RenderFormer: Transformer-based Neural Rendering of Triangle Meshes with Global Illumination

TL;DR

RenderFormer introduces a transformer-based neural rendering pipeline that directly generates images from triangle meshes with global illumination, eliminating the need for per-scene training and enabling flexible, high-quality rendering.

Contribution

The paper presents a novel transformer-based rendering pipeline that models both view-independent and view-dependent light transport directly from triangle representations.

Findings

Effective rendering of scenes with complex shapes and lighting.

No per-scene training or fine-tuning required.

High-quality global illumination effects achieved.

Abstract

We present RenderFormer, a neural rendering pipeline that directly renders an image from a triangle-based representation of a scene with full global illumination effects and that does not require per-scene training or fine-tuning. Instead of taking a physics-centric approach to rendering, we formulate rendering as a sequence-to-sequence transformation where a sequence of tokens representing triangles with reflectance properties is converted to a sequence of output tokens representing small patches of pixels. RenderFormer follows a two stage pipeline: a view-independent stage that models triangle-to-triangle light transport, and a view-dependent stage that transforms a token representing a bundle of rays to the corresponding pixel values guided by the triangle-sequence from the view-independent stage. Both stages are based on the transformer architecture and are learned with minimal prior constraints. We demonstrate and evaluate RenderFormer on scenes with varying complexity in shape and light transport.