FORUMS: list search recent posts

ATSC Broadcast Color Spec

COW Forums : Broadcasting

<< PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
Angelo Lorenzo
ATSC Broadcast Color Spec
on Mar 2, 2011 at 1:25:04 am

Out of curiosity, since the switch to digital broadcasting, is the broadcast community still recommending Rec.709 with 16-235 levels or since the picture itself isn't part of additional data transfer, that you've seen Rec.709 with full swing 0-255?

Out of safety/ignorance I've been conforming all my tv work with 16-235.


Return to posts index

Andrew Rendell
Re: ATSC Broadcast Color Spec
on Mar 2, 2011 at 11:49:27 am

I don't know (interesting thought though).

At the end of last year I had to provide 2 versions of the same items both on tape and electronically, so I made one set (the ones going to tape) normally, i.e., 16-235, then made the electronic versions and encoded them to the specs I was given. As the specs didn't mention the video levels I didn't alter them, so they got 16-235, and they passed tech review. So I guess I'll stick with 16-235 until told otherwise...


Return to posts index

Chad Brewer
Re: ATSC Broadcast Color Spec
on Mar 3, 2011 at 1:18:26 am

Maybe Dave LaRonde (who I know from reading here is well versed in FCP) or someone else can help explain. These color issues between different NLE's have messed with my mind from time to time.

We are an FCP based facility and I know very little about AVID.
Every time a client gives us a Quicktime from their AVID in rec.709, it appears to be mapped over 16-235 and black/white levels are wrong coming out of FCP, wrong on all the scopes regarding what digital black and white should be, basically it's washed out as black comes in around 10IRE and whites barely hit 100IRE. If they export from AVID with RGB instead of rec.709, then the levels seem to be on (or at least close enough) for outputting from FCP. I could be wrong, but it seems to me that FCP is based on full swing 0-255 because everything including graphics that adheres to that space is proper coming out of FCP, verified on scopes. FCP plug-ins that allow RGB value inputs are also using 0-255 scales.

So, are the color specs really different between different NLE's or is rec.709 truly 16-235 across all HD video and FCP has an internal conversion algorithm that maps it out as 0-255 when it's really 16-235? My scopes tell me that 0x0x0 is 0IRE and 255x255x255 is 100IRE...

Chad Brewer
Senior Broadcast Videotape Operator
TeleVersions, LLC


Return to posts index


Angelo Lorenzo
Re: ATSC Broadcast Color Spec
on Mar 3, 2011 at 5:58:32 am

I could be wrong as I don't cut with FCP, but it sounds like it MAY be their gamma shift issue. I conform/color correct all of my footage with After Effects on a calibrated monitor because it's an ICC color managed application.

I found this on another forum:
FCP assumes that you are working on a 1.8 system and therefore need a gamma adjustment to match the 2.2 broadcast gamma. So FCP automatically adjusts the gamma for certain imported footage. The rules are as follows since FCP 5.1.2:

- Quicktime movies that use Y'CbCr codecs (like Uncompressed, DV, etc.), get no adjustment (assumed to be 2.2 from the start).
- Quicktime movies using RGB codecs (like Animation) and still images (like TIFF) use the setting in Imported Still/RGB Video Gamma (either Source, 1.8, 2.2, 2.22 and Custom). This is kind of new in recent versions (I can't remember when it was added), but in the past, all RGB video codecs got an automatic adjustment of 1.22 (got brighter). Now it's controlled by the same setting as for still images.

Things were more complicated before FCP 5.1.2 where there rules were different and files were also adjusted on export (for example TIFF would get a .824 gamma shift - made darker). This doesn't happen anymore.

You can test the whole thing by creating a gradient in FCP and viewing it with the Waveform Monitor in the Video Scopes:

- Create a grad, view it in the scope and you'll see it as a straight line going diagonally.
- Export the grad as a TIFF. Import the image back into FCP with the Imported Still gamma set to Source. The grad will now appear brighter and the scope will show a bump in the mid tones duo to the applied gamma adjustment of 1.22.
- Now change the gamma for the imported clip (cmd + 9 on the clip in the sequence). Under Format > Gamma change it to 1.8. Nothing will change because any still imported into FCP is assumed to be set to 1.8.
- Now change the gamma to 2.2 and you'll see it revert to the original grad with the same straight slope in the scope.

There's more to this once you start adding Motion (which again assumes that everything coming in is 1.8 and exports everything as 1.8).

Working for TV and need to be sure of your content? Use a broadcast monitor. Only then can you spot all these differences and will you know exactly what will be displayed on a TV.


Return to posts index

Andrew Rendell
Re: ATSC Broadcast Color Spec
on Mar 3, 2011 at 10:18:41 am

Chad So, are the color specs really different between different NLE's or is rec.709 truly 16-235 across all HD video and FCP has an internal conversion algorithm that maps it out as 0-255 when it's really 16-235? My scopes tell me that 0x0x0 is 0IRE and 255x255x255 is 100IRE...


It's nothing to do with different NLEs, although some will modify inputs and outputs differently to others. Black=16 and White=235 is correct. Quick guide to rec.709 here: http://en.wikipedia.org/wiki/Rec._709
(IIRC the full spec is behind a paywall)


Return to posts index

Andrew Rendell
Re: ATSC Broadcast Color Spec
on Mar 3, 2011 at 10:31:02 am

Slight revision my last post - I've just found the ITU-R 709 official recommendations here: http://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.709-5-200204-I!!PDF-E.p...


Return to posts index


Chad Brewer
Re: ATSC Broadcast Color Spec
on Mar 4, 2011 at 1:23:31 am

Angelo/Andrew,

Thanks for the responses. I don't believe it's a gamma issue at all as I've done color correction on these types of clips where gamma never had to be corrected, only lifts and highlights were translating incorrectly.

I've previously read the rec.709 specs in detail and I'm trying to gather how FCP handles the values. What I'm looking for clarification on is whether the video signal values are 16-235, but FCP has been designed to handle them/map them to 0-255 like computer graphics software, e.g. it behaves consistently like a program like Photoshop. Some say 0-15 is reserved for super-black and 236-255 is for super-white. If that is the case, then in FCP super-black and super-white are either below 0 or above 255 as translated on this scale...

So, is it just an issue of how FCP is designed to interpret what is always 16-235 for black/white in the video signal? Seeing the differences between AVID and FCP is what has made me curious about this.

Chad Brewer
Senior Broadcast Videotape Operator
TeleVersions, LLC


Return to posts index

Andrew Rendell
Re: ATSC Broadcast Color Spec
on Mar 4, 2011 at 10:57:15 am

As far as I'm aware, FCP doesn't change the levels unless you tell it to (the difference with Avid is that Avid puts that option in your face when you import, whereas FCP doesn't), so where you have footage that comes from a video camera the levels will be as the broadcasting spec and signals over 235/100% are essentially overexposure (giving you the chance to correct it rather than just clip it).

As an experiment, I've just imported an image from photoshop and peak level is at 255, so FCP isn't changing anything automatically.


Return to posts index

Rafael Amador
Re: ATSC Broadcast Color Spec
on Mar 4, 2011 at 3:05:25 pm

[Angelo Lorenzo] "Out of curiosity, since the switch to digital broadcasting, is the broadcast community still recommending Rec.709 with 16-235 levels or since the picture itself isn't part of additional data transfer, that you've seen Rec.709 with full swing 0-255?"
8b Digital Video is always mapped at 16-239 (Luma 0-100%).
There is room for highlights and "SuperBlacks", so you will need to adjust those (and Chroma and RGB) to the very specs of your distribution system.
rafael

http://www.nagavideo.com


Return to posts index


Chad Brewer
Re: ATSC Broadcast Color Spec
on Mar 4, 2011 at 7:13:00 pm

Hey Rafael,
So, can you offer answers to my posts in this thread regarding how the signal appears to be mapped from 0-255 in FCP. For example, if I choose a color in FCP that is 255x255x255 it is 100IRE on the scope where 235x235x235 is not 100IRE. Is this just how FCP happens to map the signal?

Thanks.

Chad Brewer
Senior Broadcast Videotape Operator
TeleVersions, LLC


Return to posts index

Chris Wiggles
Re: ATSC Broadcast Color Spec
on Mar 4, 2011 at 8:42:54 pm

It appears to me, at least I have always assumed, that FCP displays much as QT displays video on playback: it expands the range to 0-255. But that the actual video is 16-235.

So for instance if you just pop a DVD in, or a test DVD, your computer will re-map that on playback by default to the 0-255 range, thus clipping below 16 and above 235 in the video.

On PCs, depending on video card and rendering options, you can actually maintain the video playback at 16-235 with no re-mapping. I've never tackled this question on a Mac.

Because you can clearly excurse beyond 100%, and below 0%, and maintain that properly with output, which is impossible to do if FCP is actually doing anything in 0-255 beyond simply expanding the range while playing it back for you so it matches the desktop and everything else (which is all 0-255).

I am going to have to do some digging into this though, because I would hope there is a way to maintain 16-235 display properly for correct monitoring of the ACTUAL video, not the re-mapped/clipped display of it.


As far as Rec709 and broadcast video: video is 16-235, period. You NEVER want to be mastering video with black at 1 (or 0). Reference black is at 16. You should have nothing of image significance below 16. Nominal reference white is 235. However slight excursions beyond 235 for either heavily saturated colors or specular highlights and things like that are not a problem. If you are mastering at graphics levels of 0-255, then anyone that is handling video properly will have severely clipped blacks because your blacks will be submerged well below the black-level of an accurate playback system which will be correctly aligned to digital 16 for black. And god forbid you master to 0, and then that gets expanded to graphics levels, an entire 15 steps of black details in your content will be clipped off permanently forever and lost altogether.

Regards,
Chris


Return to posts index

Angelo Lorenzo
Re: ATSC Broadcast Color Spec
on Mar 5, 2011 at 2:39:37 am

Chris, the latter half of your post answered my question perfectly. I actually did some additional digging and found that digital still uses the value of 0 and 255 for sync pulses.

Like I mentioned in a previous post, I use After Effects to conform my footage as it's color managed. It allows up to 253 for spectral highlights.


Return to posts index


Rafael Amador
Re: ATSC Broadcast Color Spec
on Mar 5, 2011 at 5:24:11 am

[Chris Wiggles] "It appears to me, at least I have always assumed, that FCP displays much as QT displays video on playback: it expands the range to 0-255. But that the actual video is 16-235.

So for instance if you just pop a DVD in, or a test DVD, your computer will re-map that on playback by default to the 0-255 range, thus clipping below 16 and above 235 in the video."


Video signal is 0-255. Is the luminance that have to be kept between 16-235.

On display, the 100% Y' value (b235, 100 IRE) must be mapped-back to 8b RGB as 255x255x255 = Pure White.
Everything on top of 100 IRE is clipped as white too because we have not more RGB values to represent them.
With 10b RGB monitors this is not like that.

DVDs do not allows this head-room for super-blacks-whites. When you making the MPEG-2/DVD compliant, all values below 0% or over 100% gets clipped.


[Chad Brewer] "All in all, I'm not versed at all in AVID, but when my clients export Quicktimes from their AVID with rec.709, the levels are wrong in FCP. When they export with RGB, black is black in FCP."
As Steven Jobs would say: " Mapping is a world of pain".
Each NLE, Graphic application, and some codecs seems to fallow his own rules.
rafael
rafael

http://www.nagavideo.com


Return to posts index

Tim Ward
Re: ATSC Broadcast Color Spec
on Mar 4, 2011 at 10:04:31 pm

Nice article/exchange about DV, super-black, super-white, etc.

http://www.larryjordan.biz/articles/lj_white.html


Return to posts index

Chad Brewer
Re: ATSC Broadcast Color Spec
on Mar 5, 2011 at 12:53:07 am

Thank you all for chiming in on this thread! The last few posts have cleared things up for me as well as support what I believed was going on all along in FCP.

Despite not starting the thread, some of these issues had been bothering me for a while. Being an FCP only editor at an FCP based facility, it first started bugging me when my clients' exports from their AVID's didn't have proper black/white levels when brought into FCP. Furthering the madness was how I now know FCP handles levels - between either FCP or the corresponding AJA, Blackmagic, etc. capture cards, the output signal is in fact proper 16-235. What was confusing things the most for me is that within FCP, all of its channel controls go from 0-255 like computers/graphics, instead of 16-235 which is correct for broadcast video. So, after having understood all of this for years in terms of video signals, it was mostly how FCP "represents" things within itself as a program that was throwing me for a loop.

All in all, I'm not versed at all in AVID, but when my clients export Quicktimes from their AVID with rec.709, the levels are wrong in FCP. When they export with RGB, black is black in FCP. I still can't figure that out, but at least I now know what's going on with 16-235 in FCP outside of that issue.

Thanks guys.

Chad Brewer
Senior Broadcast Videotape Operator
TeleVersions, LLC


Return to posts index


Rafael Amador
Re: ATSC Broadcast Color Spec
on Mar 5, 2011 at 6:09:57 am

Hi Chad,
When we bring 8b RGB to that sequence, 0-RGB (Black) should be 16 (0 IRE > Y' = 0), while 255-RGB (White) should be 235 (100 IRE > Y' = 100%)).

The problem is that FC behaves in different ways depending where the graphics come from:


a) If you are importing the graphic (PS, PNG,..):
- By default are mapped to 16-235.
- But if you set the famous "SuperWhites" option, the mapping will be to 16-255.
Make a test with a white still from PS and see how the Waveform changes when you check/uncheck "SuperWhites".

b) If you are generating the White in FC: That will depends on the tool that is generating the white.

- The two Color Mattes, now works properly, mapping 0-255 to 16-235. Those values doesn't change whether your sequence is set to SuperWhites or no.

- The TEXT GENERATOR/ BORIS: Generates 0-255 RGB values that will be mapped depending if the sequence is set to SuperWhites or not.

[Chad Brewer] " if I choose a color in FCP that is 255x255x255 it is 100 IRE on the scope where 235x235x235 is not 100IRE."
Those values you are setting, are RGB. They haven't been been mapped yet.
The secret is the SuperWhites.
Your 235-235-235 will be pure White (100 IRE) as soon as you change your sequence to "SuperWhites".
So if you use in Boris-3D the default 234x234x234, that will be Gray on a normal sequence.
If you change to "SuperWhites", you will see that the Gray becomes White on your scope.

Don't miss this:
http://www.kenstone.net/fcp_homepage/video_levels_nattress.html
The graphics for the RGB-YUV mapping are very clear.
rafael

http://www.nagavideo.com


Return to posts index

Chad Brewer
Re: ATSC Broadcast Color Spec
on Mar 6, 2011 at 3:32:45 am

Rafael, thanks for the further helpful info. Much appreciated!

Chad Brewer
Senior Broadcast Videotape Operator
TeleVersions, LLC


Return to posts index

<< PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
© 2017 CreativeCOW.net All Rights Reserved
[TOP]