Do HD studio cameras use studio swing pixel mapping? - AVS Forum
Forum Jump: 
 
Thread Tools
post #1 of 8 Old 10-04-2012, 11:50 AM - Thread Starter
Member
 
DrewL's Avatar
 
Join Date: Dec 2008
Posts: 34
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Hi, I think this qualifies as a technical question :-) Ideally for a CCU operator or video engineer who works at a TV station.

The legal range of Y values for broadcast in 10-bit space is 64-940, aka "studio swing". Many HD cameras, though, take advantage of the full 0-1023 range ("full swing"). If you want to broadcast footage from them you have to legalize it at some point.

For the "real" HD cameras in the studio, do they clip at 940, or is it up to the CCU operator to adjust gain (and pedestal up) to legal levels and avoid extrusions?


I'm always finessing my knowledge of color management and have been wondering whether newer cameras just gave up on the Rec.709 spec in that respect, or acquisition in full swing has always been the intention of Rec.709.
DrewL is offline  
Sponsored Links
Advertisement
 
post #2 of 8 Old 10-04-2012, 04:47 PM
Advanced Member
 
fastl's Avatar
 
Join Date: Sep 2007
Location: Boston
Posts: 584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 7 Post(s)
Liked: 11
Anything between 3 and 1020 are considered to be LEGAL video amplitudes. VALID video levels are 16 (black) to 940 (white). Consequently, 3-to-16 and 940-to-1020 are LEGAL but not VALID, and are allowed primarily to accomodate over and undershoot effects that can occur in video processing without being flagged as illegal signal levels. They are NOT supposed to be generating viewable levels above white and below black (whiter-than-white and blacker-than-black), contrary to what all of the "gurus" around here keep claiming!
fastl is offline  
post #3 of 8 Old 10-04-2012, 09:17 PM
Advanced Member
 
Rory Boyce's Avatar
 
Join Date: Feb 2000
Location: Fair Oaks, CA
Posts: 861
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Fasti is mixing 8 bit and 10 bit data ranges. Picture black should be at quantizing step 64 as the original poster indicated. With that correction the second post is correct. No picture source should be intentionally outputting incorrect levels. Quantizing levels 00 and FF for 8 bit video and their 10 bit equivalents are reserved for serial digital timing information and are expressly dis-allowed. Fixing the video after the fact is going to be a problem if it can not be sent to something to fix it using serial digital video. Computer video boards have used zero to full scale values that would be illegal video levels in professional digital video for many years which does not matter if the video is just going to a monitor.

Rory
Rory Boyce is offline  
post #4 of 8 Old 10-05-2012, 02:47 PM
AVS Special Member
 
sneals2000's Avatar
 
Join Date: May 2003
Location: UK
Posts: 7,044
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 53 Post(s)
Liked: 47
Quote:
Originally Posted by DrewL View Post

Hi, I think this qualifies as a technical question :-) Ideally for a CCU operator or video engineer who works at a TV station.
The legal range of Y values for broadcast in 10-bit space is 64-940, aka "studio swing". Many HD cameras, though, take advantage of the full 0-1023 range ("full swing"). If you want to broadcast footage from them you have to legalize it at some point.
For the "real" HD cameras in the studio, do they clip at 940, or is it up to the CCU operator to adjust gain (and pedestal up) to legal levels and avoid extrusions?
I'm always finessing my knowledge of color management and have been wondering whether newer cameras just gave up on the Rec.709 spec in that respect, or acquisition in full swing has always been the intention of Rec.709.

My understanding is that 10bit and 8bit 709 HD follow the same level space as 601 SD - with 8bit 0 and 255 reserved for signalling, and 1-15 and 236-254 luminance levels there for over and undershoot (particularly valid when dealing with analogue SD sources - and possibly also worth remembering that there are LOTS of analogue HD triax systems in use with both Sony and Philips/GrassValley HD cameras. HD cameras still can be employing analogue links in their chains - though many people find this difficult to believe!) and to avoid ringing that would result from clipping edge transition over/undershoot.

Black is still at 16, and white is at 240 - however AIUI you should preserve 1-254 to avoid nasty surprises - though you shouldn't be putting video there intentionally. Will also the vision guys at work where they set the clippers on white levels.

8 bit and 10 bit systems in studio are designed to be interoperable and use the same 8 bit MSB level space - so the 10 bit values are based on the same levels with 2 more LSBs.
sneals2000 is online now  
post #5 of 8 Old 10-05-2012, 05:07 PM
Advanced Member
 
fastl's Avatar
 
Join Date: Sep 2007
Location: Boston
Posts: 584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 7 Post(s)
Liked: 11
...Fasti is mixing 8 bit and 10 bit data ranges.

Meant to say: Consequently, 3-to-64 and 940-to-1020 are LEGAL but not VALID..... classic example of typing faster than brain is working.

Anyway, Sony's top of the line camera (F65) has their S-curve log gamma, which is reputed to compress roughly 14 F-stops of image dynamic range into the 10-bit signal range. So they can actually capture more dynamic range than you can display, and certainly more than you transmit when "linearized".
fastl is offline  
post #6 of 8 Old 10-05-2012, 06:14 PM - Thread Starter
Member
 
DrewL's Avatar
 
Join Date: Dec 2008
Posts: 34
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
To round up:
  • Values 0 and 1023 (or 0 and 255) are reserved.
  • Values 3-63 and 940-1020 are not supposed to be seen, but (contrary to what I thought) are supposed to be carried, and may contain image information.
  • The purpose of those values is to allow for quantization errors.
  • I suppose any device in the chain may pass or clip super-blacks and super-whites, but may not 'break' due to their presence.

1. Something that Rory touched on: are we mixing up Rec.709 and SMPTE 292M? Are there different requirements and implications for invalid levels in baseband video vs. transmission?

2. As a bit of empirical verification, I hooked up my Sony EX1 camera to a Leader waveform scope via HD-SDI. The camera had no problem producing super-whites and the scope had no problem dealing with them.

3. Does anyone know why so much headroom was chosen? Does it have to do with corresponding analog IRE values?

4. Broadcasters may choose to fail QC for invalid levels. It seems that there's nothing technically wrong with those levels, they just don't want video that may look crushed due to some oversight, so they err on caution.

5. Still waiting for an answer to my original question :-)


6. fastl, not sure what the F65 and dynamic range has to do with the topic smile.gif It's a dandy fine camera, but when it comes to dynamic range and gamma curves, all 'cinema look' cameras do this. Rec. 709 only allows for 5 stops of dynamic range, pretty abysmal by today's sensor standards. The cameras compress between 7 and 14 stops of latitude into the same bit range, and it's up to the DP and colorist to extract the desired image from that.

Thanks y'all!
DrewL is offline  
post #7 of 8 Old 10-06-2012, 01:51 AM
AVS Special Member
 
sneals2000's Avatar
 
Join Date: May 2003
Location: UK
Posts: 7,044
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 53 Post(s)
Liked: 47
Quote:
Originally Posted by DrewL View Post

To round up:
  • The purpose of those values is to allow for quantization errors.
That's my understanding. The Rec/ITU 601 levelspace was decided in the very early 80s - and in those days it was designed to allow component digital devices to be used within largely analogue composite environments where there would be composite decoding and A/D conversion on input and D/A conversion and composite encoding on output. The source content may also have been pretty artefacty - as it would have been through a largely analogue production chain - edge overshoots and undershoots would have been commonplace. If you clip over/undershoots you potentially generate nasty harmonics when you convert back to analogue which would appear as ringing (HF edge artefacts) so it is important to preserve these - hence the requirement to capture levels below black (for undershoots on low level transitions) and above peak white (for overshoots on high level transitions) when you're a digital island (and these clipped transitions can also cause problems with digital processing downstream too). All very sensible decisions in the early 80s.
Quote:
3. Does anyone know why so much headroom was chosen? Does it have to do with corresponding analog IRE values?
I guess the levels were chosen based on the likely levels of over/undershoot that could be expected in a 'production' PAL or NTSC analogue composite signal (I think SECAM production was probably increasingly in PAL already by this point)
Once 601 kit (D1 VTRs, DVEs, Harrys, Paintboxes, Slidefiles/DLSs etc.) started appearing in the 80s the 601 standard became dominant - initially with parallel 656 interconnects (on 25 way D-types) then in the early 90s the SDI addition (not the original 656 serial standard which wasn't used) became widespread. Even if 656 wasn't used then 601 level space was.

I guess a level space change could have been made in the switch to HD - but there was a lot of sense in keeping to the same levels within broadcast interconnects where SD and HD signals are often used within the same studio or facility.
Quote:
4. Broadcasters may choose to fail QC for invalid levels. It seems that there's nothing technically wrong with those levels, they just don't want video that may look crushed due to some oversight, so they err on caution.

AIUI most broadcasters in the UK doing manual QC would fail you if 'real' picture information was outside the 16-235 range (8bit) but if there were brief incursions, particularly on analogue-sourced archive, then that would be deemed better than clipping and causing ringing and should pass?

Any bit of broadcast kit should preserve 1-15 and 241-254 (8 bit) AIUI for this reason.

(There are similar issues in SD with the horizontal differences between analogue and digital standards. In 625/50-land an analogue line - 4:3 or 16:9 - is 52us, and this translates to 702 samples at 13.5MHz. However a digital Rec/ITU 601 line is 720 samples long - and these 720 x 576 digital images are thus slightly wider than 4:3 or 16:9... Digital sources should really produce a 'wider than 4:3 or 16:9' image of 720x576, but analogue sources will usually arrive somewhere around 702x576 (depending on analogue blanking errors). If you then shrink a 720x576 image in a DVE - what do you do about the edges? Do you preserve the 9 samples that don't have picture content on the left and right in some sources - say a 702x576 digitised-analogue source, or do you crop the 9 samples either side?)
sneals2000 is online now  
post #8 of 8 Old 10-06-2012, 05:48 PM
Advanced Member
 
fastl's Avatar
 
Join Date: Sep 2007
Location: Boston
Posts: 584
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 7 Post(s)
Liked: 11
.....To round up: Values 0 and 1023 (or 0 and 255) are reserved. Values 3-63 and 940-1020 are not supposed to be seen, but (contrary to what I thought) are supposed to be carried, and may contain image information.....

Values 0 to 3 and 1020 to 1023 are reserved; they don't want you within 3 LSBs of the synchronization levels. 3-63 and 941-1020 are supposed to be seen, but not seen as unique levels above normal peak white, contrary to what the home theater types are often promulgating. You obviously can't go below black unless you are emitting "negative photons". Note that the Videotek legalizer has default settings of 4 and 1019, but you can set them up to 283 and down to 633.
fastl is offline  
Reply HDTV Technical

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off