a certain software are really good, and you can fine tune them (UE4 for example), it would be much better than starting every material from scratch, like we do now with Vray.

Edited December 10, 2014 by Lupaz

[aristocratic3d]

BTW, You cant make a robot lie. if you see a robot lying then rest assured its a technical error. ha ha.

 

I Dont become hypothetical too much. let NASA build some house in Mars first and then pack your luggage to travel there.

And keep in mind no matter where you go, once you will die.

[Rich O]

Interesting re Lumion, but when it comes to setting up bespoke objects for a render I think this is where using actual DCC software comes in handy. Sometimes you just need to do stuff yourself.

Also I think the material library point is rather moot, as there already exist asset and material libraries and I'm sure we all have our own as well. I guess it lowers the bar of entry somewhat but until there is software like an extreme version of RichDirt to automatically add those natural imperfections to materials with just a few sliders, libraries can only go so far without human input.

Convincing lighting and post processing on the other hand... I'm really liking the idea of stuff like Brigade for that.

[numerobis]

In fact, UE4 has global illumination, it's called Light propagation volume (LPV) but it's low quality. Good for games maybe, but not architectural visualization.

As far as i know, the implementation of LPV in UE4 is still in beta stage and yes, very low quality and buggy at the moment. I think the LPV version in Cryengine is developed further but is still limited.

[philippelamoureux]

As far as i know, the implementation of LPV in UE4 is still in beta stage and yes, very low quality and buggy at the moment. I think the LPV version in Cryengine is developed further but is still limited.

 

You are correct! it's unfortunate!

[eagle_ear]

prediction for the future: more predictions about the possible future, sometime in the future.

[Nic H]

IMO its the end of arch vis as we know it

foreign freelance armies will soon swarm the world

[braddewald]

Today Ronen Bekerman posted a tutorial by Lasse Rode on his UE4 approach.

 

Article: http://www.ronenbekerman.com/striving-for-real-time-photorealism-in-architectural-visualisation/

Video:

 

I've got to say, if it's as easy as he makes it sound, I'm starting to wonder if trying it out myself is in the near future...

[philippelamoureux]

Today Ronen Bekerman posted a tutorial by Lasse Rode on his UE4 approach.

 

Article: http://www.ronenbekerman.com/striving-for-real-time-photorealism-in-architectural-visualisation/

Video:

 

I've got to say, if it's as easy as he makes it sound, I'm starting to wonder if trying it out myself is in the near future...

 

It's not overly complicated but, imo, it's just a pain in the ass. The Ue4 engine is a fantastic tool. I wish I could work with it all the time. Do you prefer to wait for the image to render...or have it render instantly (realtime) but spend alot of times manually unwrapping UV's and tweaking them to perfection to hope achieve decent shadows in your scenes?.

 

I imported a evermotion scene in ue4 for test purpose a while ago. https://www.youtube.com/watch?v=4coAflBE__w As you can see the lighting is shit (it's LPV) but it was easy to import in Ue4. Materials we're easy to make too. Dynamic sky etc is really cool in ue4. Recorded with shadowplay. I think a strong point of Ue4 is it's material editor. Materials can looks very good in that engine when you know what you're doing.

 

I tried my hardest to get that part fun, but wasn't able too. UV mapping, placing islands, pelt mapping etc is something I truly hate. I prefer to have 15 hours rendering time while I sleep lol.

 

On the other hand, making anims/fly tru is a pure joy in Ue4. Here's a link to a guy working on a realtime raytracing solution for Ue4. Dunno if it's gonna work for real in the end but it looks cool so far. The day there won't be the need to unwrap uv's....Ue4 is going to absolutely rock!

 

Pictures on page 4 : https://forums.unrealengine.com/showthread.php?44278-Realtime-Dynamic-GI-Reflections-AO-Emissive-plugin-AHR/page4

 

Studio Xoio made this render with ue4...http://xoio.de/winter-2015-interaktive-simulation/. Not bad render for using a real time engine!!! Probably lot of post prod work tho.

[Graphite]

 

This is not Ratatouille. Not everyone can cook. I mean render.

 

This.

[RyderSK]

Even once the easiest, one-button solution (be it whatever tech forms up to it, real-time game engine or other) becomes available on market, it won't change much. It's mere continual cycle of technological advancement of tools that's been going on for quite a long already, it just happens to be more exciting at certain times.

 

Even if we hypothetically discard the fact it's still in major case the creativity that drives our work forward and makes it useful and attractive for clients, the clients would nonetheless not have time nor interest to do the work unless it would be extremely financially beneficial for them, which in case of the worthy clients isn't issue anyway. Most of the time I am not even dealing with architects directly, but rather hired by 'suits' (whether it's developer shielding the actual project or marketing agency building up collateral) and they don't do any creative work nor have intention of.

 

 

From technical scope regarding purely Real-time vs Offline (Vray) rendering, I've been toying with real-time engines for more than 6 years, far longer than I do rendering, so I will mention bit my view. I did my architectural project in CryEngine. The evolution is great, but is far slower than people make it out to be. Even the global illumination in current most popular factor (LPV) has been around since beginning, yet it still looks just as horrible as it did 6 years ago. Actually the beta version from Lionhead which is in Unreal4 (Unreal has no dedicated rendering research team, unlike Crytech, that's why they rely on 3rd parties) is even worse than the iteration CryEngine had 5 years ago. It didn't move anywhere. Now we have Voxel cone tracing coming up, from nVidia labs, which again will have to be licenced and integrated by other parties (Epic/Crytech/etc..). It looks bad, very bad compared to the standard we're used to seeing in average Vray-type of rendering. And that's now, it will look bad in next 5 years as well. It's not moving that fast as people make it out to be. If you would watch closely the industry (gaming) outside of the "Hype bursts" that occasionally happen when it crosses into other industry medias like ours, you would notice.

Baking workflow, while improving steadfastly, is still extremely laborious and unproductive towards commercial short term projects like Archviz. If you got opposite feeling from Lasse's article, he's simply coming across as overly positive :- )

Which leaves us in practical conclusion than currently, you can either sell highly expensive"great looking real-time", but it will be very laborious on artist's part (way more than offline rendering), and the necessary skill is hugely outside of reach of anyone outside of the visualization industry, or sell fast, but bad looking real-time, which has been getting easier to attain for outsiders, like architects themselves. The quality can often help in design process, but it's not useful for any other use.

 

I agree that we're getting breadth of tools and resources that make our work far easier but at same time, that makes it even more confusing for new-comers and industry outsiders. But this is still disregarding the original major drive, creativity, which is the added value clients pursue within our industry.

 

Each year, another wave of fear and excitement, but I personally don't think anything special (revolution) is happening at all :- ) Neither in near future.

Edited December 17, 2014 by RyderSK
some dyslexy here and there..

[Nils Norgren]

Here is few items from search results, after feeling deja-vu reading through the fundamental question of the thread (disregard real-time vs render, just think "industry killer"), remember, no matter the ease, quality and price of cameras, photographers have managed to scratch out a living.

 

Some of these posts are classics, peruse them, notice a thread emerge, don't forget to note the date of the posts...

 

http://forums.cgarchitect.com/6184-future-architectural-visualization-3.html#post49233

 

http://forums.cgarchitect.com/23988-changes-our-industry.html

 

http://forums.cgarchitect.com/29542-impression-autodesk-2009-danger-cg-arch-industry-disappearing.html

 

http://forums.cgarchitect.com/61031-chasing-dying-industry.html

[philippelamoureux]

Even once the easiest, one-button solution (be it whatever tech forms up to it, real-time game engine or other) becomes available on market, it won't change much. It's mere continual cycle of technological advancement of tools that's been going on for quite a long already, it just happens to be more exciting at certain times.

 

Even if we hypothetically discard the fact it's still in major case the creativity that drives our work forward and makes it useful and attractive for clients, the clients would nonetheless not have time nor interest to do the work unless it would be extremely financially beneficial for them, which in case of the worthy clients isn't issue anyway. Most of the time I am not even dealing with architects directly, but rather hired by 'suits' (whether it's developer shielding the actual project or marketing agency building up collateral) and they don't do any creative work nor have intention of.

 

 

From technical scope, I've been toying with real-time engines for more than 6 years, far longer than I do rendering. I did my architectural project in CryEngine. The evolution is great, but is far slower than people make it out to be. Even the global illumination in current most popular factor (LPV) has been around since beginning, yet it still looks just as horrible as it did 6 years ago. Actually the beta version from Lionhead which is in Unreal4 (Unreal has no dedicated rendering research team, unlike Crytech, that's why they rely on 3rd parties) is even worse than the iteration CryEngine had 5 years ago. It didn't move anywhere. Now we have Voxel cone tracing coming up, from nVidia labs, which again will have to be licenced and integrated by other parties (Epic/Crytech/etc..). It looks bad, very bad compared to the standard we're used to seeing in average Vray-type of rendering. And that's now, it will look bad in next 5 years as well. It's not moving that fast as people make it out to be. If you would watch closely the industry (gaming) outside of the "Hype bursts" that occasionally happen when it crosses into other industry medias like ours, you would notice.

Baking workflow, while improving steadfastly, is still extremely laborious and unproductive towards commercial short term projects like Archviz. If you got opposite feeling from Lasse's article, then you didn't pay much attention or he's simply coming across as overly positive. Since we talked of it quite much, it's not.

Which leaves us in practical conclusion than currently, you can either sell highly expensive"great looking real-time", but it will be very laborious on artist's part (way more than offline rendering), and the necessary skill is hugely outside of reach of anyone outside of the visualization industry, or sell fast, but bad looking real-time, which has been getting easier to attain for outsiders, like architects themselves. The quality can often help in design process, but it's not useful for any other use.

 

Each year, another wave of fear and excitement, but I don't think anything special (revolution) is happening at all :- ) Neither in near future.

 

Amen!

 

You can go to ue4 official forums and see that 99% of the ''arch viz'' people do there is literally trash. Once in a while a good project will emerge but we all know the time and effort these guys put to make their ''proof of concept''.

 

At the same time I guess it's possible to make money selling low-end real-time visualization ''experiences'' but i'd hate to make work that don't make me proud in the end!!!

 

One thing i'd like to see improved rapidly is the interactivity, even for offline renderer. Material previewing etc. I think most render softwares are trying to improve in this field as we speak. It's something that enhance our workflow tremendously imo.

 

The unreal engine workflow, you have to try it to understand (hate) it lol.

[lasse1309]

hey,

 

well, honestly - it is very fast, you basically only have to seperate your modelling from the app.. even the unwrapping thing has become easier a lot. it generates automatically on import, and in most cases that is good enough!

 

cheers

Lasse

[philippelamoureux]

hey,

 

well, honestly - it is very fast, you basically only have to seperate your modelling from the app.. even the unwrapping thing has become easier a lot. it generates automatically on import, and in most cases that is good enough!

 

cheers

Lasse

 

Yea but I think it's easy with simple models, but what about very complex models? organic shapes and all...? I tried without success but I not experienced in that field...might be me! Also, we've mostly seen small interior scenes done with ue4...what about a full scale exterior project? I'd like to see that.