Does the DLSS setting affect still renderings?

Please note: Should you experience issues with Enscape or your subscription, and in case of any urgent inquiries/questions (e.g. regarding our upcoming licensing changes) please reach out to our dedicated support team via the Help Center or Support button as detailed here.
  • Further tests on a newly acquired Meta Quest 3 (with ultra high graphics on a high-end machine) made me realize that DLSS severely degrades real-time viewing quality in Enscape scenes.

    Only later did I find out a comment, that it is recommended by the Enscape team to turn DLSS off if your intention is to view your scenes in VR.


    DLSS, to be very blunt and simple, is an AI "interpretation" feature by NVIDIA, that analyzes your scene and adds in its own frames in between the frames produces by the software (in this case Enscape), making your experience smoother. Now a question arises which I couldn't find an answer for anywhere - does DLSS affect still renderings in Enscape?


    As I understand, DLSS does affect video renderings, correct?
    But does it make any difference if you are rendering just a picture?


    Here's 2 renderings of the same scene with same settings, rendered with and without DLSS.



    As you can see, visually it seems that there is no discernable difference.

    However I have to add, that rendering a still 4K picture with DLSS ON takes a bit less time, so I am assuming the GPU is doing its AI shenanigans in the background, "interpreting" and adding frames and stuff.

    But then again, there is no added value in the quality of the rendering.


    Here's the full pictures side by side.



    DLSS is making rendering times a bit faster apparently, that I understand. What I don't understand is at what cost? In a still rendering I saw no difference. But I read that DLSS does affect animation exports (by interpreting/adding frames). As good as AI can get, as a professional I don't trust AI generated imagery because its interpretation almost always skews straight lines. AI has trouble in the art world with fingers, and in the architecture world it does not comprehend how important parallel/straight lines are.


    Can someone confirm whether DLSS is compensating visual/design quality for faster return times?

    And does the checkmark change anything if I am rendering a still image?


    Thank you!

    • Official Post

    Hey ViggoPaulman


    you might be interested in our blog post & webinar on the topic:
    https://blog.enscape3d.com/enscape-nvidia-dlss-support

    External Content youtu.be
    Content embedded from external sources will not be displayed without your consent.
    Through the activation of external content, you agree that personal data may be transferred to third party platforms. We have provided more information on this in our privacy policy.

    DLSS, to be very blunt and simple, is an AI "interpretation" feature by NVIDIA, that analyzes your scene and adds in its own frames in between the frames produces by the software [...]


    That's not entirely accurate for the current Enscape (DLSS 2 based) integration, but a specific feature of DLSS 3 called "frame generation", which we haven't adopted yet.
    DLSS in Enscape is only upscaling the contents of the frame at hand, but of course since the GPU can render in a lower resolution you can already get a notable speed-up thanks to that.

    DLSS will be active for walk-through and video rendering. Still images do get up-scaled if DLSS is enabled, but this up-scaling doesn't need the neural network as it's much simpler and can be solved without it. In theory there can be minor visual differences (in screen-space effects) due to the up-scaling, but usually still images are almost identical to the native resolution results, as you've noted.

    For video renderings your results may vary: 2 years ago when we first shipped it, DLSS video quality was usually even superior to native resolution rendering, due to our temporal anti-aliasing implementation, which couldn't compete when it comes to sharpness/ghosting for fast camera movement.
    However in the mean-time we've replaced that temporal anti-aliasing implementation with AMD's FSR2 algorithm, which delivers much better quality.


    Bottom line:
    If you want best possible quality in your videos and don't have any issues with limited VRAM or long rendering times, deactivating DLSS makes sense.
    However if you're doing e.g. 4k or even higher resolution videos and you risk running out of VRAM and rendering times become a bottleneck in your process DLSS usually doesn't degrade quality notably even compared to the native resolution FSR2 temporal anti-aliasing we have today.


    And generally if you need fluent frame-rates for smooth navigation in your real-time walkthroughs it makes a lot of sense to have it enabled, especially on high-resolution (4k) displays.

    Hope that helps!

  • you might be interested in our blog post & webinar on the topic:
    https://blog.enscape3d.com/enscape-nvidia-dlss-support

    External Content youtu.be
    Content embedded from external sources will not be displayed without your consent.
    Through the activation of external content, you agree that personal data may be transferred to third party platforms. We have provided more information on this in our privacy policy.

    Really interesting talk. You guys should do these types of technical talks more often :)