Announcement

Collapse
No announcement yet.

FStorm VS Octane

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

  • #31
    Thanks for your feedback and tests guys!
    The main purpose was to see how to speed up interiors in general - I think we got some good results and tricks that we can translate for real projects.

    As Antony mentioned, a good setup seems to be the middle way; a good balance between GI Clamp, Light Samples, Portals and Exposure.
    Last edited by RobSteady; 10-19-2016, 09:05 AM.
    FStorm 1.3.0d I Max 2018 I Win7 I i7 5930k I 64GB I 3 x 1080Ti

    Comment


    • #32
      So i rendered Fstorm (0.35) on default settings. left for 13 mins and then did a comparison with VrayRT 3.40.03 for 13 mins, with a denoised version too.
      using two MSI GTX 760 4GB, driver 373.06, windows 10

      Comment


      • Sylk
        Sylk commented
        Editing a comment
        Your FStorm settings are bad. Not a good comparison for image quality.

      • manish_mv
        manish_mv commented
        Editing a comment
        this is not correct comparison fstorm lighting is different from vray and the scene has no textures in it also. comparison should be correctly done with a completely finished scene.

    • #33
      Sylk i'm not sure what you mean, I opened the OP scene, reset Fstorm to default settings and hit render. i left for 13 min because i was getting bored. i then set VrayRT to render for the same amount of time. those were the results.

      The Vray setup is the same as the Fstorm setup. Sun and sky with portals at the windows. default settings for VrayRT except for changing it to CUDA and setting the time limit.

      In this particular scene i would say that VrayRT is miles ahead in speed and quality, but with a more complicated scene with textures, displacement, reflections, more lights maybe FStorm will have higher quality and be faster?
      Last edited by c0mpani0ncube; 10-20-2016, 09:26 AM.

      Comment


      • adams
        adams commented
        Editing a comment
        Well that seems a fair enough analysis, especially the bored part.

    • #34
      Originally posted by c0mpani0ncube View Post
      So i rendered Fstorm (0.35) on default settings. left for 13 mins and then did a comparison with VrayRT 3.40.03 for 13 mins, with a denoised version too.
      using two MSI GTX 760 4GB, driver 373.06, windows 10

      what are your vrayRT settings?
      m.

      Comment


      • #35
        adams is there a problem with me getting bored? i felt the image had converged enough to make a comparison against vrayRT

        wastzzz default vrayRT settings. only switched to CUDA engine and i set a time limit of 13 mins which is how long the FStorm render for. my GPU are obviously slower than other because to let the full 1000 samples in Fstorm run was estimating an half an hour but i wasn't going to wait that long.

        Comment


        • #36
          As this thread is already open I take the chance to share my test for a better understanding on the conversion material between Octane and Fstorm for Speed and Quality performance, said that I like to point out that as Octane does not convert Corona ( yet?) I had to manually set the scene so that is not the best comparison but I guess very close.
          The scene is a 3Dsky.org model and Corona ready that Fstorm converted almost without a glitch, although the HDRI is missing from the pack so I can't have the exact environment.
          What I generally find is that the Glass value converted from Corona ( ior 1.52 ) does not compare much the Octane one, so I had to lower Octane ior to 1.3, and the emitting filament material is set easily in Fstorm thatn Octane, the converted one is set to Power 33.705 while in Octane I had to adjust both Power and Temperature ( were Power is set to 0.001 ) Fstorm seems handling the matter is a simple way.
          I like the glass better for some reason, on a thin glass surface ( no double ) it seems reflecting at IOR 1.52 more natural than Octane were I had to adjust Trasnmission color and set IOR to 1.2, the rest octane did it faster than Fstorm in this case but I'm sure I'm not good at setting things up correctly for a better comparison,,hard to say, I love the way Fstorm is getting us all by the storm ( lol ) .. surely Andrey is an hell of a programmer genius for sure, this is another revolution and being him behind Octane Core as well he definitely revolutionized the world of rendering.

          So take this as it is, but both renders were stopped at 1000 Samples with:

          Fstorm set on his default with multiplier at 0.001 for the environment.

          Octane: PT Diffuse Depth 4 | Glossy 16 | Coherent 0.0 | GI Clamp 1.0 and a touch of Post glare

          Comment


          • #37
            alancodger

            Both are great renders but definitely Fstorm is the more realistic result. Big difference in times. 1.27min (Octane) against 7.27min (Fstorm) or is that a typo on the pictures since the .27 is the same?


            Not sure why c0mpani0ncube is getting so much flak, seems a fair comparison out of the box so to speak. Granted those stalwarts here who know how to tweak Fstorm to produce the renders in the gallery, will achieve faster/better results but the same can be said about Vray (although I think the newer versions have somewhat pre-optimised defaults, so could be unfairly balanced). Besides, personally I actually think the Fstorm render in c0mpani0ncube's post, has the most natural lighting so far, closer to the corona result.



            Off topic but does Vray RT GPU have issues with specular highlights even on fully supported materials? Comparing RT CPU to RT GPU it doesn't look like it adds the light cache phase even though it builds it. I haven't tried Vray in max so it might be dcc plugin specific but I have used the same core.
            Last edited by rvectors; 10-21-2016, 05:53 PM.

            Comment


            • #38
              Well somehow I should agree although I still love to compare the 2, not that I won't have both installed as any Octane users here, what strikes me is that I got some feeling that emitters are less noisy than octane with few adjustment instead of going through several setting like in Octane, I mean, the entire issue is that is so close to it BUT better? Will keep working on materials. Thanks

              Comment


              • #39
                This is another test on a little basic scene Than I was working on, the Octane scene has been converted almost fully and I really wish I could do the same with Octane but I love both renders truly, I can see though Fstorm has a certain contrast in color, looks better,weird it looks more blue-ish past the conversion... but I I can get it back in Octane using the Linear response.. so nothing dramatic, testing, again Octane seems quicker here although looking at the screen I can say that Fstorm did the job at 500 samples really..

                Original renders 1753X1240 reduced in photoshop for this forum @ 1400x900px

                Comment


                • #40
                  anyone tested these scenes using hdri? i dont have octane but i guess someone can compare, it seems most of us are lighting scenes with hdris but not sure if it makes the scene render faster or slower or more noisy

                  Comment


                  • #41
                    Another test to understand textures color and generally peformance.. no doubt Fstorm clear the noise faster and like I wrote before Fstorm got more contrast than Oct. both are fine 2 me but yes I loveFstorm speedv and simplicity in his settings

                    Comment


                    • mitviz
                      mitviz commented
                      Editing a comment
                      the second one is octane right? it does have something about it i love

                  • #42
                    I have some inquiries if it’s not too much trouble , guide me about this. These inquiries are displayed underneath here


                    https://fstormrender.com/forum/forum/off-topic/12062-unreal-engine-5-announced

                    Comment


                    • #43
                      Sitting tight for most appropriate response straightaway. I’m excessively confounded here

                      Comment

                      Working...
                      X