Announcement

Collapse
No announcement yet.

Modo render to v-ray cheat sheet?

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

  • #16
    Really nice results!
    For production don't go below 24 for Max subdivs,but for testing your settings should be quick and fine. Maybe go up a tiny bit on the noise threshold(like .25 or .3) if you want to save more time. But it will depend on the scene a lot,because if you have tiny details in the scene the denoiser can affect it. Take a look here for example,you can see the effect of denoiser,
    https://www.youtube.com/watch?v=AeUEMPJ3uHU
    Links in the description of the video too.
    Muhammed Hamed
    V-Ray GPU product specialist


    chaos.com

    Comment


    • #17
      Originally posted by Muhammed_Hamed View Post
      Really nice results!
      For production don't go below 24 for Max subdivs,but for testing your settings should be quick and fine. Maybe go up a tiny bit on the noise threshold(like .25 or .3) if you want to save more time. But it will depend on the scene a lot,because if you have tiny details in the scene the denoiser can affect it. Take a look here for example,you can see the effect of denoiser,
      https://www.youtube.com/watch?v=AeUEMPJ3uHU
      Links in the description of the video too.
      Thank you!
      So apologies for the naive question. Which noise threshold Are you referring to? I'm still trying to sort out all the different thresholds and subdivs lol.

      Also when people say samples, are they referring to subdivs ?

      BTW, I was able to reproduce the blotchy denoise issue, it only seems to happen on RT renders. I will post some pictures soon. I was able to get rid of them by letting the RT GPU run longer, but at that point it was longer than the production CPU render time.

      thanks for all your help!

      Comment


      • #18
        You are welcome!
        I was referring to those settings when I said they are good as test settings,
        • Min Shading Rate: 6
        • Min Subdivs: 1
        • Max Subdivs: 6
        • Threshold: 0.02

        Use those settings when you need a very fast render for testing something maybe. But for production rendering don't go below 24 for the Max subdivs. Let's say you are rendering animation and you want more render speed per frame. In that case you can use min subdivs of 1,Max subdivs of 24 and threshold of .025 .. If it is not as quick as you want go up on the threshold again like .03 .. So on so forth with keeping an eye on the denoise result until you get the best render times without losing any details.

        Also when people say samples, are they referring to subdivs ?
        The actual number of the samples is the square of the subdivs.

        but at that point it was longer than the production CPU render time.
        Vray RT CPU is not as fast as the production renderer (Progressive/adaptive). In some scenes I found RT CPU to be very slow compared to adaptive (Up to 30% or more). And You will need lower threshold with RT CPU to get a result that is as clean as what the adaptive sampler does.
        Vray RT CPU/GPU also gives slightly different results sometimes,so keep that in mind.
        And it would be nice to share the blotchy issue so that devs help you with it.
        Muhammed Hamed
        V-Ray GPU product specialist


        chaos.com

        Comment


        • #19
          Originally posted by Muhammed_Hamed View Post
          You are welcome!
          I was referring to those settings when I said they are good as test settings,
          • Min Shading Rate: 6
          • Min Subdivs: 1
          • Max Subdivs: 6
          • Threshold: 0.02

          Use those settings when you need a very fast render for testing something maybe. But for production rendering don't go below 24 for the Max subdivs. Let's say you are rendering animation and you want more render speed per frame. In that case you can use min subdivs of 1,Max subdivs of 24 and threshold of .025 .. If it is not as quick as you want go up on the threshold again like .03 .. So on so forth with keeping an eye on the denoise result until you get the best render times without losing any details.
          Ok thank you, I will try that.

          Originally posted by Muhammed_Hamed View Post
          The actual number of the samples is the square of the subdivs.
          Ohhh, I think I recall richard mentioning that in the video now.

          Originally posted by Muhammed_Hamed View Post
          And it would be nice to share the blotchy issue so that devs help you with it.
          I'm attaching the blotchy issue.
          I'm just doing a modelling exercise right now trying to model an apple tv remote I have.
          I have a basic ground surface, and then a studio hdr, and the remote.
          RT - GPU seems to not do well with the denoiser when I set the max noise to 0.05. But if I set it to 0.01, the blotchyness goes away, but the render time is very long.
          Conversely, the production render did it in 9 minutes and very similar quality to the rt-gpu version that took 26.

          I have a 6gb Nvidia 1060 gtx btw. For some things, rt gpu seems MUCH faster, but for things like this, not so much.

          Comment


          • #20
            Ok weird, I did another render with rt-gpu, this time, I set the same 0.01 stop noise condition, but I also changed settings to:
            • max subDivs: 24 (from 1
            • Threshold: 0.025 (from 0.015)
            Render time was faster? I guess because I chose 0.025 for the threshold? Or my gpu wasn't running as efficiently last run? not sure.
            I wish there was a way to tell if the RT is actually in progress or finished. Sometimes the 'timer' counter pauses, so I can't really tell if its finished.

            Comment


            • #21
              Modeling is very nice. Good job!

              I'm attaching the blotchy issue.
              I think I seen this problem before. It was related to adaptive lights,but I might be wrong. You should post a thread in bugs forum so Vladimir tells what is going on. But first I think you are doing something wrong.
              for RT you should ignore the image sampler settings. They doesn't apply to RT(Min shadin rate,Min subdivs,Max subdivs,noise threshold). To control RT you have to use the max. noise,GPU ray bundle size and GPU rays per pixel. I don't recommend changing anything except the Max. noise. If you are running into interactivity problems use low thread priority from Vray Modo settings. But I recommend turning it back off while doing the final rendering because it affects performance a lot. Also,the render time is not bad for your 1060. This is the expected performance from it and your CPU should catch up if you have like 6th/7th Gen Quad core core i7 .Keep in mind that Modo and windows take a chunk of the GPU performance and VRAM,so it is always good to have a dedicated card for display and windows perfromance and another card(s) for rendering.
              Back to the settings you used,keep in mind that the threshold you use for Adv. production renderer will give slightly different noise pattern if used with RT(Max. noise). RT always needs lower noise threshold to match cleanliness of the production renderer(At least from my testing).
              For the first render you used .15 threshold,then you used .05 threshold for the second render and got the weird problem. I just want to say that this scene is not simple for AA. You have depth of field and a lot of blurry reflections. It is very important to keep an eye on the sample rate pass. It is really useful to know what is causing the noise in the render and what settings you need to change. I have watched Grant Warwick sampling tutorial many many times and I recommend that for you until you get your head around it.
              Actually your 4th render is the best in my view and render time is not that bad. You don't need to wait until RT or progressive is 100% done. If you see that the render is clean enough,take a look at the sample rate render pass to see what the sampler did so far,then you can pause the render and let the denoiser finish the job.
              Your second render is good also maybe go down on the Max. noise to .035 or so and maybe use an agressive denoiser profile. Test those out until you get something you like,but I really think that the 4th render is the best already and the render time is good for this scene(Unless you have got dual Xeons lol)

              Use a monitoring app to make sure the card is working without problems in the first place(no throttling) and you will be able to tell if the render is finished or not. Use CAM by NZXT or GPUZ. I have CAM running all the time on my other monitor.
              Would be nice to post a thread in the problems section to report the weird behavior in the second render.
              Muhammed Hamed
              V-Ray GPU product specialist


              chaos.com

              Comment


              • #22
                Awesome.

                Thank you Muhammed! I'm actually spending a week off of work to learn as much modo as I can, haha.

                I'll post the denoiser issues I see on the bugs page. I've noticed it on a few scenes I've tried. Are the HDR's considered adaptive lights?

                I didn't realize the sampler settings didn't apply to RT! That's good to know!

                I'll get one of those apps to monitor for sure. I have a ryzen 1600 cpu right now (upgrade from my MBPro), and I usually leave my system alone when i do a render like that. I don't run the gpu in low priority mode, I just let it crunch away.

                Do you recommend any online tutorials that teach v-ray well? I've got a Pluralsight subscription, but a lot of the tutorials are for max or cinema, not modo
                I'm assuming the general layout is pretty much the same across the board, but the things i'm not sure about are specific material chains, like how 3dsMax does materials vs Modo shading.

                Thanks again for all the help, really learning a lot!

                Comment


                • #23
                  Are the HDR's considered adaptive lights?
                  Adaptive lights is an algorithm that Vray uses to calculate lights faster. So you have adaptive lights on with any kind of lighting including HDRI.

                  I have a ryzen 1600 cpu right now (upgrade from my MBPro),
                  That is a beast of a CPU. Will be faster than any Macbook Pro. And yeah if you don't use the system while rendering,there is no need for low priority because it reduces the performance a lot with your 1060

                  Do you recommend any online tutorials that teach v-ray well?
                  Check out Mograph plus introduction to Vray 3.4 for 3Ds Max . He updated the guide for 3.5 too. He went through everything in the user manual with very good examples. He worked in 3Ds Max,but I can guarantee that all the workflows you will be able to do in Modo very easily. And if you needed help just post in the forum and I will help you. After a bit of time you will be familiar with how Modo deals with Vray and you will even find Modo has better controls than Max or Maya.
                  Any videos on Max or Maya you should be able to follow along with,but don't watch old videos before Vray 3 because many things has changed. Also Mograph plus will only teach you the tools and how to use them,but he doesn't touch on production workflows for materials/lighting..etc ,so if you want something on those techniques check out Mastering lighting and Mastering Vray by Grant Warwick. He touched deeply on materials/lighting/Post production. I have watched all those tutorials so if you need help post here and I will be happy to help you.

                  C4d Vray tutorials are different and will confuse you,so just focus on Maya and Max's videos.

                  but the things i'm not sure about are specific material chains, like how 3dsMax does materials vs Modo shading.
                  it is very very close to 3Ds Max's workflow. I started using Modo 10 months ago coming from C4d. C4d has no material nodes,so every renderer for C4d has its own node based material system (Arnold-Cycles-Octane) which I used and most users have no problem using a different node based system for each renderer. It takes like a day or 2 to get your head around it.
                  Muhammed Hamed
                  V-Ray GPU product specialist


                  chaos.com

                  Comment


                  • #24
                    Awesome, thank you very much for all the info. I finished my remote project, and will post in the images forum!

                    Comment


                    • #25
                      I'm attaching the blotchy issue.
                      This looks like the normal artifacts when you apply the denoiser on a very noisy image.
                      In RT (CPU or GPU) the denoiser will start up from time to time (based on the "image effects update freq." value), denoise the image and show it in the effectsResult channel.
                      Maybe it started when the image had some significant noise / fireflies and that's where the splotches are coming from.

                      Greetings,
                      Vladimir Nedev
                      Vantage developer, e-mail: vladimir.nedev@chaos.com , for licensing problems please contact : chaos.com/help

                      Comment


                      • #26
                        Originally posted by Muhammed_Hamed View Post
                        Really nice results!
                        For production don't go below 24 for Max subdivs,but for testing your settings should be quick and fine. Maybe go up a tiny bit on the noise threshold(like .25 or .3) if you want to save more time. But it will depend on the scene a lot,because if you have tiny details in the scene the denoiser can affect it. Take a look here for example,you can see the effect of denoiser,
                        https://www.youtube.com/watch?v=AeUEMPJ3uHU
                        Links in the description of the video too.
                        24 max subdivs? I wouldn't put a hard rule on subdiv levels because a lot is going to depend on the resolution you're using and other elements to your scene. For print you can always get away with lower, and even for higher res online viewing of a still, you don't always need 24. Often you can just as easily get away with 8-16 IMHO on high res assets esp. if you're comping back in the denoiser elements where needed/works best.

                        I'm not saying you don't need higher max sub divs at times. I've needed as much as 65 with a very tight threshold. Just that it's always best to see what you can get away with before defaulting to something that could exponentially blow up your render times when it might not be needed.

                        It also doesn't matter what you set the max sub divs or max threshold to at a certain point because Vray will top out at one. If I have 50000000 max sub divs, and a threshold of .9...well it's never going to use that many max sub divs. I prefer to work max sub divs and threshold in tandem.

                        Say I have some noise in the scene I want to take out/reduce. I draw a region around it.

                        First step for me is to actually lower the threshold.
                        I do that until I see no change. Then I up the max sub-divs a bit.
                        Rinse-repeat
                        If I end up at a bit of noise that just will not get fixed, even with a very low threshold and high max sub divs, I then start slowly adding min sub divs.
                        If THAT fails, I then mess the min shading rate. I've only had to do that once so far though.

                        This is my opinion for the production render though set to adaptive. Progressive or RT...that's different.
                        Last edited by GidPDX; 11-07-2017, 07:03 PM.

                        Comment


                        • #27
                          You are definitely missing the point. This is the first time I see someone still does scene optimization AFTER new sampling system was introduced in Vray 3.3
                          Since 3.3 I never needed to draw a region to play with the max subdivs,min subdivs or the min shading rate. I just use numbers and they always give me best results/render time. This is the point of the new sampling system and keep in mind that every default value there is not just a random number.
                          The sampling system is smart enough to figure out where and when to use the Max. Subdivs. Using the noise threshold you control WHERE the max subdivs is used. Like I said and with confidence I say if you want faster render DON'T touch the max subdivs(below 24) and instead use a higher noise threshold. That way you are getting consistent noise pattern across the render and most importantly,the very few areas that usually need higher subdivs are not cut off!!
                          Using lower Max. subdivs than 24 will not make the render faster at all compared to using a higher threshold(and keeping the Max. subdivs untouched). You will get areas that is completely cut off(you can tell using the sample rate render element) and you will get very strong noise in those areas.
                          Here is how you use the adaptive sampler in Vray,
                          https://www.youtube.com/watch?v=OS3XP9TDCwM
                          There was a CG garage podcast with Lele,the guy who worked on the 3.3 sampling system with Vlado. He talked about why they changed the sampling system in the first place and how it is supposed to be used after the update,
                          https://labs.chaosgroup.com/index.ph...3-for-3ds-max/


                          Muhammed Hamed
                          V-Ray GPU product specialist


                          chaos.com

                          Comment

                          Working...
                          X