-
So widescreen NTSC is not 16:9…
I’m making DVD’s of an HD feature film – something I haven’t done in a long while. This is the first time I’m using Adobe Media Encoder (2014) to make the DVD assets (from the ProRes master).
Late into the process I discovered AME added slim yet annoying pillar-boxing to the transcode, just as discussed here. Last time I used Apple Compressor and there was never such an issue.
A bit of digging and quick math show that despite what everyone would tell you, the aspect ratio for widescreen DVD’s is not 16:9 (or 1.78:1). It’s a bit wider, at 1.815:1. So for example, Adobe’s official page on aspect ratios, which says “D1/DV NTSC Widescreen PAR 1.21:1 Footage has a 720×486 or 720×480 frame size, and the desired result is a 16:9 frame aspect ratio“, is wrong. The numbers don’t add up.
I have no idea why a 1.21:1 PAR was standardized instead of 1.185:1. It seems like AME is one of the few tools that respects it but in a world where no one else does, it does more harm than good.
1. Where has this issue been hiding until now? What do other tools do for HD to Widescreen-SD downconversions?
2. Would you set AME’s “Source Scaling” to Stretch To Fill or not? In the aforementioned discussion everyone is against it, because it “distorts the image”. But when was the last time you saw a 1.815:1 display?? I checked, even Apple’s DVD Player software uses the wrong aspect ratio of 16:9.
3. Are you aware of any TV’s or DVD’s that do the correct 1.815:1 to 1.78:1 center-cut? Because I’m not.
I’m inclined to do the downconversion with the Stretch option, and disrespect the pixel aspect ratios, since any DVD nowadays would play on a 16:9 TV, and everyone seems to have forgotten about the 1.815:1 thing.
Your thoughts?
Note: you may very well say this is off-topic. There are better-suited cow forums for this, but they’re relatively barren and I need an answer within a few hours. Sorry.
-
7 Replies
-
The common viewer is so very accustomed to seeing distorted images these days. They watch TV shows from the 70’s and 80’s in 16×9 and don’t think anything is wrong with the picture. Not to mention how they overlook the MPEG degradation of the picture over satellite or cable.
Since moving to the Creative Cloud suite we have always used the Stretch to Fill option. It is so slight that not even our most critical producers notice anything is a amiss when they review the final product.
Joe
Living the SuiteLife!
Stuff for editors http://www.cafepress.com/suitelifehttp://www.facebook.com/pages/SuiteLife/1524456414462851?ref=hl
-
The rumor I read years ago was that the BBC after saying they would
adopt AME requested that.All of my clients are annoyed by the bars added and all of them
direct me to “scale to fill” even though there is a slight loss
top and bottom.They all said they’d rather have that than stretch slightly, especially
for dvd viewers on pc/mac systems which is the only time you would see the bars. -
My point is that you’re both still wrong (as I was yesterday):
Chris, if you’re using “scale to fill” to maintain the “correct” aspect ratio (and crop the top&bottom), you’re giving your clients a slightly distorted image.
And Joe, your producers don’t notice anything is amiss because nothing is. Doing it wrong with Stretch to fill seems to be the only way to do it right…
Why? Because everyone’s display devices disregard the standards and apply an additional (or different) aspect ratio correction. If there’s a standard and all the displays ignore it in a consistent manner, well then I’ll ignore it too…
I’ll be happy to hear if I’m wrong about this. -
there’s a hybrid non-linear stretch method posted here.
https://forums.creativecow.net/thread/2/1056501
-
Thanks Chris, but not relevant to this discussion.
I’ve been doing a bit more reading to get to the bottom of this. As en engineer and online editor, I underestimated how lax the “official” SD aspect ratios were. There’s a fairly good discussion at https://en.wikipedia.org/wiki/Pixel_aspect_ratio, and I found https://toolbox.sgi.com/TasteOfDT/documents/video/lurker/pixelaspect.html“>another insightful discussion from 1999 SGI forum, albeit one that adds as much as it confuses.
So what’s going on?
Warning – this is not a discussion for beginners; it will make your head spin 😉1. Anything that has to do with SD is based on analog standards. We’re so used to live in our exact, digital HD world that we forgot the relatively lax tolerances that analog video called for. Mainly, the “active image area” was well-defined but what laid beyond greatly varied. The “entire frame” had many different sizes and aspect ratios that were never standardized! ” Consecutively, the pixel sampling frequencies in use today are industry conventions, not stated in any spec.”
2. The standards were often lacking. CCIR-601never defined pixel aspect ratios. (Ironic, since it’s called “encoding parameters of digital television for standard 4:3 and wide-screen 16:9 aspect ratios”.) It barely even mentions pixels; everything is derived from analog sampling frequencies.
The first revision of ISO/IEC DIS 2-11172 (better known as MPEG1) defined aspect ratios which “the committee seems to have made up”, disregarding Rec. 601. Rev.2 didn’t fare much better, and added “we presume that a certain degree of tolerance is allowable.”
The spec for MPEG2 (which all DVD players use) allowed for the proper aspect ratios but didn’t enforce them, leaving plenty of room for interpretation.
Lastly, SMPTE RP 187 (from 1995) was supposed to help clean the clutter, but instead made up its own impractical pixel aspect ratios, so it was ignored by the entire industry…3. What’s causing our specific pillar-box issue and AR mismatch? SD video was never meant to be 720 pixels wide(!) – but 704 pixels. The extra width (8 pixels on each side) was added by Rec.601 as a safety margin corresponding with the horizontal blanking period, meant to be buried in the overscan area. 704×480 at a 1.21:1 PAR indeed translates to 16:9!
In our HD age all this got increasingly forgotten, so DVD and TV manufacturers treat those extra pixels as “active image”, we editors and producers get upset if stuff doesn’t look right at the edges, and the aspect ratios get mangled in the process. -
if you really want to get specific:
NTSC (D1) is 720 x 486 defined by SMPTE group as D1 NTSC (ITU-R 601)
720×480 is for DVD’s and the DVC codec.
704 is implying non-square according to 10/11(standard 0.9) ratio conversion to 640 square pixels 640/480 = 4×3 (1.33 aspect ratio) or 853.33 size (704×1.2121) for 16/9 aspect in square pixels.
720×486 was the wrong par but Actual SMPTE RP 187-1995 format:
overscan once again takes 3 pixels from the top and bottom now leaving 480.This is because the pixel aspect ratio is different from the production aperture.
The picture needed to be centered after the crop for a perfect pixel ratio, called the clean aperture. Why? because of overscan. And guess what happens when codecs ignore the overscan? your pixel aspect ratio goes awry!quick tip:
for 16:9, width x PAR / height = 1.77, for 4:3, width x PAR / height = 1.33 -
Yup, I’m totally getting specific. 🙂
Thank you for mentioning RP187, the introduction in it basically sums everything up:
Most video standards documents specify tolerances for individual picture blanking edges; center and aspect ratio are only implied by nominal blanking values, and vary as the blanking edges move within permitted tolerances. Modern techniques of post-production […] demand a degree of precision not afforded by the implied definitions of center and aspect ratio.
Chris Wright wrote:
This is because the pixel aspect ratio is different from the production aperture. The picture needed to be centered after the crop for a perfect pixel ratio, called the clean aperture. Why? because of overscan. And guess what happens when codecs ignore the overscan? your pixel aspect ratio goes awry!
Exactly! Case closed. 🙂
Reply to this Discussion! Login or Sign Up