Announcement

Collapse
No announcement yet.

Vectorscope in 10 bit

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

  • Vectorscope in 10 bit

    Latest Edius WG9

    If I put up the 100x100% colour bars 10Bit 3840 x 2160 50p

    the G and the MG are outside the circle with a screen sampled value of 939, the other colours are inside the circle but have a value also of 939, shouldn't those also lie just outside the circle?

    Just to confirm, the 75 x 75% bars are exactly in their spots as they should be.

    It seems the vectorscope has gone non linear between 75% and 100% and the software values do not have enough headroom. This is proven by adding colour balance and increasing the chroma the spots do not increase at the same rate.

    Minus 18 on colour balance/chroma they are correct after that it goes non linear.

    The waveform RGB monitor is correctly showing 100%

    This is something that needs addressing.

    Mike
    Sys4: Z10PE-D16WS MB 2xE5-2696 Xeon 72 active logical cores. EWG9. 64G RAM. Aorus GTX1080Ti. 55" Q7 1500 NIT HDR 4K TV/Storm 3G Elite/Decklink 4K 12G/8CH audio monitoring, Yamaha RXA-870 A/V. Sys1-3 EWG8 + RX-E1+HDBX1000 MIP in HP xw8600 2 x X5492 CPU 8 cores, 8Gig RAM, Quadro FX3800. All sys Fibre to central media pool - 5TB Axus Yotta RAID + QLogic Fibre Switch. Central VCR rack plus YUV & audio to viewing room with Yamaha AX1 7.1 100 watt per channel amp with 1000W sub 63" HD 3D Samsung TV

  • #2
    Mike,

    Do you have an earlier version of EDIUS so you can compare?

    First, I have to say that I do not have WG9, I have WG8 and I know that WG9 scopes were rewritten to add extra modes.

    My first point would be to make sure you are seeing correctly the data: In the past I was tricked by 'digital signal' into seeing something that was not there. That was due to the fact that unlike analog signal the digital can start and end in a space of one dot and on the digital scopes where the display is so small sometimes it is hard to tell if the dot is on or off. My solution to that is to add Gaussian Blur filter to that signal. That adds smooth transitions between colors and gives you that 'analog' look from years ago.
    Here is screen capture of a project as you described: the filter applied to the color bars with Vectorscope and Color Balance filter visible:

    Click to Enlarge

    As you can see, the Vectorscope now is displaying lines between the color information points. In my WG8, as you can see all points are outside of the circle. That is correct, because all these calibration targets are for 75% calibration signal.

    In this next picture I adjusted the chroma so the Green falls back into the target. The chroma slider has the range of -128, 0, +128.

    Click to Enlarge

    The adjustment I made was -33 (-33 / 128 = -25.78%) meaning that the 100% signal was reduced by 25% hence it is on target a 75%.

    Again, this is what I see on my WG8, you may have different experience on WG9. I would suggest you add Gaussian Blur to the color bars and see if you still see what you described earlier.
    Rob

    Comment


    • #3
      Hi Rob

      The vectorscope has always been odd in this respect, its not version specific.

      Using the Edius colour bars 100 x 100 you can see it just goes into limiting but only on certain vectors, rather like the limit wasn't a circle around the graticule circle but an ellipse.

      I am used to using Tektronix physical vector scopes which of course don't have this issue. The other thing I note is that you can make the spots quite bright on a tube vectorscope but they are very tiny spots in Edius which are difficult to see.
      Sys4: Z10PE-D16WS MB 2xE5-2696 Xeon 72 active logical cores. EWG9. 64G RAM. Aorus GTX1080Ti. 55" Q7 1500 NIT HDR 4K TV/Storm 3G Elite/Decklink 4K 12G/8CH audio monitoring, Yamaha RXA-870 A/V. Sys1-3 EWG8 + RX-E1+HDBX1000 MIP in HP xw8600 2 x X5492 CPU 8 cores, 8Gig RAM, Quadro FX3800. All sys Fibre to central media pool - 5TB Axus Yotta RAID + QLogic Fibre Switch. Central VCR rack plus YUV & audio to viewing room with Yamaha AX1 7.1 100 watt per channel amp with 1000W sub 63" HD 3D Samsung TV

      Comment


      • #4
        Originally posted by createmedia View Post
        The vectorscope has always been odd in this respect, its not version specific.
        I agree.


        Originally posted by createmedia View Post
        Using the Edius colour bars 100 x 100 you can see it just goes into limiting but only on certain vectors, rather like the limit wasn't a circle around the graticule circle but an ellipse.
        Two points here:
        1. At 100% you are at it's limit. I would not be surprised if EDIUS had something kicking in at the very last moment, but I am unable to confirm that.
        2. In a way, the footprint of all 75% colors does resemble ellipse. Here I created a slide with all Vectorscope points: R->Mg->B->Cy->G->Yl and back to Red with a gray box in a middle so when blurring it would look like a solid object:

        Click to Enlarge

        It's not a circle...
        All these colors are at 75% including Blue and Yellow. The B and Yl are scaled different from R, Mg, Cy and G in Rec.709.

        I think I got the point you are making:
        Click to Enlarge

        Here is the same 75% slide with +25% gain in chroma channel applied to it.
        If you look at the left and right sides of the Vectorscope, you should see that the signal looks like if it was cut off, where the Red and Green direction looks like no limiting is applied. That just reconfirms (to me) that Yl and B are encoded at a different scale from the rest of the colors and what you see there are the limits Yl and B will ever reach (meaning '709 transfer function not the Vectorscope scaling)

        Originally posted by createmedia View Post
        I am used to using Tektronix physical vector scopes which of course don't have this issue. The other thing I note is that you can make the spots quite bright on a tube vectorscope but they are very tiny spots in Edius which are difficult to see.
        Yes I agree and that's why I use the Gaussian Blur to make those details visible.
        Rob

        Comment


        • #5
          Originally posted by createmedia View Post
          ...Using the Edius colour bars 100 x 100 you can see it just goes into limiting but only on certain vectors, rather like the limit wasn't a circle around the graticule circle but an ellipse....
          Mike,

          I have run some excel simulations using the BT.709 transfer function and I can confirm that there is distortion of the output favoring Green-Magenta axis so it will look like tilted 50° ellipse.

          Click to Enlarge

          This is an excel spreadsheet. If you are interested in playing with it, just PM me and I'll send you the link to it.

          Getting back to EDIUS, it now does looks like there is something different about how the Vectorscope is displayed. Not necessarily wrong, just different. And what you called 'limiting' I am still convinced that those are just the legal RGB cut off boundaries.
          Rob

          Comment


          • #6
            Many thanks for all that, I guess the programmers need to look at this aspect.

            Mike
            Sys4: Z10PE-D16WS MB 2xE5-2696 Xeon 72 active logical cores. EWG9. 64G RAM. Aorus GTX1080Ti. 55" Q7 1500 NIT HDR 4K TV/Storm 3G Elite/Decklink 4K 12G/8CH audio monitoring, Yamaha RXA-870 A/V. Sys1-3 EWG8 + RX-E1+HDBX1000 MIP in HP xw8600 2 x X5492 CPU 8 cores, 8Gig RAM, Quadro FX3800. All sys Fibre to central media pool - 5TB Axus Yotta RAID + QLogic Fibre Switch. Central VCR rack plus YUV & audio to viewing room with Yamaha AX1 7.1 100 watt per channel amp with 1000W sub 63" HD 3D Samsung TV

            Comment

            Working...
            X