I work in an architectural visualisation company. We have been creating images and animations for quite some time now, but, despite wanting to, never really moved into realtime game engines.
I have decided to give Unity a try.
A current project of ours is a 4 storey art gallery and museum. We have been producing CGIs of the building for a while - in fact, our images helped the architect win the competition for the building design in the first place.
We are now creating some updated images for a couple of the interior galleries. I have an idea that would take it much, much further though. I want to take our 3D model of the gallery (modelled in AutoCAD but textured and rendered using Vray in 3dsmax) and put it into a ganme engine, like Unity. I then want to add the artworks and exhibits to the gallery and launch it as an online virtual art gallery and museum for the world to see. There are a million ways this could develop, but I need to make a start and see what is possible.
Is Unity a good starting point? From what I have seen, it could be. Are there any potential pitfalls I should be aware of? For instance, my major concerns are that it should be very stable (ie not crash/freeze) and that most people’s home computers should be able to view and navigate the gallery with relative ease. Are file sizes going to be huge?
Here is a link to one of the CGIs we created for the building. This is the basis of the model we would be using for the virtual tour. http://bit.ly/cvw5bX
I’d appreciate any thoughts to this absolute beginner!
If you want to use the models as-is, then Unity (or any other engine) isn’t going to be appropriate. The main thing is that you’d have to redo the 3D models and textures so they are appropriate for realtime 3D on a wide range of computers. If you’re OK with that, then indeed Unity would work well since many people use it for visualization.
There are a few hurdles to go through, first would probably be that there is a bit of a jump needed to get a handle on the scripting basics. camera controls, loading assets downstream, and shaders being probably the biggest holes to learn to fill for arch.
there are quite a lot of assets out there to pull from as starting points, but not many area targeted for our use, so again, some scripting is needed.
but the community is very large and very very helpful!!
that said, there are a few model organizational hurdles that are different from what we’re used to in the offline rendering world…
There isn’t really any mesh instancing currently, the performance hurdles are more in the number of objects and number of materials sides of life. some thought needs to be taken when baking your textures currently, most of the time is spent in the unwrapping phases since the automatic modes are a bit too automatic. hopefully the upcoming addition of the beast light mapping built in will help out here in the future.
Its certainly doable, theres some custom scripting needed if you want to try and set up the kind of easy quick pipe needed to keep up with arch-ish revisions. but one off projects are pretty easy to handle.
Hi Eric. Thanks for the reply. I suoppose I guessed that our models would be a bit too detailed for a straight conversion. I am happy to simplify the models where appropriate.
Texturewise, some of our bitmaps are huge (4-6k pxels wide in some cases) so these would be resized - I can use 3dsmax’s proxy bitmaps I guess.
But initially, I just want to create a rock-solid, stable ‘white’ model that just works well. I will then (hopefully) add detail until I get a happy balance.
The first issue I have come across is the lighting in the free version of Unity. It appears that I cannot use realtime shadows. This is a bit of a pain as this is a big part of the aesthetic in any architectural project. I understand why it is limited, of course. I guess when I am more accomplished, I could request a fully functional trial of the PRO version.
I will at some point give texture baking a try. Maybe that will allow me to get around some of the problems with shadows?..
There is certainly some real potential here for ArchViz work. I love the way you have a ‘timeofday’ slider in your basketball scene. Is that based on reality? If it is, that is VERY useful.
I also like the way you can switch between cameras and have the interactive view glide between them.
Our image production process involves initially creating a white block model of the building to be rendered. We then create a series of compositions that we feel best show off the building. We issue these to our client for their selection and then work up the chosen view to a high level of realism.
If we could issue our client with an interactive model like this, and have a drop down menu of all the different views we think are suitable, and have the client switch between them, this would be a real service enhancement from our point of view. If the time of day slider is, indeed, based on actual sun times at particular locations on the planet, this, again would be very powerful in allowing our client to choose the best time of day for the shot (or more likely, allow us to TELL the client which is the best time of day!).
Actually, architectural visualization almost always uses baked lighting and shadows, which gives far higher quality than real-time can possibly achieve (at least at the current level of technology).
I agree that texture baking is the way to go, vs. using the dynamic lights and shadows. I’m still new to this myself, but I’m finding that the number of ‘Draw Calls’ quickly becomes prohibitively large with each new light added - which has a very noticeable impact on peformance, especially if you add shadows to the lights. I originally bought Unity Pro thinking the dynamic shadows would make my Revit models come to life, but I’m quickly finding that lightmapping is the only way to go. It can be very time consuming, and hopefully Unity 3 with built-in Beast will make it easier, but if you want the build to be accessible and smooth running on a variety of machines, lightmapping is the way to go.
The virtual version of the gallery is a fantastic idea! You might consider adding a multi-user dimension to it at some point, if you haven’t already. There are lots of ways to achieve this, but one method I’ve been experimenting with is ReactionGrid’s ‘jibe’ platform, that enables you to upload your Unity file and host multiple users exploring the world simultaneously. Its still in early beta, but something like jibe could really add a social dimension to the gallery space.
I think Dave is right though - it will definitely require a relatively deep understanding of scripting - something I’ve been struggling to learn for several months (being a non-programmer, it is taking me a while) but I think its absolutely necessary to build a meaningful archviz project in Unity.
Maybe we can talk Dave into selling us his controls and scripted elements as a Unity prefab for architecture-specific projects? His are by far the best archviz projects I’ve seen done with Unity.
Hope this helps! Let us know how things are going, or when we can tour a preview build!
