Spotlights
July 16, 2025
How CD PROJEKT RED used RealityScan to help create The Witcher 4 Unreal Engine 5 Tech Demo
Find out how CD PROJEKT RED captured the beauty and chaos of real-world nature with RealityScan to create the real-time fantasy forest in The Witcher 4 Unreal Engine 5 Tech Demo.
At this year’s State of Unreal presentation, CD PROJEKT RED gave the world a glimpse into the future with The Witcher 4 Unreal Engine 5 Tech Demo. But it wasn’t just the return of a beloved universe that got attention—it was the world itself.
The rocks, trees, roots, and natural formations shown in The Witcher 4 Unreal Engine 5 Tech Demo looked incredibly realistic. That’s because, in many cases, they were actually real—captured with RealityScan, Epic’s next-generation photogrammetry tool.
The demo follows witcher Ciri, a professional monster slayer, as she explores the never-before-seen region of Kovir in the world of The Witcher 4.
CD PROJEKT RED used RealityScan to scan complex natural objects with incredible detail, streamlining the capture pipeline and bringing more realism than ever to the Continent.
The rocks, trees, roots, and natural formations shown in The Witcher 4 Unreal Engine 5 Tech Demo looked incredibly realistic. That’s because, in many cases, they were actually real—captured with RealityScan, Epic’s next-generation photogrammetry tool.
How RealityScan captured a forest
The demo follows witcher Ciri, a professional monster slayer, as she explores the never-before-seen region of Kovir in the world of The Witcher 4.
CD PROJEKT RED used RealityScan to scan complex natural objects with incredible detail, streamlining the capture pipeline and bringing more realism than ever to the Continent.
Duplicating the chaos of nature
Creating believable natural environments is one of the hardest challenges in game development. Nature is messy; rocks have undercuts, tree roots tangle and twist, and so on.
This chaos can cause even experienced scanning teams to struggle with certain challenges:
- Incomplete image alignments for large or irregular datasets
- Tedious manual masking to isolate subjects
- Quality control bottlenecks across massive asset libraries
Why CD PROJEKT RED chose RealityScan 2.0
CD PROJEKT RED adopted the new RealityScan 2.0 pipeline to scan high-fidelity assets from real-world locations, primarily rocks, trees, and forest floor elements. These objects weren’t scanned in controlled lab setups—they came from real forests complete with uneven lighting, occlusions, and background clutter.
RealityScan 2.0 workflows and features made all the difference in a number of ways:
AI Masking eliminated cleanup bottlenecks. Scanning a root system or mossy stone usually means hours of post-processing to clean up the background. With the new AI-powered background segmentation, CD PROJEKT RED was able to separate complex assets from the environment quickly and cleanly—directly inside the app. This was especially useful in fieldwork scenarios, where it’s difficult to control the background.
AI Masking eliminated cleanup bottlenecks. Scanning a root system or mossy stone usually means hours of post-processing to clean up the background. With the new AI-powered background segmentation, CD PROJEKT RED was able to separate complex assets from the environment quickly and cleanly—directly inside the app. This was especially useful in fieldwork scenarios, where it’s difficult to control the background.
Smarter alignment handled even messy organic data. RealityScan 2.0 introduced a high-quality feature detection mode. CD PROJEKT RED used this mode as the default, leading to higher alignment precision with fewer components (more unified scans), and better camera coverage across dense datasets. For the kind of wild, asymmetrical objects the team was capturing, this level of alignment stability made all the difference.
Visual quality analysis identified gaps. Large scan datasets can hide problems like missed areas or inconsistent coverage. The new Quality Analysis tools gave CD PROJEKT RED an instant overview of where the point clouds or meshes were solid (green), and where they needed more input (red). This meant fewer rescans, faster QA, and more confidence before assets entered Unreal Engine.
Once processed, the high-fidelity assets were brought into Unreal Engine 5, forming the backbone of the richly detailed world shown in The Witcher 4 Unreal Engine 5 Tech Demo.
RealityScan’s output worked seamlessly with the engine, preserving mesh integrity and enabling CD PROJEKT RED’s artists to build layered, immersive environments without needing to touch up assets in external DCC tools.
And because the masking, quality analysis, and alignment all happened within one tool, the studio’s asset pipeline moved faster, with less back-and-forth and more visual consistency.
What CD PROJEKT RED showed with The Witcher 4 Unreal Engine 5 Tech Demo is that entire environments—not just hero objects or props—can be grounded in real-world data.
RealityScan 2.0 provided the tech to make that possible:
Want to see the results? Check out the State of Unreal 2025 presentation, opened by CD PROJEKT RED and The Witcher 4 Unreal Engine 5 Tech Demo, to see it in action.
You can also learn more at the RealityScan website, and install RealityScan or RealityScan Mobile with the links below.
From scan to scene: Unreal Engine integration
Once processed, the high-fidelity assets were brought into Unreal Engine 5, forming the backbone of the richly detailed world shown in The Witcher 4 Unreal Engine 5 Tech Demo.
RealityScan’s output worked seamlessly with the engine, preserving mesh integrity and enabling CD PROJEKT RED’s artists to build layered, immersive environments without needing to touch up assets in external DCC tools.
And because the masking, quality analysis, and alignment all happened within one tool, the studio’s asset pipeline moved faster, with less back-and-forth and more visual consistency.
A new standard for worldbuilding
What CD PROJEKT RED showed with The Witcher 4 Unreal Engine 5 Tech Demo is that entire environments—not just hero objects or props—can be grounded in real-world data.
RealityScan 2.0 provided the tech to make that possible:
- AI masking reduced manual work.
- High-quality alignment handled complex objects.
- Visual QA ensured reliability at scale.
Want to see the results? Check out the State of Unreal 2025 presentation, opened by CD PROJEKT RED and The Witcher 4 Unreal Engine 5 Tech Demo, to see it in action.
You can also learn more at the RealityScan website, and install RealityScan or RealityScan Mobile with the links below.
Download RealityScan
Ready to scan the world?
RealityScan is free for students, educators, hobbyists, and companies earning under $1 million USD in annual gross revenue. Above that threshold? Find out about pricing or get a free 30-day trial today.