About quality render, sampling and render times

It might be an idea to give quality render setup.

At some high point increasing quality means exponential render time, but it would be good for users to choose themselves between, for example, a 90% quality and 1 hour rendering and a 100% quality and a 10 hours rendering.

Fast rendering is great, but as we can always run a render overnight, if it allows few gains in quality I’m sure most of users will go for it for final work.

Also it seems the video rendering in D5 uses a lot of denoising process, which at the end decreases sharpness.

Best regards.

1 Like

I agree in having choices but this is a realtime render engine after all. If ultimate quality is what you are after, then I am afraid no realtime rendering engine will give you that, except Unreal, but then even Unreal will stop been realtime if you have bake the light solution and wait hours for it. Realtime is made of compromises, fewer bounces, acceptable but not best available quality.
I can render 8-10 sec 4k supersharp, noise free complex video segments regularly in 10-12 minutes using a single RTX3090. No other GI raytracer can do that as far as I know and that is “the feature” that makes D5render unique and so valuable for some of us.

1 Like

I’m talking about the still existing gap quality in between still images and video render in D5, even rendered in 4K.
Quality export setting are current option like in Lumion or Twinmotion.
The border Between realtime engine and path trayced is every day narrowing ; with you specs you can can probably run cycles engine almost in realtime too.

So it’s great to render a video in 10 min with a 3000$ graphic card but that’s not the point.

Realtime is still an abuse of language and most of Professional inquiries are not presented realtime in D5 (or whatever software) itself but in controlled products like videos or still images because architects are very carefull and precise to what they show to clients.

There’s no debate in D5 already great value quality level; I’ve been among the first lucky users and what have been achieved in 1 year is just mind blowing and benefit to all of us.

D5 stand for comunity version and not expensive like Lumion seing architects as milking cows is also a very valuable philosofy and at the end probably its hardest performance to date.

1 Like

Well if the developers can do what you are asking then we all win. I use Vray professionally for interior renderings for 20 years and I know no current realtime render engine will match it regarding quality therefore I may not need the different quality settings but sure it can be better if we have the option. It is probably more important to people using it for both stills and animations.
Regarding the card :grinning:, it was €1588 when I got it. Just now the prices have doubled with the massive shortage everywhere. Cheers.

1 Like

I think that the quality gap between UE and VRAY has come to a point where it is near. Is there a difference, yeah. Is it huge/big, for me personaly NO.

That being said since D5 is UE based and evolving at a HUGE pace i think we are going places… If what the D5 team has done in one year keeps going at the same speed i can only dream where the software can be next year… I think they also have the possibility to update their engine to the upcoming UE5.

I understand the question for a quality option altough personaly i would prefer the ability to render a part of the frame (as in vray, corona,…). It’s kind of intensive to constantly re-render the whole image when you are working on a specific part.

Enjoy your weekend boys and keep those discussions alive, i know the D5 team is reading and always looking on how they can improve the software .


I agree that final output quality selection would be great, ie a slider from low to high, high being fully path traced, or much higher light bounces/sampling than current setup.

The curent biase is a good speed vs quality balance, but for a final image I’d be happy to wait longer to achieve proper lighting/shadows.

Twinmotion has just added controllable pathtracing, and looks like Enscape will with the addition of Corona.


interesting them for the future. The functional ideas are also aimed at individual working methods.
It will be difficult, if it is expected to have the same or similar features.

I compare the wishes of the client in an architecture competition, many specific specifications, where and how, also limit our creativity.
Sometimes it’s in the details where the wishes are.

However, this huge pace and advancement I haven’t experienced with the other render programs in our office. For the future I wish the same clarity in the development of functions as 100 different trees, but I know the D5 team is doing everything right.


I recommend everyone to read our blog to learn more about path tracing and D5 GI.

We do more work than path tracing. :eyes:


I read all the blogs and was amazed :). Also the vision about the desired progress and future possible paths are very interesting!


Watched this blog.

I can be wrong, but what I see in fact in D5 Render about GI is a combination of two technologies, NRD + RTXGI.

NRD - NVIDIA Real-Time Denoiser | NVIDIA Developer
RTXGI - RTX Global Illumination | NVIDIA Developer
And for some reason, the NRD function is used by half or not used at all.

Pic 1 - In D5 I see Raw Ray Traycing, but I don’t see NRD being used.
Pic 2 - Render in D5 Render with only Raw Ray Traycing (RTXGI is OFF).
Pic 3 - Render in D5 Render (RTXGI is ON).

As you can see in Pic 2 and Pic 3 near the bus stop (framed in red), the denoser applied in D5 Render does not currently remove noise from Raw Ray Traycing or from translucency.
Maybe NRD technology would do it.

Please, look at the pictures I sent in the original scale.
Thank You!

Pic 1

Pic 2

Pic 3

Noticed that GI settings for video are better (clean) than for static rendering.
It would be nice if this could be improved in new versions of D5 Render.

Video screenshot

Static render