Renderfarm services for unreal scenes!

Since lightmass use CPU+RAM to build lighting in a scene, I hope one day some company will open their farms for Unreal projects. Think about it, hundreds of pc building your scene :slight_smile: Incredible quality scene built in minutes!!!
This would be even be easier to setup than GPU renderfarms and the software is free. It’s not like they have to install a bazzillion vray+max licenses!!!

I will investigate more this way Running ue4 on an amazon instance is one thing but distributing the building on multiple amazon instances is another! That would be great. Multiple Xeons working on a scene at the same time :open_mouth:

I emailed Rebusfarm to ask them if ue4 lightmass on their renderfarm would be something possible eventually. Looks like they don’t share my/our enthusiasm lol.


thank you for that suggestion. We will look into it, but frankly, I doubt that’s going to happen anytime soon. So far, you are the first to request this, and with something like UE4 for Archviz, chances are it will go away just as quickly as it appeared.

best regards,


Hmmmm. Interesting Reply.

I guess the realtime aspect of unreal is a direct menace to their business. People throw hundreds of $ to render movies made with offline renderers. Building the lighting in unreal could be done in a few hours, even minutes on their farm I suppose. Not much $ to do here I guess!

Very interesting indeed. No body is requesting UE4 because it’s just starting in Architectural Visualization, It has long road to cover. Rebusfarm is in no position to declare it’s fate, that too in this demotivating way.

Don’t worry my friends, we all know what are the prospects of using UE4 and it’s not just Arch Viz that we can create but whole other options are at our disposal - game level design, advertisements, VR development and many more.


I don’t think that a huge render farm is the solution for all of us.

Rafareis123 has a super-powerfull machine which is able to get results quickly…but the most important thing is that he knows what he needs to tweak in order to get vray-like results, and that’s where is the most important trick of all!!!

At the end of the day, if you need to wait 30 minutes or 4 hours for me it doesn’t make a difference because in both cases the end result will be stunning, and that is what’s important.

With the “LEt’s make Lightmass EPIC” topic most of us understood lots of things and most of us are experimenting to get better results…you have to wait for the results? Yes, but when you’ll do another scene you already know what needs to be done and you’ll get better with just experimenting with various scenes, that’s it…as stated there’s no “one click solution” for each scene, but if you already know how to get better indirect lighting, avoid light leaks and splotches you already saved 10-15 hours or rendering time.

Save your money, buy a powerfull pc later on but in the meantime let’s religiously wait for the documentation which Rafareis is kindly giving to the community :slight_smile:

The thing is, Rafael’s scene is 2 room, a couch and, a table and 4 chairs… Of course it’s not going to take 30 hours to build. But what about large scale projects? or projects with a lot of vegetation. If we can’t do these kind of scene with lightmass we are in trouble! I have not seen a single medium/large scale project with the quality of the best interior scenes we’ve seen because these projects are impossible to build!

also, waiting 30 mins to make fast iterations or modification to a scene would be much muuuuch better than waiting 4 hours between tests. Final build isn’t everything, sometimes you have to build the lighting several time to achieve the look you need. And if your client want to change a single piece of geometry, you’ll notice the difference between 30mins and 4 hours, let alone 20+ hours!

But yea, let’s see what we can learn and improve about lightmass! Can’t wait to get more infos from the devs.

I think they underestimate the power of Virtual Reality. Probably they never tried an Oculus Rift and explored a virtual house by themselves. That`s not an mayfly

Infact that is another big issue which VR developers will have…

The new CV1 will need to run at 90fps to be smooth, so you can have use the render farm you want to render, but at the end of the day the performance is the most important thing, that’s why I will gladly spend 3k for a rock solid workstation rather then spending money on a render farm.

Post processing ( especially reflections ) are the most FPS-consuming thing right now, so some heavy tweaks needs to be done in order to achieve a good balance between quality and performance…

On the other hand if you’re using UE4 and do not plan to go full VR then a render farm could help with testing and rendering time, but I think that is going to take a bit of time for the available ones to adapt to this “new” method.

There’s even a bigger problem… will clients have the hardware to use VR content? We are the producers, it’s not a problem to have a beefy workstation but the clients will need one too.

Like Tim Sweeney said in a conference a while ago, currently we have the palm pilot of VR tech, wait until we get the ipad!

You could recommend your clients to by an oculus ready pc to ensure the VR experience they want also works.

Part of me agrees with him. Unreal has a lot of issues to overcome to be used in a serious production environment. You can’t be guessing what works and what doesn’t and as far as I’m concerned lightmaps are a BIG problem. We really need to be able to find solutions to “little” problems quickly and right now we have big problems - not good.

What’s going to happen when your client has a model built already and they want it done quickly but the model isn’t built suitable for lightmaps? You’re not going to be able to make a new one.

If you put your clients inside of their virtual houses inside of VR I think they could care less about perfect shadows and reflections. Resolution will suck for a while anyway. Archviz people seem so hellbent on quality they forget the simple pleasure being able to move around in your own space as opposed to just looking at pretty pictures.

With real time GI around the corner, I ask what’s better? Perfectly baked but static GI or or the ability to change the time of day and turn lights on and off in the house? Download and play around with this if you haven’t already. It’s Cryengine, but completely dynamic GI with just 2 bounces.

Oh and I’m shocked at the ignorance in the reply of rebusfarm. Whatever makes them think that UE4 will go away for archviz? Maybe I just don’t understand the space enough to grasp how any client would prefer pretty V-Ray pictures over the possibility to actually walk around in a house?

Their reply is kinda disappointing, but they may be right… UE4 became famous for realtime archviz because of its great visuals and active community that makes it easy to learn. But UE4 is still an IDE for game development. It’s easy to see how people are struggling to get Lightmass to attend archviz standards. What is going to happen when software like Lumion and Autodesk Stingray start to appear with same visual quality and way easier to use? I think people will massively migrate… I bet those companies are aware of the UE4’s VR and interactivity capabilities and will find a way to deliver similar features on a much more friendly IDE for archviz artists…

IMHO, if Epic Games want to keep their archviz users, they must invest on specific IDE and lighting features for archviz. The lightmass dev participation on that lightmass topic should be just the beginning…

That cryengine svoti demo still looks like it’s a videogame. It’s nowhere near as impressive as what lightmass does :

As a customer, even for VR, I’d prefer hyper-realism and static vs interactive that looks game-ish. But that’s me tho.

Update : I’ve tested the cryengine scene on my pc and it’s worse than I expected. No contact shadows, very bad indirect shadows around objects. It runs at 60 fps tho. Also, forget svoti or vxgi if you wanna make vr stuff!

Fair enough. I just doubt that’s true for most customers. The CryEngine demo is very pretty already (doesn’t look particularly videogamey to me) but has flashy dynamic lighting. I’d rather see what kind of light the sun is going to throw into my future house rather than having the most hyper-realism. Or how things are going to look at night with just interior lighting. And don’t forget this is running dynamic GI here and now, quality will undoubtedly increase a lot in coming years.

Yes you are right, You can look at what Brigade can do (from Otoy) and have a glimpse at what dynamic G.I could be in the future. That’s running on like 200 gpus iirc.

I can attest and surprisingly so, most of our clients are okay with an imperfect game like render. Especially, when we can turn animations around in little time and cost. Like it or not, its a disruptive media. I love V-Ray and have spent thousands on Rebus, but the majority of our clients cant see the difference. Those that master the workflow will produce animations quicker and cheaper. It only gets better from here.

There’s several end users. There’s designers using it as a tool then there’s clients wanting to know what they’re getting. The designers won’t care as much about quality for the design process but a lot of Architects are going to want the best quality they can get for the best price.