Research roadmap

3DGS · Neural rendering · Vulkan

A structured path from 3D Gaussian Splatting foundations to original contributions in surgical visualization. Four phases — each building toward NVIDIA-relevant skills and novel research territory.

Phase 1 — now

3D Gaussian Splatting foundations

Understand 3DGS from first principles, get first scene rendering on DGX Spark, integrate into VertexNova.

Gaussian ellipsoids, tile-based rasterization, alpha blending, spherical harmonics. How it maps to existing OIT knowledge from surgical robotics visualization work.

Short technical overview of gsplat, render pipeline stages, and practical API usage on DGX Spark.

gsplat 1.5.3 from source on GB10 Blackwell. SM 12.0 arch, --no-build-isolation, PyTorch 2.7 compatibility.

04 First 3DGS scene rendering active

Download pre-trained garden scene. View in nerfstudio. Understand what the .ply file contains and how the rasterizer works.

05 3DGS viewer in VertexNova planned

Create vne3dgs module. Parse .ply Gaussian attributes. First OpenGL render pass — blobs on screen before beauty.

Phase 2 — next

Neural rendering depth

Understand the full neural rendering landscape — NeRF vs 3DGS, 4DGS for dynamic scenes, training from own images.

01 3DGS vs NeRF — rendering engineer's comparison planned

Speed, quality, editability, storage tradeoffs. When to use each. From a real-time rendering background, not an ML background.

02 COLMAP pipeline — first own scene planned

Capture photos → COLMAP → sparse point cloud → 3DGS training. First surgical-style scene reconstruction.

03 nerfstudio full training pipeline planned

vertexnova/nerfstudio-spark container. Train splatfacto on custom data. Measure training time on GB10 vs reported benchmarks.

04 4D Gaussian Splatting — dynamic scenes planned

4D-GS achieves 82 FPS at 800x800 for temporal sequences. Apply to surgical anatomy — moving instruments, tissue deformation.

Phase 3 — NVIDIA signal

Vulkan + ray tracing + NVIDIA stack

Close the key gaps identified for NVIDIA Omniverse and neural graphics roles. Public demos that signal readiness.

01 Vulkan backend in VertexNova planned

Ship the Vulkan renderer. Profile with NVIDIA Nsight Graphics. Write a technical blog post. This is the single biggest NVIDIA signal.

02 Vulkan ray tracing first pass planned

VK_KHR_ray_tracing_pipeline. Simple path tracer. Shows understanding of RTX hardware — core to NVIDIA's stack.

03 CUDA compute in VertexNova planned

Parallel mesh processing pipeline using CUDA. Connects C++ rendering background with GPU compute fluency.

04 OpenUSD + 3DGS integration planned

OpenUSD added official GS support in April 2026. Import a USD scene into VertexNova or write a USD-to-splat pipeline. Direct Omniverse bridge.

Phase 4 — original contribution

Medical 3DGS — novel territory

Apply 3DGS to surgical visualization — a domain nobody has seriously explored. Potential paper or patent.

01 Surgical scene reconstruction with 3DGS planned

Endoscopic footage or CT multi-view → COLMAP → 3DGS training → real-time Vulkan render. Unique combination of OR rendering experience + 3DGS.

02 2DGS / Gaussian Frosting for mesh extraction planned

Extract meshes from trained Gaussian scenes. Bridge between traditional mesh-based surgical planning and neural rendering.

03 Research writeup / paper draft planned

Document findings. Target a venue like MICCAI or IEEE VR. Patent potential given surgical visualization background and existing co-inventor credits.

Target

NVIDIA team alignment

Which NVIDIA teams each phase maps to.

Phase 1–2 Neural Graphics SDKs

3DGS + gsplat + DGX Spark work maps directly to NVIDIA's Neural Graphics team hiring.

Phase 2–3 Omniverse / Isaac Sim

nerfstudio + OpenUSD + simulation background from robotic surgery platforms.

Phase 3 RTX / Graphics Research

Vulkan ray tracing + VertexNova multi-backend engine + CUDA compute pipeline.

Phase 4 Healthcare / Robotics

Medical 3DGS is novel territory. Surgical visualization + neural rendering is a unique combination nobody else has.