Download

ArchVis construction. - Modular verses single object construction comparison.

Since UE4 uses physics based rendering I did a bit of testing as to environment construction that I would normally do using 3ds Max using the Realistic Rendering sample available for anyone to download.

https://youtube.com/watch?v=4kBryBatURQ

As a benchmark I did not change any of the lighting set up and only turned off AO for the final result but even then the result of going from modular to single object construction is obvious.

Very interesting test, man! Thanks for sharing it! :smiley:

Hey FrankieV,
Really nice video and very good tests which actually allows you to better understand what’s going on there.
I also tried two different methods:
One is to build a 1mx1mx10cm blocks to be used as modular parts and a unique mesh ( with the chamfer trick which works good ) for each single room and a big ( external ) mesh to enclose all those meshes, since they’re one side.
I still find difficulties to get nice clean lighting but mostly because I’m trying to get very good and detailed contact shadows for furnitures, and the way I’m doing it is to mess with the indirect lighting smooth and static level…

It would be interesting to get the files you tested available to check so that everyone can see the various situation and how the lighting behaves using different approaches.

PS: The unique mesh with the chamfer, the lightmap is 2k or 4k?

Interesting video Frankie, I’ve learnt a few things. I’m not convinced about the one big mesh for the whole room tho. I had more success building modular (with lot of detached faces on my meshes) to use the most of the lightmap 0 to 1 space. What is the lightmap resolution of your room mesh?

Thanks for the video FrankieV, it was very informative. I might try this out myself when I have the time.

The light map resolution is 1024.

The experiment was to see what the difference would be between modular and single modular construction as to being significantly different and not necessarily better. Turning off the AO was as far as I wanted to go and make the comparison based as close to the original parameters as possible.