Eradiate Spring News

We hope you are all doing fine in these troubled times. Belgium was locked down for two months but the Eradiate team members were among the lucky ones who can continue working from home: the development did not suffer any setback. And we have some news we would like to share with you! In this newsletter, we talk about synergies with the computer graphics community and we give an update about the development status of Eradiate.

Boosting our synergy with the graphics community #

As we’ve been mentioning often, Eradiate has, from the start, been designed with the intent of benefiting from many technological advances made by the computer graphics community. We are obviously not the first ones to think about it: after all, Eradiate’s ancestor Raytran started back in 1993 as a fork of the Rayshade ray tracer. We also had very current examples at the Eradiate Workshop 2019 of radiative transfer scientists using, for their applications, tools and techniques developed in the computer graphics community (see presentations by Dr Najda Villefranque and Dr Sylvain Douté here and here). It is no coincidence that the Mitsuba renderer (the one underlying Dr Douté’s Hypsim simulator) has been one of our main sources of architectural inspiration from the start: its comprehensive design fits the needs of ray tracing for Earth observation applications and it is packed with features which make it ideal for contributed development.

It is fortunate that Mitsuba’s creator Pr Wenzel Jakob is based relatively close to the Eradiate team’s location; therefore we contacted him to discuss potential interactions between our communities and he was kind enough so as to give a talk at the Eradiate Workshop 2019 (presentation here).

A few months later, Pr Jakob’s team released a rewrite of Mitsuba. With similar abstractions to its predecessor, Mitsuba 2 leverages modern C++ to provide many new features, a lot of which are relevant to Eradiate. After pondering our options, we finally decided to embed a customised version of Mitsuba 2 as Eradiate’s computational kernel instead of reinventing the wheel. First, because Mitsuba 2 does not contain features contradictory with the goals we have for Eradiate. Second, because the Mitsuba 2 codebase is of excellent quality, well-documented and highly comprehensive. Finally, because Mitsuba 2 is under active development with a large community of contributors, which provides many opportunities to exchange code and scientific and technical knowledge. We believe that committing to such cross-community collaboration will help us deliver higher-quality software with the resources we have.

Towards the alpha version #

We expect to deliver a first set of features this Summer, following the established roadmap. This alpha-stage release will be delivered to ESA and will focus on monochromatic simulation and homogeneous participating media. Public release will come later during the beta phase. We are working hard on adding the functionality we are missing to our new computational kernel, but also on the machinery required to generate our scenes, as well as documentation and tests.

We leave you with an image generated using the very first contribution made by the Eradiate team to the Mitsuba 2 codebase. See you soon for more news!