About the Book
Build photoreal game levels from real scenes and ship them in Unity and Unreal with workflows that hold up under production pressure.
Turning scans into playable content is hard. Camera poses drift, floaters creep into windows, and frame time slips when scenes scale. Teams need a reliable path from capture to engine that keeps quality high and performance predictable.
This book gives you a practical pipeline. You will capture consistently, train fast with Gaussian splats and NeRF variants, export engine ready assets, and integrate them cleanly in URP, HDRP, and UE5 with Lumen, VR, and WebGPU delivery covered.
Choose the right representation and SH degree for speed, memory, and fidelity
Run stable capture, pose recovery, calibration, and undistortion for clean datasets
Train with Splatfacto and Splatfacto W, add depth regularization, and avoid floaters
Use gsplat and accelerated rasterizers, know when they help and why
Set visibility, depth handling, anti aliasing, and mip strategies that stay stable
Measure quality with PSNR SSIM LPIPS and camera path benchmarks that catch regressions
Convert PLY to compact SPZ with SH preservation, orientation, and tiling for streaming
Stand up batch exporters and converters that do not break on large jobs
Stream city scale scenes with tile grids, CDN friendly manifests, and clear budgets
Integrate in Unity URP and HDRP, master transparency sorting, queues, and depth prepass
Integrate in UE5 with translucent materials, Lumen lighting, and stable sorting
Build mesh proxies with SuGaR for collisions, nav, and occlusion wherever needed
Mix splats with Nanite meshes and classic assets without visual seams
Plan VR for Quest with Vulkan first, stereo correctness, foveation, and near plane control
Deliver on web with WebGPU and web viewers, manage file size and bandwidth
Profile bottlenecks, thread wisely, and scale out with nDisplay and multi machine setups
Apply pruning, scale regularization, and culling policies that reduce shimmer
Automate metric gates, screenshots, and video for repeatable regression checks
Follow release checklists, change guards, and production sign off steps that prevent surprises
Work through playbook case studies, from a one day indoor room to a streamed city block
Extras included, grounded in the manuscript, to speed real work: production playbooks, a release checklist and change guards, camera path benchmarks, and QA procedures with regression tracks.
This is a code heavy guide. Working snippets in Python, Shell, JSON, YAML, C++, and GLSL demonstrate training drivers, exporters, SH evaluation, rasterization loops, streaming managers, and profiling harnesses so you can plug them into real projects.
Get the workflow your team can trust, purchase your copy today.
