Ohao Engine — real-time path tracing · ~67 fps · DLSS Ray Reconstruction · ReSTIR GI

Shaoxuan ·
Frank · Yin

Vulkan · CUDA · Neural Rendering · Agents · Simulation

I build renderers, physics engines, and the pipelines that feed them — from hybrid deferred-plus-ray-traced Vulkan to CUDA path tracers with MIS and OIDN denoising, down to the capture and reconstruction infrastructure that turns photons into assets.

MSc Computer Science, Linköping University — Graphics & Visualization. Ships hardware RT pipelines (Vulkan KHR, CUDA), real-time path tracing denoised with NVIDIA DLSS 4 Ray Reconstruction and SVGF, MIS path tracing with environment-map importance sampling and Owen-scrambled Sobol, three offline denoiser backends (Intel OIDN / NVIDIA OptiX / NRD), differentiable rendering (nvdiffrast + Disney BRDF), and synchronized multi-camera capture systems. Recently completed an AI-algorithm internship at Freedo Technology (MLS-MPM physics on 3D Gaussian Splatting). Also builds in the open: an inverse rendering lab inside Ohao that recovers PBR materials and lighting from multi-view captures, plus agent tooling — an open-source Godot MCP that lets AI agents build and playtest scenes, and a multi-agent CLI framework. Open to graphics / engine / GPU-systems / AI-infrastructure / agent roles — available full-time.

61K Lines · Vulkan Engine Standalone C++ (50K+) + GLSL (11K) renderer — deferred PBR path and hardware ray-tracing path sharing one scene representation.

11 dB PSNR · Thesis Result 3D Gaussian Splatting gain over NeRF on reflective surfaces — measured on a 209 GB dataset I captured and released.

12 cams Capture Pipeline Synchronized multi-camera rig with staggered triggers + Bluetooth turntable — built end-to-end in C++ / C# (~8K lines).

40 FPS CUDA Path Tracer Monte Carlo with BVH on RTX 4060 at 1280×720 — Russian roulette termination, OIDN denoising, consumer-GPU throughput.

… Contributions · 2026 Across github.com/Qervas — shipping, not posting.

GitHub contribution heatmap for @Qervas — 4,858 contributions in 2026

Featured Work

Damaged Helmet PBR render — close-up with rich metal, glass, and cable detail

Damaged Helmet · deferred PBR · full material stack

Lantern scene rendered in Ohao — emissive warm glow with wooden cross structure

Lantern · emissive driving scene lighting

MetalRoughSpheres · BRDF calibration

Path-tracing pipeline · Intel OIDN denoising on a glossy scene

Path-traced helmet at 16 spp before denoising — visible noise across glossy materials

Helmet · 16 spp · raw path-traced

Same helmet at 16 spp after Intel OIDN denoising — production-quality output

Helmet · 16 spp · OIDN denoised

Ohao Engine

C++20 · Vulkan KHR RT · GLSL · ~50K C++ + 15K GLSL across 121 shaders

Standalone Vulkan engine written from scratch. Hybrid renderer: a deferred G-buffer path and a hardware ray-tracing path share one scene representation, bindless resources, and acceleration-structure workflow. Architecture spans render (22K LOC), physics (11K LOC), and a gpu / Vulkan abstraction layer (8K LOC).

Deferred path: Cook-Torrance GGX BRDF, cascaded shadow maps, SSAO, screen-space reflections, subsurface scattering, TAA, bloom. Path-tracing path: Vulkan KHR ray-tracing with multiple-importance sampling (balance + power heuristic), environment-map importance sampling via precomputed marginal/conditional CDFs, Owen-scrambled Sobol QMC, variance-driven adaptive sampling, and three offline denoiser backends (Intel OIDN, NVIDIA OptiX, NVIDIA NRD).

Real-time path tracing (new, July 2026): the same path tracer now runs interactively with a denoiser chain built for 1-spp frames — NVIDIA DLSS 4 Ray Reconstruction integrated through the raw NGX Vulkan API on Linux (guide buffers, motion vectors, HDR pipeline; feasibility-probed on Blackwell), full SVGF with temporal reprojection through camera motion and variance-guided à-trous filtering as the vendor-neutral fallback, environment importance sampling + MIS in the realtime raygen, solid-angle sphere-light sampling, and an adaptive N-spp loop that spends spare GPU headroom on quality.

Physics layer: GJK / EPA collision detection over a BVH. Top renders use Khronos glTF Sample Assets — a sanity-check that the engine reproduces the reference results graphics engineers actually verify against.

Inverse rendering lab (new): recovers multi-surface PBR materials and lighting from multi-view captures — train-only fit with a Diff-IR deferred sibling and a hybrid Diff-fit → path-trace refine. Capture-gated on held-out views: ~32.5 dB holdout, ~34.4 dB relight (+20.5 dB vs a wrong initialization).

GitHub Repo ↗

3DGS · +11 dB PSNR vs NeRF on reflective surfaces

Aerial photogrammetric reconstruction of Gränsö Castle

Gränsö Castle · aerial photogrammetric reconstruction

Method comparison matrix: Input / RealityCapture / NeRF / 3DGS across studio objects including vase glass, stone, skull

Method matrix · RC / NeRF / 3DGS · glass, stone, skull

Neural Rendering Dataset — MSc Thesis

C++ · C# · WinUI · Hardware Control · ~8K lines

Built CamMatrixCapture: a 12-camera synchronized capture system with staggered trigger firmware, Bluetooth-controlled turntable automation, and a state-machine driven capture workflow. Indoor object captures feed NeRF / 3DGS training; outdoor work includes drone reconstruction of Gränsö Castle.

Released a 209 GB public dataset focused on challenging materials — reflective, transparent, translucent surfaces that traditionally break neural rendering methods.

The benchmarking contribution: 3D Gaussian Splatting reaches ~11 dB PSNR above NeRF on reflective surfaces — a surprisingly large gap, with clear implications for production capture pipelines.

Gränsö Reconstruction (video) ↗

Selected Projects

NIUA Godot MCP architecture — AI agent to MCP server to Godot editor to playtest evidence

NIUA Godot MCP · open source

Node / TypeScript · MCP · Godot 4.6 GDExtension

From-scratch MCP server plus a bundled Godot editor addon that lets AI agents create, inspect, run, debug, and export Godot 4.6 games through tools — ~26 tool domains (scene, nodes, scripts, run, resources, UI, workflows…), an editor bridge over local HTTP with token auth, and a run_playtest_evidence contract (input → assert → screenshot) so the agent checks its own work. Installers for Claude / Cursor / Codex.

Repo ↗

NocturneAI architecture — coordinator, ReAct agents, tools, workflow engine

NocturneAI · multi-agent CLI

TypeScript · React Ink · ReAct · Workflow Engine · in dev

Autonomous multi-agent framework for the terminal. Clean layered architecture (core / application / infrastructure / presentation), an AgentCoordinator for multi-agent orchestration, a workflow engine with plan validation, plus plugin manager, metrics, and background-task manager. ReAct loops with tool registration / execution; GitHub Copilot's OpenAI-compatible API as the LLM backend. ~30K LOC.

Repo ↗

MLS-MPM Physics Simulation

Taichi · MLS-MPM · APIC · Fluid + Solid Mechanics

Real-time MLS-MPM physics (Hu et al. 2018) in Taichi — fluid dynamics coupled with solid mechanics and fracture on one particle-grid solver. Scenes: 2D dam break, 3D water tearing through breakable bridge pillars, and building / flood destruction. APIC particle↔grid transfer with weakly-compressible equation-of-state pressure.

Live demos ↗ Code ↗

sx-engine hero render: ray-traced glass sphere over desert with cloud-reflecting environment

sx-engine — CUDA Path Tracer + Physics

CUDA · Taichi · C++ · OIDN · ~3K LOC

CUDA path tracer with BVH, Schlick-Fresnel refraction, and optional Intel OIDN denoising. 16 parameterized liquid presets (ocean · honey · lava · milk · blood · crystal …) each with their own IOR and depth-absorption. Paired with a Taichi-GPU Position-Based Dynamics cloth solver — stretch + bend constraints, material-specific stiffness — exporting per-frame OBJ meshes back to the CUDA renderer. 58 production stills shipped.

Repo ↗

Glossy reflective pot from the thesis capture — the kind of object this inverse-rendering pipeline is designed to preprocess

Turntable Inverse Rendering · in progress

Python · PyTorch · nvdiffrast · GGX/Disney BRDF

Foundation for a differentiable rendering pipeline that will learn per-point BRDF from turntable captures, then relight glossy objects uniformly so Structure-from-Motion actually converges. Full GGX stack shipped — D (GGX / Beckmann / anisotropic), G (Smith / Schlick-GGX), F (Schlick / roughness-aware) — plus nvdiffrast backend, turntable camera / mesh / lighting layers. Neural decomposition + training loop next. Built on NVIDIA nvdiffrast.

Repo ↗

SPH particle fluid simulation with velocity-coded color visualization, confined in a tiled pool

SPH Water Simulation

C++ · CUDA · OpenGL · BSc thesis

GPU-accelerated Smoothed Particle Hydrodynamics solver with Marching Cubes surface extraction and CUDA ↔ OpenGL interop for real-time visualization. Uniform-grid spatial hashing for neighbor search, cubic-spline density kernel with gradient / Laplacian variants for pressure and viscosity forces, plus surface-tension kernel. Bachelor thesis — the foundation my current fluid work builds on.

Repo ↗ Demo video ↗

Fully oxidized metal sphere PBR render — final stage of the procedural rust progression

Real-time Material Aging & Weathering

OpenGL · GLSL Compute Shaders · PBR · Procedural Noise

Compute-shader playground for procedural material decay on a PBR sphere — noise-driven rust progression (0/25/50/75/100%), paint cracking and peel, and a dynamic puddle / weather system that wets and dries the ground in real time. Built around an OpenGL compute pipeline with full BRDF response so albedo and roughness shift together as the surface ages. Coursework for Linköping's TNM084 Procedural Methods.

Repo ↗

Custom Qt-based Xiàngqí board mid-game with the trained DQN agent on move

Xiàngqí AI with Deep Q-Networks

C++ · Qt · CUDA · DQN · Self-Play

Reinforcement-learning agent for Chinese Chess — full Xiàngqí rules engine in C++/Qt, a Deep Q-Network with experience replay and a target network, and CUDA-accelerated training driven by self-play in the spirit of AlphaZero. Win-rate sweeps against random and self-play opponents, plus a playable GUI to face the agent. TNM114 group project at Linköping with Oskar Tengvall.

Repo ↗

$Cornell-box-style path-traced render with diffuse, glossy and refractive interactions converged$

Path Tracing & Photon Mapping

C++ · CUDA · Monte Carlo · KD-Tree

Two global-illumination renderers written from scratch for TNCG15. Monte Carlo path tracer with a CUDA backend for per-pixel parallel ray generation and progressive accumulation — real-time-feasible against the CPU baseline. Photon-mapping pass scatters photons from light sources, then estimates radiance via KD-tree nearest-neighbour lookup for sharp caustics and indirect lighting. Comparative study of where each technique wins.

Repo ↗ Demo ↗

Skiing-action classifier test panel: original video, OpenPose skeleton, ST-GCN class label, and attention map

Skiing Action Analysis · OpenPose + ST-GCN

Python · OpenPose · ST-GCN · Action Recognition

Pipeline for classifying skiing techniques from video — OpenPose extracts 2D body keypoints, a Python preprocessor stitches per-frame skeletons into spatial-temporal graphs, and an ST-GCN trains on those graphs to label actions like big-bend carving. Best model (epoch 39, lr 0.1, batch 32) hits 50% Top-1 / 90.9% Top-5. Co-authored, published in Intelligent Computer and Applications Vol. 12 No. 4, April 2022.

Paper ↗

Experience

Freedo Technology

AI Algorithm Intern · Remote

Feb 2026 — Jun 2026

Built a GPU physics system coupling an MLS-MPM solver to 3D Gaussian Splatting on a multi-GPU A100 server — a static scanned scene can deform, fracture, and flow in real time. ~2.25× end-to-end speedup via CUDA-graph capture and mass-conserving particle reduction.
Wrote the simulation-and-render pipeline in Taichi, NVIDIA Warp, and CUDA; productized it as a FastAPI + React workbench with streaming splat playback.

Independent AI Creative Platform (NDA)

Founding Engineer · Part-time

2025 — Present

Generative-media platform for creative tooling — GPU-backed image / video generation jobs orchestrated through an async queue with retry, prioritization, and tenant-isolated quotas.
Built the full vertical: TypeScript frontend, Python / Node backend, Postgres data layer, object storage, OAuth, billing integration, and a deployment / monitoring story. Reads as full-stack — anchored in the GPU work that pays the bills.

Beijing Guoyao Xintiandi Information Technology

C++ Development Engineer Intern · Beijing

Aug 2022 — Sep 2022

Rewrote rendering pipeline from legacy OpenGL to modern OpenGL for a Qt-based 3D medical visualization tool — shader programs, VBOs, modern state management.
Built a custom C++ network protocol layer for data transmission between client and backend.

Skills

LanguagesC++20, CUDA, Python, C#, TypeScript, Rust, GLSL, HLSL

Graphics APIsVulkan (incl. KHR ray tracing), OpenGL 4.3, DirectX 12

RenderingDeferred rendering, hardware RT, path tracing, PBR (Cook-Torrance GGX), global illumination, TAA, SSAO, SSR, CSM

GPU ComputeCUDA kernels, compute shaders, CUDA-OpenGL interop, Nsight / RenderDoc / PIX profiling

Simulation & MathSPH fluids, Marching Cubes, GJK/EPA collision, BVH, differentiable rendering

Neural RenderingNeRF, 3D Gaussian Splatting, synchronized multi-camera capture pipelines

Engines & ToolsCustom C++/Vulkan, Godot (GDExtension), Unreal Engine, Blender

SystemsCMake, Git, Linux, Docker, PostgreSQL, low-level performance profiling

How I Work

I care about infrastructure, not app surface area. The systems I find interesting are the ones other engineers build on top of: a renderer, a capture pipeline, a physics engine, a GPU scheduler. I profile before optimizing, I read the paper before reimplementing, and I'd rather ship one well-understood 50K-line engine than three shallow demos.

Education

Linköping University

MSc Computer Science — Graphics & Visualization

2023 — 2025

Graduated February 2026.
Top grades (5/5) in Advanced Global Illumination, Modelling & Animation, Computer Games Design, Large Distributed Projects.
Thesis: Neural Rendering Dataset Collection (see Featured Work above).

Beijing Information Science and Technology University

BEng Computer Science

2018 — 2023

GPA 3.53 · Advanced Mathematics 97 · CPU Design 91.

Shaoxuan ·Frank · Yin

Ohao Engine

Neural Rendering Dataset — MSc Thesis

NIUA Godot MCP · open source

NocturneAI · multi-agent CLI

MLS-MPM Physics Simulation

sx-engine — CUDA Path Tracer + Physics

Turntable Inverse Rendering · in progress

SPH Water Simulation

Real-time Material Aging & Weathering

Xiàngqí AI with Deep Q-Networks

Path Tracing & Photon Mapping

Skiing Action Analysis · OpenPose + ST-GCN

Shaoxuan ·
Frank · Yin