Announcement

Collapse
No announcement yet.

Understanding DMC Sampler

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

  • Originally posted by adanmq View Post
    Hi
    3LP. Fixed 4subdiv = 4² = 16 Samples.

    I want to share my approach to get your feedback in case i´m doing something crazy. It´s working great for my in all kind of situations.


    All material, lights, etc set as default 8 subdiv.
    "Divide shading subdiv" active

    1. First i setup AA using Fixed, focus on the most difficult part of the image and raise AA until i get clean edges. (Ex. AA=4 but vary in each scene)
    2. Disable Adaptive DMC (just for the initial calculations, in order to get exact values, i will change it later). Setup Adaptive amount = 0
    2. Locate the noisiest part of the image (no mater were the noise comes from). Render region it and raise "Min Shading rate" until i get the level of noise i want (You usually need to use big numbers here 128-256 works for me most of the time). (Ex. You get clean noise using 256, now, in the "brute force" tab, "Per Pixel" secondary samples, in this example 4096/4096, take note of this number.
    3. Locate the part of the image whit less noise, and lower the "min shading rate" as much as i can before noise appear on this region (For example: 32). in the "brute force" tab you can take note of the new per pixel information. (using 32 should be 512/512).
    4. So now i know the minimum number of secondary samples your scene need (512) and the maximum (4096) in order to get the level of noise i choose, every single secondary sample above 4096 will be a waste of time.
    5.In order to let Vray move between this 2 values, use MSR=256 and raise the Adaptive amount until the "Brute force" "Per Pixel" tab says (512/4096). (In this example 0.75).
    6.Focus again in the noisiest region of the scene and lower "noise threshold" until its clean (you need to use very low values in order to force Vray use the 4096 samples, but in the worst case scenario the max secondary samples still very low(4096)), or you can raise a little MSR to give Vray more space.

    The beauty of this approach is that you don´t need to focus on where the noise comes from, because you are changing all the secondary subdivs at once using MSR and giving Vray enough space to use adaptive. The problem i have using Adaptive image sampler is the incredible big margins of adapivity i get, and the inevitable oversampling of the heavy AA zones of the image plus the tedious process of optimize every material/light individually. And in my personal tests i get better render times.

    PD: Once more time excuse my English.
    thought I'd try this, but struggling to follow.

    Do you switch back to adaptive after using fixed to test the AA ?

    with your example settings of fixed 4, msr low 32, msr high 256 it would be useful to know what all your final settings are, just so I can see if I am doing it all correctly.

    thanks!
    www.peterguthrie.net
    www.peterguthrie.net/blog/
    www.pg-skies.net/

    Comment


    • yes, the final settings in this particular example are Fixed=4, MSR=256, Adaptive amount=0.75 Noise threshold=0.0007 (you need to enable extra decimal in max). Everything else default.

      Basically you determine the most difficult part of the image and raise MSR until its fine (this is just a test phase). I disable adaptive just to get the exact per pixel samples this area needs and be able to know the exact value Vray is using to sample those areas, but after the initial "test" phase re-enable Adaptive again to let Vray use less samples in the easiest areas, but you need to very low Noise threshold to "force" Vray sample the difficult areas, otherwise when you re-enable adaptive you will get noise in the difficult areas. Sounds tricky (mainly because my bad English i suppose) but for mi its much more intuitive and easy to setup than full adaptive methods.

      Sounds scary at first but if you look at the sample hints you will see that the worst case scenario it´s not so bad, using full adaptive you can get very high sample rate in some areas of your image.

      I´m sure joconnell or Lele will explain it much better.
      My Spanish tutorial channel: https://www.youtube.com/adanmq
      Instagram: https://www.instagram.com/3dcollective/

      Comment


      • thanks for that - think i've got it now!

        so do you think this would work for a wide variety of scenes? arch viz?
        www.peterguthrie.net
        www.peterguthrie.net/blog/
        www.pg-skies.net/

        Comment


        • For me works great mainly in Arch Viz were you can have complex GI or tons of materials/lights and setup each one individually can cost a lot of time (i don´t fully test it in exteriors).

          Once you get used to it, it´s really fast to setup compared to full adaptive methods and gives me clean results and most important consistent render times. But i must test it further in exteriors and heavy DOF scenes.
          My Spanish tutorial channel: https://www.youtube.com/adanmq
          Instagram: https://www.instagram.com/3dcollective/

          Comment


          • It certainly should work for everything, you're pretty much finding what your minimum is, your maximum is and then letting everything else in between pick whatever it needs for itself. I'd say the speed comes from cutting out all the loops at the low end of sampling and it'll certainly be good for scenes with too many materials to tweak by hand.

            Comment


            • i think this is one of the toughest scenes I have ever had to deal with:

              Click image for larger version

Name:	Capture.JPG
Views:	1
Size:	173.6 KB
ID:	855630

              do you think its still a route worth taking or should i save myself days of testing and accept that universal will probably be best?
              www.peterguthrie.net
              www.peterguthrie.net/blog/
              www.pg-skies.net/

              Comment


              • I reckon it'd be worth gathering together a few things like this to profile - if you don't have the time yourself, package it up and let myself and lele have a fiddle with it. It'd be nice it some patterns started emerging in then and then we can form general patterns of attack.

                Comment


                • +1 for that John. It wouldn't be too hard to compare a scattering of scenes and get good starting points for certain types of scene.

                  Comment


                  • I love this topic and hate it at the same moment.

                    Its great that we can fine tune our scene in such amazing matter but I hate it because in real world I just want to press render and everything else should be handed automatically, shaders samples, light samples and everything else.
                    CGI - Freelancer - Available for work

                    www.dariuszmakowski.com - come and look

                    Comment


                    • Yeah it's a tough one. Thankfully there is a few options for setting and forgetting and you've got progressive now too. We're getting kind of greedy with the amount of complexity in lighting and shading we're using these days but still want tiny render times with convenience

                      Hopefully something like the 12 gig titan or the next nvidia card after that will make your life a lot easier Dariusz?

                      Comment


                      • ive been testing this latest method, but in my scene im having issues. i do the setup stage, with adaptive 0, and get it pretty noise free with 256 msr. obviously, its dog slow.

                        trouble is when i then re-enable adaptive (at a value of 0.8 as per my testing) even a noise threshold of 0.00000001 gives me a way more noisy result in the difficult areas than with adaptive off.

                        i can keep adding 0's but it doesnt seem to be making any difference.
                        Last edited by super gnu; 26-05-2015, 07:52 AM.

                        Comment


                        • Originally posted by peterguthrie View Post
                          Despite trying several times to find a workflow that suits me, I default back to universal settings on most projects.

                          adanmq's method sounds promising, but its still too complicated for me..

                          In general, I'm usually fairly happy with universal, but would LOVE to be able to tell vray to spend a bit less time on areas like trees/grass and more time on things like plain white walls (where noise is undesirable)

                          Now that we have MSR and even the new vray/properties subdivs multiplier (in the nightlies) there must be some clever 'for-idiots' approach that could be almost as universal as vlados famous universal settings... anyone?
                          Would be interesting if you could post a slice of the scene which you have trouble with for people to poke at.

                          I usually use a sort of hybrid method where I shoot some rays from shaders and lights and let the dmc sampler "fill in the blanks".
                          Dmitry Vinnik
                          Silhouette Images Inc.
                          ShowReel:
                          https://www.youtube.com/watch?v=qxSJlvSwAhA
                          https://www.linkedin.com/in/dmitry-v...-identity-name

                          Comment


                          • Originally posted by peterguthrie View Post
                            In general, I'm usually fairly happy with universal, but would LOVE to be able to tell vray to spend a bit less time on areas like trees/grass and more time on things like plain white walls (where noise is undesirable)
                            Use the min shading rate and in the 3.20 builds, use the "Subdivs mult" in the object properties. This allows you to set up relatively high MSR for use for flat areas, and tell V-Ray to use lower sampling for trees/grass/other high-detail areas. This works without touching any shader/light subdivs at all.

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

                            Comment


                            • Originally posted by vlado View Post
                              Use the min shading rate and in the 3.20 builds, use the "Subdivs mult" in the object properties
                              this sounds quite interesting, really looking forward for 3.20 to be released.

                              Comment


                              • I also look forward for SP2
                                But I'm really looking into a tutorial someone could put together to explain how to use this new workflow and settings, I'm so lost with all those different way of doing now.

                                Stan
                                3LP Team

                                Comment

                                Working...
                                X