At SIGGRAPH 2018 OTOY and Epic Games demonstrated a preview of Octane 2019 integrated with Unreal Engine 4. Features shown in the preview included:
- The automatic conversion of Unreal Engine scenes and materials into OctaneRender and full integration of OctaneRender feature that will be included in Octane 2019.1.
- Use of the Brigade Engine scene graph for real-time path-traced games and interactive content powered by Unreal Engine.
- AI Light, AI Scene, AI Spectral and Volumetric Denoising, and Out-of-Core Geometry, UDIM Support and Light Linking for final rendering.
- Octane Vectron and Spectron – Octanes recently announced procedural node-based Volumetric Geometry and Lighting technology that promises “infinite detail” and granular lighting control (features were added in OctaneRender 2018)
- The ability to use OTOY’s ORBX scene format, enabling users to drag and drop scenes created in many leading authoring tools and mix and match them with Unreal Engine content in a responsive Live Viewport.
- Optimized support for NVIDIA RTX Ray Tracing, Vulkan, DXR, CUDA and Metal iOS/Mac OS backends built on OTOY’s cross-platform RNDR framework. AI Viewport: mixed reality media, including upscaling, AI style filters, light field and holographic output for next-generation volumetric displays and HMDs.
Integration is planned for release in the first half of 2019. It will be included in OctaneRender’s new rental scheme with a baseline offer for up to 20 GPUs, network rendering and more than 20 DCC integrations for $20 a month. A private beta will be available in August with a public beta available later this year. You can read the full press release on OTOY’s website.
AI, AI, AI… looks more marketing talk than real AI. Those are just machine optimized algorithms i guess? Or does those AI denoising algorithms really learn from mistakes and improve? Can user teach them or are those already optimized by developer -Otoy have done all teaching?
Actually, it’s quite a reasonable use of the term AI based on modern standards. Yes, it’s mostly supervised learning stuff, which is very much part of modern AI. You might want to check out the recent SIGGRAPH primer on deep learning – https://cgpress.org/archives/cgtutorials/deep-learning-a-crash-course
Yes it works – as octane 4 users can confirm!
What are some good use cases for Octane in UE4?
Real time visualization through baking or real time path tracing with denoising