-
Ultrascope RGB/YUV parade seems to be wrong
Hi,
I came across this when playing out animation footage with very saturated colors that also have high transients (e.g. sharp vertical lines): the RGB parade display of the Ultrascope shows overshoots (sharp spikes) that go well below 0 mV or above 700 mV, meaning the signal is out of gamut. But I believe the Ultrascope is wrong there.You can easily recreate this by generating a fully saturated red solid in After Effects and then applying the Grid filter (Generate > Grid), which is an 8-bit effect, on it. This looks perfectly legal on the Ultrascope; the Red is at 700 mV. Set the Grid color to white and the fill mode to normal (so that the solid color is visible again). The RGB parade on the Ultrascope will show out-of-gamut spikes in the red channel that mostly point up (to over 800 mV!) whereas some, strangely, point down. You can even soften the grid and turn the brightness down to 90 %, and it is still out of gamut.
If you change the grid color to black, you will get the respective downward spikes (below 0 mV) in the green and blue channels.
You get similar effects when setting the solid color to blue instead of red.I do not see why the signal should suddenly be out of gamut by adding a white or 90% white grid, so I think the Ultrascope has a real problem here.
I did these tests in 1080psf24, but switching to 720p50 or 576i50 showed the same results.
It also did not matter whether I output from a Decklink or a Kona 3 card, the Ultrascope display stays the same. It also stays the same when I render a Quicktime and play it out in FCP.
My Ultrascope setup is an HP XW4600, Windows7, driver version 1.42.Do other Ultrascopes show the same?
Can somebody with a different outboard scope recreate this? What do you get?Regards
Robert -
3 Replies
-
interesting, I’ll try to give it a shot later today
-
Funny I will also try it. Right now I had no problem at all. At first when we just installed the Ultrascope, I compared it side by side with our Tektronix and the Ultrascope showed the same levels with different test patterns. So I gave it the OK.
I will try the test that you are gaving and compare it with both. See what happens.
Creating something that does not exist to exist.
-
Hi Robert,
Our engineers have reproduced the same source data and investigated the outputs, conversions, etc. First of all, the display on the Tektronix 7120 scope we tried looks just the same. As this is considered to be the benchmark, we have concluded that a fix is not necessary.
The image you have generated in After Effects in RGB,has extreme transitions in the different colour channels ie. going from red (1,0,0) to white (1,1,1). You can get away with this in computer graphics, but not video/broadcast, as this requires infinite bandwidth for instantaneous transition from 0 to 1. Since broadcast is bandwidth-limited, it is necessary to filter the output, which AfterEffects presumably does. Also, it is converting from full-scale RGB to YCbCr on output.
In all this conversion and filtering a certain amount of ringing artefacts gets introduced at the hard edges, so the pure switch from a red column to a white column ends up having more of a ramp of a few columns between them on output. When this signal is sent to UltraScope, it already has artefacts in the YCbCr. When this gets converted back again to RGB for the parade display, the ringing becomes even more evident.
Thanks for your support of UltraScope.
Regards,
Hersh Burston
Senior Product Manager
Blackmagic Design
Reply to this Discussion! Login or Sign Up