Pointopia — Haochen(Jimmy) Xu

Haochen(Jimmy) Xu

jimmyhcx@gmail.com
@jimmyhcx
/in/jimmyhcx

Unreal Engine Lighting & Cinematic Artist

Bridging lighting, camera, and spatial storytelling through real-time worldbuilding.

I am an Unreal Engine Lighting & Cinematic Artist specializing in lighting, shot design, and spatial storytelling for narrative-driven real-time experiences. My work focuses on shaping mood, composition, and emotional pacing through light, camera, and environment.

With a foundation in architecture, I work across games, short films, music videos, and previs pipelines, collaborating closely with art teams to create cinematic real-time experiences for film, games, and virtual production.

CV

CONTACT

POINTOPIA

BACK

POINTOPIA
Unreal Lidar Visual/Digital reconstruction

Co-Designer
Sep 2023-Dec 2023

Created by:
Haochen Xu
Firdavs Yuldashev

DESCRIPTION

Our goal was to explore point cloud as a design tool and a universal system where all types of models can unify. Downgrading and simplifying everything into a single point cloud could open up more opportunities for us to imagine more complex landscapes. Formats like mesh and NURBS overcomplicate the process for us and, as a result, restrict the level of complexity we can achieve when modeling. So the answer may not be making the workflow more complex, but rather reducing it to a point cloud and using it as a universal unit. The potential of the point cloud also lies in the fact that the differences between all the models that are crafted, scanned, or AI-generated can become indistinguishable. At the beginning of the semester, we picked and documented the abandoned Hawthorne Mall as our site of choice. Using a Leica LiDAR scanner, we were able to capture a point-cloud 3D scan of the entire building, which gave us a lot of room to work with.

In the process of using point cloud as a design tool, we utilized and incorporated multiple different formats and ways of modeling, such as crafted mesh objects, 3D scans, AI-generated models, and so on. All of the models would then undergo a process where we simplified them into point-cloud data and populated the point-cloud scan of the mall. It resulted in something that we found very interesting and novel — a playground made of tiny dots that seemed to blend everything so well and blur the edges of objects to an extent that made everything read as a single unified landscape rather than distinguished formats, models, and styles. We imagine that in the future, perhaps a software package will exist that could use point cloud as its base model unit to make the interoperability of different formats much easier than current software packages do.