Enviroment | Visualisation
Interiors – 2022

selection of UE4 projects

Here are a couple of examples of 3d environments prepared within Unreal Engine. Mostly, I’ve used premade assets and was responsible for level design, composition and lighting. Levels were later used for Machine Learning processes.

My client presented some restrictions that needed to be taken into consideration. We had to use an old version of Unreal Engine which didn’t support ray-tracing and other modern solutions. Also, I couldn’t use post-process volumes, as it was overwritten by client’s rendering pipeline. Lastly, I had to utilize existing library of assets, but was able to add small details when needed.

Warehouse

For this project I designed whole level from scratch. Assets used were provided by a client. I had a complete freedom of composition and was loosly informed what needs to be included. 

Apartment

Here I recieved already made environment, but a client needed enrichments within this level, so I populated empty spaces with more assets and redefined lighting. Also, I had to fix several assets with broken UVs, double vertices, etc.

Office

Same as above I recieved already made environment, but a client needed enrichments within this level, so I populated empty spaces with more assets and redefined lighting. Also, I had to fix several assets with broken UVs, double vertices, etc.

House

Same as above I recieved already made environment, but a client needed enrichments within this level, so I populated empty spaces with more assets and redefined lighting. Also, I had to fix several assets with broken UVs, double vertices, etc.

Summary

For these environments it was required to use 4.21 version of an Unreal Engine due to pipeline technical limitations. Other limiting aspect was necessity of utilizing assets from existing library. Luckily, there was a possibility to improve some of existing models and materials in the process. It was interdisciplinary project where I had a freedom to decide how final levels get to look.

Creation of these levels was just a beginning of work. Later I had to generate camera trajectories and triggered events for Machine Learning purposes. Thanks to these environments I was able to deliver diverse data and as a result client was able to improve object recognition ratio.

Pin It on Pinterest