Repository navigation
[Feature] Feature Request: Quad-View Foveated Rendering (QVFR) with NVIDIA MVR/SPS support for ETS2/ATS VR #32
Description
What should the mod do?
I would love to see support for eye-tracked Quad-View Foveated Rendering (QVFR) in ETS2/ATS, ideally combined with NVIDIA Multi-View Rendering (MVR/SPS) and the existing DLAA/DLSS implementation.
The goal would be to render a small, high-resolution focus region around the user's gaze while rendering the peripheral areas at a much lower resolution, reducing GPU workload without sacrificing visual clarity where it matters most.
Why? What would it do for you in the game?
First of all, thank you for your fantastic work on bringing DLAA/DLSS to ETS2 and ATS, especially for VR users!
I'm currently using the AMS2 QVFR mod with my Pimax Crystal Super 50 PPD and RTX 5090, and the results are absolutely stunning.
My current settings in AMS2 are:
- Render scale: 1.0
- FOV crop: 15% top / 5% bottom
- Focus resolution: 100%
- Peripheral resolution: approximately 30%
- Focus region: approximately 20%
- Eye tracking enabled, combined with DLAA
With this configuration, I can run almost everything on high settings while maintaining excellent image quality. Fine details are much clearer, especially in challenging weather conditions and at night, without the distracting aliasing I used to experience.
The performance improvement is remarkable, although CPU limitations mean that 72 Hz currently works best for my setup.
Considering the extremely high rendering resolution of modern VR headsets, I believe QVFR could be a major improvement for ETS2/ATS, particularly when combined with your existing DLAA/DLSS injection.
It would make VR truck simming much more enjoyable by allowing higher graphics settings and potentially better performance without compromising sharpness in the area you're actually looking at.
Another major advantage of QVFR would be the possibility of increasing the focus region's render resolution well beyond 100%, while keeping the peripheral resolution much lower.
This could be especially beneficial in ETS2/ATS, where being able to read distant road signs, recognize small objects and see fine details far ahead is important for immersion.
The game already offers internal scaling up to 400%, but applying such high resolutions to the entire VR image would be extremely GPU-intensive. With eye-tracked QVFR, we could potentially achieve much of the visual benefit by supersampling only the area we're actually looking at.
For example, using 150–200% focus resolution with approximately 30% peripheral resolution could be an interesting starting point, depending on performance and technical feasibility.
How often would you use it?
Every drive
Where would it live?
- A menu row (Picture / Motion and mirrors / Upscaling and VR / VR performance / Frame rate and CPU / Panel / FPS counter)
- A
dlaa.inikey only - A hotkey
- Don't know
Game
Both
Flat or VR
VR
Your setup (helps to judge the cost)
RTX 5090, Pimax Crystal Super 50PPD. OpenXR
Anything else
Here are two interesting projects that might serve as technical references:
- AMS2 QVFR + DLSS Mod
https://github.com/hypeapps/ams2-qvfr
This is the mod I'm currently using in Automobilista 2. It demonstrates how effective eye-tracked QVFR can be when combined with DLAA/DLSS.
- Sim-QVFR
https://github.com/Williem3/Sim-QVFR
Another interesting project exploring QVFR and NVIDIA Multi-View Rendering (MVR) across multiple simulators, including LMU, AC EVO, ACC and X-Plane 12.
Both projects might provide useful ideas or implementation references.
I understand that ETS2/ATS uses the Prism3D engine and that integrating QVFR could be considerably more complicated than the existing VRS implementation.
I'm not expecting an immediate solution, but I'd love to know whether you think something like this could be technically feasible in the future.
Thanks again for all your hard work and for making VR truck simming better!
Activity
MikeSta4 commentedon Nov 12, 2026
One more thought that came to mind after submitting this feature request:
Besides using QVFR to increase the focus region's render resolution, the GPU performance saved by rendering the peripheral areas at a much lower resolution could also be used for more demanding graphics and lighting improvements.
For example, enhanced lighting, shadows, reflections, or even future DLSS 5 Neural Rendering implementations.
Especially in ETS2/ATS, where realistic lighting and atmospheric effects make such a huge difference to immersion, this could be another major benefit of QVFR.
Of course, compatibility with additional graphics injectors would be a separate challenge, but the idea of combining eye-tracked QVFR with advanced rendering effects is really exciting.
JRDevo commentedon Nov 12, 2026
Thanks for the detailed request and the two links -- they helped a lot.
Short answer: I looked at it, and it is not something I can add soon.
Why:
What you can use today:
I keep the issue open as a long-term idea. If SCS adds multi-view support, or someone finds a clean way to get extra views out of Prism3D, I will look again.
JRDevo commentedon Nov 12, 2026
You can already get a share of this today. VR performance > foveated (VRS) with eye tracking cuts GPU work at the edges, and you can spend that on the game's own shadow, reflection and lighting settings.
DLSS 5 Neural Rendering needs DX12. ETS2 / ATS in VR run on DX11, so it is not possible in this mod
MikeSta4 commentedon Nov 12, 2026
Thanks for the detailed explanation!
Just one technical question out of curiosity:
As far as I understand, the AMS2 QVFR mod works by replacing
openvr_api.dll, while Dan Suzuki's LMU mod uses a launcher. Neither requires modifying the game's EXE file on disk.I assume they hook into or patch the rendering pipeline at runtime instead?
Would a similar DLL-based approach theoretically be possible with Prism3D, or does its rendering architecture make this fundamentally different?
I'm not suggesting this would be easy, just trying to understand the technical limitations a little better. Thanks again!
MikeSta4 commentedon Nov 12, 2026
Hi JRDevo,
Thanks again for your explanation about using the GPU headroom from eye-tracked VRS for higher in-game lighting and shadow settings.
Unfortunately, even at maximum settings, the default lighting in ETS2/ATS still looks rather dated to me, especially at night and in certain weather conditions.
I was wondering about something else:
Would it theoretically be possible to use your DLAA/DLSS injector and eye-tracked VRS alongside a separate DX11-based lighting enhancement plugin?
For example, if your mod uses
dinput8.dlland the lighting plugin usesdxgi.dll, and all upscaling and anti-aliasing features of the lighting plugin are disabled, leaving only its lighting improvements active.I understand that different DLL names don't necessarily prevent conflicts, especially when both plugins hook into the rendering pipeline.
Do you think such a combination could work, or is there a fundamental technical limitation? Perhaps you know of an alternative approach to improving the game's lighting while keeping your DLAA and eye-tracked VRS?
Just curious about the technical possibilities. Thanks again for your time!
JRDevo commentedon Nov 12, 2026
possible in theory with a DLL, and the LMU multi-view path is the right model. It is still a large project and I can't promise it. My next step would be a small test: can the mod draw ETS2's G-buffer twice in one call (eye + focus view) without breaking the picture. If that works, I will say more here.
JRDevo commentedon Nov 12, 2026
Yes, in theory that can work. Windows loads both files, because they have different names. This mod hooks the D3D11 calls themselves, so it still sees the game's drawing when another plugin sits in front of D3D11 (as ReShade does).
I have not tested such a combination myself, so please treat it as an experiment.
If you try it, please use the v0.16.0-rc4 pre-release and send dlaa_inject.log with log_level = 4 in dlaa.ini. That version's log lists the other DLLs and plugins that loaded, and it shows if DLAA / DLSS and the eye-tracked VRS still start. Then I can tell you whether they work together, and fix it if they don't.
rimaille commentedon Nov 13, 2026
Just a small comment to say you can already use dlss5, in combination with optiscaler. Didn't test it, my hardware is low end, but with your 5090 you can, there's a lot of videos on youtube explaining how to do this.
JRDevo commentedon Nov 13, 2026
Since v0.11.0 the mod knows about OptiScaler. It looks for OptiScaler next to the game exe, and the log then shows env: OptiScaler detected.
AFAIK NVIDIA has not released a DLSS 5 SDK or DLL yet. The DLSS 5 file these mods use leaked from a game build; it is not an official NVIDIA download.
It would really be far better when SCS releases DX12 and NVIDIA releases the SDK... but I'm happy to troubleshoot what I can on this mod's side to try to get it working.