Announcement

Collapse
No announcement yet.

Recreating Beauty Pass with ALL V-Ray Elements in Nuke

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

  • Recreating Beauty Pass with ALL V-Ray Elements in Nuke

    Hi,
    could someone help me, i would like to understand all vray render elements, and how to put them together
    i understand that diffuse*rawGI plus diffuse*rawLighting gives the TotalLighting for which the other passes can be added, reflection,refraction, etc.

    but i still have 10 questions, i hope you guys can help me with...

    1)so, if i want to control refraction more i also multiply refractionfilter*rawrefraction (and add that on?) , correcT?

    2)how do i control the shadows more, as adding them on does not really work?

    3) is there any image/graph that shows all of the elements combined (with regard to all the raw channels for maximum control?)

    i think this would be a nice addition to the documentation, just showing a big graph with simple (non-raw) and complex elements hooked up to recreate exactly the RGB color pass...

    (i have searched online everywhere,but nowhere can you find a graph with all vray elements)]


    4)also what can you do with the illuminance channel?

    5) how do i extract all channels from an EXR in Nuke?

    6) is there a way to save 16bit raw image files, instead of 32bit?

    7) would it actually speed up rendering if vray would only calculate half-floating instead of full 32bit

    why not support direct exr output?

    9)in my current test i have multiplied rawreflection times reflection filter, but there is a slight reflection missing which is there in my normal reflection pass--- how come?

    10)what is the correct way to deal with the files, i know that i have to unpremultiply first, to premultiply before viewing, doing all color correciotn inbetween, but, this is not working 100% , reflections turn up missing and im not sure what is going wrong here, any idea?S




    your help is greatly appreciated!
    Last edited by phizikl; 27-01-2010, 05:00 PM.

  • #2
    1)so, if i want to control refraction more i also multiply refractionfilter*rawrefraction (and add that on?) , correcT?

    Yes and no. While this is the theory behind it it does create various issues. This is true for all elements that would need multiplication. Multiplication basically
    doubles your Filtering and hence causes edges.

    3) is there any image/graph that shows all of the elements combined (with regard to all the raw channels for maximum control?)

    There is one in an old thread. Search the forums, i dont have the link handy atm


    i think this would be a nice addition to the documentation, just showing a big graph with simple (non-raw) and complex elements hooked up to recreate exactly the RGB color pass...

    Good call. Second that.


    5) how do i extract all channels from an EXR in Nuke?

    What do you mean ? All channels inside an exr are present within a single branch in nuke. That's part of the beauty. You dont have to (and shouldnt actually) extract them
    to seperate branches.


    6) is there a way to save 16bit raw image files, instead of 32bit?
    yes using the maxscript listener or a startupscript if you want it that way for every scene you work on. The property to set is

    renderers.current.output_rawExrUseHalf =1 (assuming VRay is your current renderer)


    7) would it actually speed up rendering if vray would only calculate half-floating instead of full 32bit
    Half float is just a number format. It aims at allowing as many colors as possible with half the bits needed. Half-Float is not the same
    as a normal 16bit format (wich are integer based). So that is not something you can calculate. Plus if i am not mistaken vray renders
    at even higher precision internally, but am not sure on that one.

    why not support direct exr output?
    It is already supported via the VFB settings and the preferred method for me.


    9)in my current test i have multiplied rawreflection times reflection filter, but there is a slight reflection missing which is there in my normal reflection pass--- how come?
    Maybe it's the specular component ? Are you rendering, saving and comping in linear space ?


    10)what is the correct way to deal with the files, i know that i have to unpremultiply first, to premultiply before viewing, doing all color correciotn inbetween, but, this is not working 100% , reflections turn up missing and im not sure what is going wrong here, any idea?S
    The elements dont have an alpha channel to multiply with.

    Regards,
    Thorsten
    Last edited by instinct; 28-01-2010, 01:27 AM.

    Comment


    • #3
      hi thorsten, thanks alot for ur reply,
      1)so how to deal with this -- getting "no edges" at the end
      5)by branching i meant, how can i use the exr to split the channels so i can grade each one seperately?
      6)nice, thanks for the tip!
      7)i just thought it would have to calculate half as much information, as long as i dont need an Highdynamic range image my thought was that this would speed up render time

      I did not know u can export EXR from VFB, thanks! i guess this is not possible for rendering out animation though?
      9) jes i think i forgot specular, i am definately in linear space, all my files are, and i use sRGB as my Look up table (with files being at 1.0 and set to linear)

      10)what do you mean the elements dont have an alpha channel? i save them as .png with alpha (all the elements) --

      i worked on 2012 with pixomondo as an intern, and there i learned about comping in nuke and about unpremultiplying first, and only last step to premultiply again, but, this is not quite working for me...

      any information or help on that?

      thanks sofar already !!!
      cheers
      ian

      Comment


      • #4
        for 1) you can't properly. This is basically by design. The only way around would be by using pixel coverage value, but this can get infinitely complex heh.

        5) in nuke you dont. You add a grade node and select the channel you want to grade instead of creating seperate branches. This simplifies scripts greately. If you insist on seperating them you can use the shuffle node

        7) The calculation is the same only the storage format for the information is different. The result of an equation doesnt affect the calculation time

        you can save to animation just fine. In the VRayVFB rollout you can specify an outputfile. Just replace .vrimg witn .exr and it will save EXRs with the framenumber added to the end

        10) You cannot unpremult or premult without an alphachannel. And elements saved from VRay do not have an alpha channel. So the Alphachannel is black and that means that a premult will make the RGB all 0
        And btw. If you're saving to png then saving linear is not really a good idea as this will prolly loose quite some information. I'd suggest sticking to float or halffloat exrs

        Regards,
        Thorsten

        Comment


        • #5
          here you can see the reflectin_pass straight from elements (thats how it should look i guess)
          then the rawflection_x_reflectionfilter - there you see there is a reflection missing.. (also more green, sunlight missing) and also the edge is a problem here.... why is all that?


          in the picture rawreflection_x_reflectionfilter_premultiplied i did i a premultiply, just to show that there is a strange reflection , which i wonder where that came frome? dont have any other channels connected...

          this is the same reflection that is making the refraction pass brigh, as soon as i use unpremultiply...

          i just looked at the alpha, which seems to be a problem, what is going on there?

          so in the last
          Attached Files

          Comment


          • #6
            here you can see the elements comped together, as you can see, i almost have the rgb_color passe,
            but

            1)that slight reflection is missing (since im trying to use raw_refleciton x reflection filter)
            2)the edge is there, u can see it around the teapot and pronounced in the back as well



            3)also, the elements do have an alpha channel, here i saved the alpha of the rawreflectoin just as an example (you just chek the -save alpha- when telling vray where to store the seperate render channels and file format, for which you can do this for .png .tif open exr, pretty much any format, they all have an alpha channel, and you can save .png in 16 million colors or 240trillion colors, i think thats quite enough to grade

            (i could have saved in better format, but this shouldnt cause any problem, only when i start to grade alot will that extra information be helpful)
            Attached Files
            Last edited by phizikl; 28-01-2010, 03:11 AM.

            Comment


            • #7
              It might come into play without grading too. Imagine very high values or negative values. This will be stripped in png's and hence change your comp. It's tough to diagnose from a few images. But it looks like a refraction is missing rather then reflection

              Comment


              • #8
                here you can see,
                refraction and raw_Refraction x refraction filter are identical

                how it should be i guess



                the specular pass is empty, and i cheked the other elements, the missing refleciton is not there

                (and it has to be in the reflection pass, since the normal reflection has that extra reflection inside, just that raw_reflection and reflecitonfilter multiplied have this part missing)

                i will gladly do another test with the format of your choice
                Attached Files

                Comment


                • #9
                  this missing reflection i could even deal with just by using the reflection pass,

                  but my main problem


                  how to get identical rgb_Color pass with no edge problems


                  this is possible?!?!


                  anyone?

                  Comment


                  • #10
                    i also just tried saving an openEXR but it does not contain all the vray elements?
                    how to accomplish this, straight from VFB?

                    Comment


                    • #11
                      if you tried saving with the save button this is not what i meant. Check here:
                      http://www.spot3d.com/vray/help/150SP1/vfb_index.html

                      And nope it is not possible to get rid of the edges when using the Elements that you have to multiply. If you use the combined result it should work mostly (Reflection instead of Raw + Filter)

                      Regards,
                      Thorsten

                      Comment


                      • #12
                        When i save a vray raw image file as .exr, then i get the all the elements in nuke (ok, that is what you meant, now i get it)

                        But Still, i multiply refraction_filter * raw_refraction

                        or reflection_filter * raw_reflection

                        and the edges are different than just using the reflection pass or refraction pass ?

                        i surely am not the first person to notice this
                        but you say, i cant get rid of this problem with the multiply , but how do people grade using the elements, since you have to use diffuse*rawlighting plus diffuse*rawGI .. or what is YOUR workflow? there are thousands of people renderin vray elements and comping them in nuke, how do they do it? they just add allthe passes? where is more information about this? seems to be overlookd somehow (there are of course some comping tutorials, but none that mention these problems here)


                        ok-- so you say, it is not possible

                        haha, wow, i mean
                        but who thought of this?


                        how do people use the rawrefraction and refractionfilters then?
                        Last edited by phizikl; 28-01-2010, 06:36 AM.

                        Comment

                        Working...
                        X