Announcement

Collapse
No announcement yet.

Understanding DMC Sampler

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by vlado View Post
    It would be interesting to see how V-Ray 3.0 handles this. Do you think you can maybe get me the scene to play with it?

    Best regards,
    Vlado
    I will see what I can do when I get a moment.
    Kind Regards,
    Richard Birket
    ----------------------------------->
    http://www.blinkimage.com

    ----------------------------------->

    Comment


    • I may be having a breakthrough moment here. Managed to drop down my light subdivs. The biggest culprit were vraymesh lights. Once I went back to my old-fashioned methods of using vraylight materials, those issues went away. I have also reduced the reflection depths of most/all materials to 1 or sometimes 2. Reflections have generally been sharpened up so I have been able to reduce the samples there too. A subtle timber bump texture on my floor, which added very little to the beauty pass, has also been ditched.
      Kind Regards,
      Richard Birket
      ----------------------------------->
      http://www.blinkimage.com

      ----------------------------------->

      Comment


      • Originally posted by tricky View Post
        I may be having a breakthrough moment here. Managed to drop down my light subdivs. The biggest culprit were vraymesh lights. Once I went back to my old-fashioned methods of using vraylight materials, those issues went away. I have also reduced the reflection depths of most/all materials to 1 or sometimes 2. Reflections have generally been sharpened up so I have been able to reduce the samples there too. A subtle timber bump texture on my floor, which added very little to the beauty pass, has also been ditched.
        What times are you getting now on that 800X800 render with materials active?
        Kind Regards,
        Morne

        Comment


        • I managed to get a full size (4000px) render done in around 1.5hrs using about 6x i7 machines. Its a simplified model with all the furniture removed. The elements look pretty clean - even the samplerate element looks ok. Still areas of red on edges and on areas of my timber floor, but not bad overall.
          Kind Regards,
          Richard Birket
          ----------------------------------->
          http://www.blinkimage.com

          ----------------------------------->

          Comment


          • Originally posted by tricky View Post
            I managed to get a full size (4000px) render done in around 1.5hrs using about 6x i7 machines. Its a simplified model with all the furniture removed. The elements look pretty clean - even the samplerate element looks ok. Still areas of red on edges and on areas of my timber floor, but not bad overall.
            Please post your times when you do the final image. Using this method and also doing 4K images with 7 X i7s I'm getting around 3 to 7 hours for interiors and 1 to 4 hours for exteriors. Most of them pretty complex. I like how clean things come out, but just thinking back I use to get these times with just a single dual core pc. Now I'm getting the same times but with 7 X high power computers. Things do look better however, but the scale is not proportionate. Now I'm throwing 5000 % more power than about 10 years ago, but things only look 3000% better. Makes me think in the future how much power I will need to render an image in 5 or 10 years time, just to get marginally better images.
            Kind Regards,
            Morne

            Comment


            • So far I have been using only the main BF subdivs setting to control my rawGI quality. However, each object has its own GI multiplier. These seems a very powerful way to increase GI subdivs only where needed. Do you guys regularly employ this technique.
              Kind Regards,
              Richard Birket
              ----------------------------------->
              http://www.blinkimage.com

              ----------------------------------->

              Comment


              • My tip is - don't care about rawGI, rawReflection, rawAnything, unless you know you're going to use it in comp. Instead focus on normal passes, they directly impact noise level of the image.

                For example black diffusefilter*noisyRawGI = not_so_noisy_GI pass, due to fact that it multiplies RawGi values to basically black... The same goes for reflections, you don't need perfect rawreflections in the middle of let's say sphere that has glossy fresnel reflections, as it's not very reflective there anyway. Pay attention to reflection, refraction, gi, lighting passes instead.

                Also make sure you understand how adaptive DMC sampling works otherwise it will cause many troubles. You can easily over-sample your image by huge margin.
                @wyszolmirski | Dabarti | FB | BE

                Comment


                • Just a thought, how much time do you guys get to play with render settings on a real project with deadlines? Typically we get 2 - 3 days to do an interior of which is taken up by modelling, lighting and texturing. We have little time for reducing render times by tweaking and checking the sample rate and noise levels via the render elements. So we have safe render settings that work in most cases. Personally I think too much time is taken up with having to reduce noise, splotches and render times when clients are becoming more and more demanding. Where it really becomes an issue is animation, anything above 10 - 15 minutes a frame is worrying for us as we then have to outsource it to Rebus or where ever which costs money.

                  I would like to see some of these processes removed and the setting up stages simplified to allow for more time as an artist to work on lighting, materials and composition. As a very simple example, I know Keyshot has its flaws but there are no splotches or noise issues to worry about as far as I understand? It makes me wonder What is the future of V-Ray? With all these renderers popping up that are just one click buttons, the time spent having to set up the renderer is becoming outdated. On the flip side, having the ability to adjust these settings to reduce render times etc is fun and is a good learning tool. Is there any way V-Ray can become more intuitive and be the best of both? Is that where RT comes in? If so it really needs to be in line with the standard renderer with no restrictions or "things it cant do".

                  Comment


                  • I think that in V-Ray 3.0 it can be very simple. You focus only on AA and min shading rate settings for very good quality and fast render times with minimum setup time.

                    For most of scenes that require a lot of AA samples (DOF, motion blur, a lot non-glossy reflections, small geometry details) it would be something like this:

                    AA sampling adaptive/progressive 1/16 to 1/25 with with min shading rate something between 2-10 (I find that in most of my recent cases 6 worked best).

                    For more simple scenes, like studio renders of products, most of the visualizations I think that something between 1/6 - 1/8 would work best with min shading rate at 16-32. It would focus less on AA, more on shading for lights, GI (if brute force) and glossy materials.

                    In both cases you control the quality with noise threshold.
                    @wyszolmirski | Dabarti | FB | BE

                    Comment


                    • Originally posted by wyszolmirski View Post
                      I think that in V-Ray 3.0 it can be very simple. You focus only on AA and min shading rate settings for very good quality and fast render times with minimum setup time.

                      For most of scenes that require a lot of AA samples (DOF, motion blur, a lot non-glossy reflections, small geometry details) it would be something like this:

                      AA sampling adaptive/progressive 1/16 to 1/25 with with min shading rate something between 2-10 (I find that in most of my recent cases 6 worked best).

                      For more simple scenes, like studio renders of products, most of the visualizations I think that something between 1/6 - 1/8 would work best with min shading rate at 16-32. It would focus less on AA, more on shading for lights, GI (if brute force) and glossy materials.

                      In both cases you control the quality with noise threshold.
                      What do you set your materials and light subdivs on then? Or do you leave them at default?
                      Kind Regards,
                      Morne

                      Comment


                      • Originally posted by wyszolmirski View Post
                        My tip is - don't care about rawGI, rawReflection, rawAnything, unless you know you're going to use it in comp. Instead focus on normal passes, they directly impact noise level of the image.

                        Also make sure you understand how adaptive DMC sampling works otherwise it will cause many troubles. You can easily over-sample your image by huge margin.
                        I'd partially agree - of course your final passes are the important ones but the raw passes are very handy when hunting down what's causing slow downs in your renders.

                        Comment


                        • All default. I believe that this would perform well in most of cases. It's probably not the fastest way, but it should save a lot of time on scene setup.

                          And if you want to improve the render time a bit, without much work. Just put noise to 0.005 and test different AA and shading rate settings on small region. I usually render just vertical strip 50px wide with the height of the render, play with settings while looking at sample rate pass. Aim isn't to make it all blue. I don't worry if it goes red/orange in some parts. But I'm mostly checking if most of it isn't red, as this means that it didn't have enough samples to clean up the image.
                          @wyszolmirski | Dabarti | FB | BE

                          Comment


                          • Originally posted by wyszolmirski View Post
                            I think that in V-Ray 3.0 it can be very simple. You focus only on AA and min shading rate settings for very good quality and fast render times with minimum setup time.

                            For most of scenes that require a lot of AA samples (DOF, motion blur, a lot non-glossy reflections, small geometry details) it would be something like this:

                            AA sampling adaptive/progressive 1/16 to 1/25 with with min shading rate something between 2-10 (I find that in most of my recent cases 6 worked best).

                            For more simple scenes, like studio renders of products, most of the visualizations I think that something between 1/6 - 1/8 would work best with min shading rate at 16-32. It would focus less on AA, more on shading for lights, GI (if brute force) and glossy materials.

                            In both cases you control the quality with noise threshold.
                            I agree with this - the new Min Shading Rate parameter is a VERY powerful new addition.
                            Akin Bilgic | CGGallery.com
                            Modeler & Generalist TD

                            V-Ray Render Optimization
                            V-Ray DMC Calculator

                            Comment


                            • In my ideal Scenario, V-Ray would have 6 values to control all sampling:

                              -AA Max
                              -AA Min
                              -AA Noise Threshold

                              -DMC Max
                              -DMC Min
                              -DMC Noise Threshold (Can be locked to AA Noise Threshold)

                              AA and DMC are NOT connected in any way, and no division of values happens - what you set is what you get.

                              So a scene set to 1/8 AA and 1/8 DMC means a max of 64 Primary Samples and a max of 64 Secondary Samples gets taken depending on your noise thresholds.
                              And in each individual aspect like Lights, Reflection, Refraction, etc - a simple multiplier value that defaults to 1.0x. If you set this multiplier to say 2.0x, your DMC sampler will then take a max of 128 samples for that Light, Reflection, Refraction, etc in the case of this scene.

                              No calculator needed, no complexity, just works logically right out of the box.
                              Akin Bilgic | CGGallery.com
                              Modeler & Generalist TD

                              V-Ray Render Optimization
                              V-Ray DMC Calculator

                              Comment


                              • Originally posted by RockinAkin View Post
                                No calculator needed, no complexity, just works logically right out of the box.
                                To be honest, I'm not entirely sure why you need a calculator even now... The max samples are either what's in the material/light/whatever, or the AA samples - whichever is higher, and this is mostly all there is to it.

                                Best regards,
                                Vlado
                                I only act like I know everything, Rogers.

                                Comment

                                Working...
                                X