Does anyone know what the luma level should be on static test patterns? or know how to read a vector/luma scope?
Please bare with me as I'm not sure I'm asking the correct question, & I need to give a few details in my story.
I capture analog NTSC video and just bought an Analog Devices EVAL ADV7842-7511 video decoder board for its 3D comb filter. Settings are CVBS-in and HDMI-out at 480i 10-bit 4:2:2 29.97fps.
I've used many external comb filter devices and have noticed when using Component-out, the image always seems brighter than when outputting via s-video. The image is even brighter with this board.
Testing with the Video Essentials laserdisc, and using the program that came with the capture card I can view a vector and luma scope. In this particular brightness test pattern the left-most vertical black bar, which is blacker-than-black should be set to the same black level as the background (may be difficult to see here).
I can adjust the capture brightness of the signal and I've lowered close to what the pattern calls for. I noticed that the value on the luma scope chart falls to around zero when the brightness seems properly adjusted.
Every capture program is different, but with the default brightness at 0, the luma chart value is about 10 and with the brightness lowered to -8, the luma chart value is 0.
Question is should certain test patterns yield a luma value of zero when everything is properly adjusted? With different images and patterns the luma value jumps around as expected. It just seems when I adjusted the brightness down to where it's supposed to be, the luma value is zero. So this might be a good indication that I have adjusted correctly?
Capture card is Osprey 827e and cap program is Ceylon, although I actually cap with VirtualDub but can access Ceylon to adjust color specs.
+ Reply to Thread
Results 1 to 22 of 22
For YCbCr black level 0IRE is code value 16, for White 100IRE code value is 235. However U.S. NTSC video colour encoder shifting black level by +7.5IRE, in decoding process this +7.5IRE called pedestal is removed in decoder so 0IRE code value is 16. Problem is if your video decoder works in Japan mode where there is no pedestal thus you may see black level at +7.5IRE.
It is common problem with NTSC - uncertainty related to pedestal removal - this is video decoder job and it should be done by capture device if properly configured. Trying to fix/correct pedestal after capture will lead to increased quantization noise as re-quantization from 92.5IRE to 100IRE scale is required.
How to read - it is easy - i think you should move slider to desired line - meter showing single line (at snapshots line 1). To check vertical pattern at once you can rotate video by 90deg so you will see whole pattern shape.
Last edited by pandy; 21st Jan 2019 at 16:06.
Are you saying it's best to capture at the correct YCbCr black level vs. adjusting it in post?
Below example of Pluge signal definition - i think you need to search for description of "your" Pluge signal.
check also https://en.wikipedia.org/wiki/Picture_line-up_generation_equipment
Perhaps SMPTE Pluge will fit your pattern
Last edited by pandy; 21st Jan 2019 at 18:07.
Can I ask how you know the one pluge that you should see has a IRE value of +2.5?
According to the Video Essentials laserdisc, the left pluge should be even with the black background in the SMPTE pattern, and same goes with "a pluge signal as viewed on a monitor". I'm just eye-balling the slider adjustment.
Thanks for the ITU link.
With limited range YUV (Y = 16 to 235, U,V = 16 to 240) everything below Y=16 is supposed to render at the same shade of black as Y=16. You can still see the data on a scope or waveform monitor*. Everything above Y=235 is supposed to render as the same shade of white as 235.
You generally want to capture with close to the correct levels. As an extreme example supposed you captured so that you only got two Y values, blacks at 234 and whites at 235. You could contrast stretch that to 16 and 235 to restore blacks but you will have lost all the detail between those two levels. So the closer you are to the correct levels the less problems you will have in post.
* In practice some capture devices may crush blacks below IRE 0 or above IRE 100 (or thereabouts) so you may not see them once the signal has been digitized.
Last edited by jagabo; 21st Jan 2019 at 21:41.
+1 what jagabo said.
But if you MUST err (please do NOT), err just on the insides of the scale (flatter contrast) and not on the outsides (where you could/would crush/clip).
If your decoder remove pedestal automatically then you should not see things darker than black...
The pluge is specified in SMPTE RP-219, along with the rest of the color bar signal, as well as explicit instructions on how to set up a monitor (4.3.1).
Below is a .bmp file with pluge patches at 12, 16 (invisible) and 20 per RP-219.
7.5 IRE setup became obsolete years ago. Now black is digital 16 which equals 0 IRE.
Forget Wikipedia. Their representation of SMPTE bars was inaccurate to begin with and the corrected representation was vandalized.
Note that the primary and secondary colors consist of 16 and 180, e.g. yellow consists of R=180, G=180 and B=16. So-called "SMPTE color bars" have been bastardized and incorrectly explained all over the Internet by people, even software developers, who are ignorant of the actual standard.
Why this matters is that the test equipment at professional video facilities, those Tektronix scopes that you and I can't afford, is calibrated to RP-219. If you submit content and the color bars are out of spec, in the worst case your content will be rejected. PBS and many of the on-line services such as Netflix are real strict about this.
Your desktop computer monitor probably doesn't have a brightness control. I solved this by using a TV set with an HDMI input which is fed by my computer's video card. The TV receiver has a brightness control and now I can set up the blacks the way they are supposed to be.
I'm trying to say that those three vertical bars on left looks like under black, black, just over black for example -2.5IRE,0IRE,+2.5IRE IMHO without touching anything and with correctly set NTSC US you should see only 1 vertical bar +2.5IRE. If you start removing pedestal after capture then you will loose some quality.(this is possible only in US NTSC where 7.5IRE pedestal exist)
In 2019 7.5 IRE = digital 32 and we know this is wrong. Per BT.709 black = digital 16.
Really, you know better than that.
Last edited by chris319; 6th Mar 2019 at 05:29.
If you're talking about amateur video from Joe YouTube User or are only interested in viewing on a monitor or cinema screen and don't care about interchangeability, you probably don't even need color bars or care about video levels.
You would benefit from reading the standard.
Commonly in digital composite NTSC encoder this is done by adding in digital adder circuit to incoming from video data, digital values representing sync pulses and pedestal and feeding this to for example 10 - 14 bit DAC to create linear analog signal - this is very convenient way of dealing with signal in digital domain. Look at this for example https://opencores.org/projects/graphiti - not the best DENC but is shows how things are done.
If it says "IRE" it's supposed to be analog. So don't confuse the two.
Also, Analog BROADCAST went out the window, not analog video in general. There is still plenty of analog video out there, and will be for years to come.
In analog land,
0 IRE=Black everywhere except US NTSC, where it has been 7.5 IRE due to setup/pedestal.
100 IRE = White EVERYWHERE.
In digital (8bit) land,
Y 16 = Black EVERYWHERE
Y 235 = White EVERYWHERE
unless you converted to PC Full setting, which is Y=0...Y=255, but that is totally NON-STANDARD with regard to analog conversions or pro video.
If it says "IRE" it's supposed to be analog. So don't confuse the two.
Analog BROADCAST went out the window, not analog video in general. There is still plenty of analog video out there, and will be for years to come.
Although Congress established a hard deadline of June 12, 2009, for full power TV stations to cease analog broadcasts and begin operating only in digital, and the Commission set a transition date of September 1, 2015, for Class A television stations to complete their transition, neither of these deadlines applied to low power television stations or TV translator stations (referred to herein as "LPTV stations"). Therefore, although all full power and Class A television stations have ceased over-the-air analog broadcasting, LPTV stations are continuing to transmit analog signals.
7.5 IRE setup was added as a buffer zone. If black were 0 IRE without setup, excursions below black might bother sync circuits. In digital this is why black and white are 16 and 235 and not 0 and 255. In ATSC, 0 and 255 are the sync levels.
Last edited by chris319; 6th Mar 2019 at 16:05.
Yes, we know there are a few LPTV stations, but basically nobody is counting them. Certainly not you, who said in the earlier post that it went out the window. Regardless, my point there was that it doesn't matter about broadcasting, because some elements of analog video recording and playback are still with us for the present, and so still is 7.5 setup. I think you knew that and were just trying to obfuscate.
But, look up IRE, it historically refers to Composite Analog Video.
That xx IRE = yy digital Y is an equivalency. Thus 100 IRE = Full White, Full White = 235 Y, and by transitive property 100 IRE = 235 Y.
Last edited by chris319; 6th Mar 2019 at 16:56.
The point that IF one continues to have to deal with analog US NTSC video (capturing, transferring, processing), one will continue to have to account for NTSC 7.5 IRE setup. And it should especially be done BEFORE the point of digitization.
The other point being the OP was quite satisfactorily answered prior to your muddying detour. I donít think, or I sure hope, that wasnít your intention, but thatís what it amounted to. The OP made it clear in the 1st post that they were working with analog US NTSC sources and reference material.
Problem with pedestal begin in capture if your HW is unable to remove pedestal before digitalization and later captured data are output as 8 bit or if 8 bit ADC is used to capture signal and later trough digital processing pedestal is removed. Those two cases will produce suboptimal conditions for signal capture as we loose signal data.
The OP made it clear in the 1st post that they were working with analog US NTSC sources and reference material.