We are now approaching the next Unreal Engine update, and not only that: Epic Games makes the engine ready for the next mayor version. UE5 was a big step in multiple ways: Not only has the technology jumped forward a good bit, but also the maturiy of certain features improved so drastically that they were partially not recognizable. The stability of the Vulkan rendering in particular has improvided a lot.
State of Synavis
One thing that I was never quite happy with regarding Synavis is the necessity of using a third party library, such as FFmpeg or GStreamer, with no immediate support, as part of the pipeline. Synavis, in the future, will natively use FFmpeg as part of its pipeline.
This has a few reasons. I think the point people might raise is why implement this when this will inevitably result in CPU encoding? - and it is a fair point. However, the CPU is relatively stale in these types of workloads and the reason why we are able to offload the encoding onto the CPU is because the other aspects of the pipeline, such as measurement and other aspects that are directly controlled by Synavis via JSON messages, are relatively cheap in comparison to compute power today.
The main problem with these larger scale pipelines right now, of course, is the fact that scientific compute resources have also become more expensive relative to how they used to be. Of course, scientific IT infrastructure still aims at providing researchers with immediate access to rendering capabilities and furthermore with access to compute resources. That being said, if you find yourself without a larger "organization" which alleviates some of these burdens, you might encounter issues. If you want to know how to get these workflows to run and whether there might be somewhere you could run the computations on, feel free to contact me using any link in the bar above.
Substrate Materials
In a nutshell, substrate is a new material system for Unreal Engine which enables the use of more complex visual mechanisms that allow the shading and rendering of multilayered materials. For Unreal Engine, this means that I will continue to adapt the SynavisUE Plugin in such a way that biologically realistic parameters can be mapped onto the standard texture in a more direct way than it was possible before. However, this is some work, as before I always directly exposed UE concepts to Synavis, without any helpful or biologically relevant conversion. This will be fixed in the next Synavis version.
Potentially we either use the sample subsurface scattering using the basic properties as parameters or we assign a new layer to the leaf innards. The main question here is parametrizability: A normal subsurface profile in UE is so far tied to a specific color and while I could make a seperate object from this, it would not really be worth it.
Some readings I am doing for this
- https://onlinelibrary.wiley.com/doi/10.1111/pce.13718
- https://link.springer.com/article/10.1134/1.1366754
- https://doi.org/10.1109/TGRS.2019.2890998
- https://doi.org/10.1016/j.dib.2024.110204
- Optics and Spectroscopy, Vol. 90, No. 4, 2001, pp. 580-591. Translated from Optika i Spektroskopiya, Vol. 90, No. 4, 2001, pp. 652-663. Original Russian Text Copyright © 2001 by Berdnik, Mukhamedyarov
- https://dl.acm.org/doi/10.1145/1073204.1073252
- https://dl.acm.org/doi/10.1145/1073204.1073251
- https://doi.org/10.1016/0034-4257(90)90100-Z
- https://www.tandfonline.com/doi/full/10.1080/15481603.2023.2168410