Output YUV data as Basic YCbCr not CCIR601

[Spirogyra]

I am converting DV home video, or VHS home video converted to DV by passing it through a DV camcorder, to MPEG2 for authoring to DVD. I capture by firewire from the camcorder using Studio 7 and the DV capture card which came with Studio 7. I use Studio 7 to edit, and then render an edited AVI. I use TMPGenc to convert the AVI to MPEG video (I separately split out the audio from the AVI as widely recommended), and TMPGenc Author to author the DVD.

Like many before me, I have been struggling to understand the check box in TMPGenc which says “Output YUV data as Basic YCbCr not CCIR601.” Like Lordsmurf, I am convinced nobody really knows what this setting really does, as I have read contradictory explanations and suggestions all over. The majority position seems to be that the box should be checked for DV video because the gamma range is already in the proper range for viewing on TV and not checking the box will cause it to be compressed further. However, I have also ready that for DV video the box should not be checked to avoid expanding gamma values, which are already in the appropriate range, to the wider computer scale, which will result in clipping of whites and darks when viewed on TV.

To resolve this question, I have, as suggested by Lordsmurf in his TMPGenc conversion article, rendered a DVD containing two identical clips which happen to have a lot of bright whites in them (video of the birth of my twin daughters in the hospital, with lots of white linen and bright lights), once with the box checked and once with the box unchecked, to see which would produce better results on the TV, which is what I want to watch them on. I reasoned that with all the whites, the correct setting should be apparent since the whites would either be clipped and the video would get washed out or the whites would be compressed and darkened or pastelized.

Well, much to my consternation, when played on the computer or the TV, I see no marked difference between either set of clips. They appear identical. Now I guess this is okay since I shouldn’t screw up either way, but based on what I’ve read, this should not be the case. And being in general a perfectionist, I would like to know I am using this option correctly. Can anybody offer any suggestions as to why I am seeing no difference between these clips. Is there anybody here who advocates leaving the box unchecked for DV video or is everybody of the majority opinion that the box should be checked for DV video?

Thanks in advance for your comments/suggestions.

[vhelp]

Hi Spirogyra,

I think firstly, that your (anyones) source should be true DV, and
not something converted from another source ie, cable; satellite
etc. That's the key here. To many missing variables, so I have
to assume for everyone.

I started on this a while ago, but stopped, due to other project
plans or personal life issues Anyways..

As to your 3rd paragraph (and some of 2nd) ...

If your source is something that was converted from VHS and NOT
TRUE DV, then you're almost always gonna have issues or disbelief
because of the fact that the source wasn't true DV to begin with.
For intance, you have VHS, and this is another color space, which
is converted already, then you re-convert it, ala DV - two conversions
in one, then you want to take that "tainted" DV source file ing TMPG
for a final enocde of 3 conversions (if not 4)

If there is to be a study w/ this Output YUV data as Basic YCbCr not CCIR601
then it has to be done on a genuien (spelling) pure source.. DV!!
and not vhs; cable; satellite etc converted to DV.

Now, checking this option on, during encoding will deffinately darken
your source if it was an Analog capture from vhs; cable or satellite.
But, the final asseptable outcome is dependant upon the user's
descresion. However, if your source is TRUE DV, say from a DV cam
that you took footage from (and not from vhs and passed through to
your DV cam) then you should see some good results w/ option checked

Again, w/ respect to your 3rd paragraph...

Your source was VHS for instance, and your passed it through (converted)
it to your DV, thinking it would be TRUE DV (which it ain't, but
is, in the sense because it is wrapped inside a DV spec) any colors
that are DV'ed will be filtered through, hence you won't notice anything
when you finalize it to a CDR or DVDR/RW media for TV play.
.
.
Also, don't forget that a TV has a different color space (I forget the
exact terminaology on this) but a TV play almost always results in
a similar final output quality when the source is very similar
ie, your test above. But, try that same test w/ a TRUE DV source and
you might be surprised as the results

The key here again, is color space and conversion. What was the color
space PRIOR to DV'ing the source (to DV) ie, was it vhs; cable or satellite ??
or, was it footage taken from a true DV cam. Thus..

DV source encoded w/ [x] Output YUV data as Basic YCbCr not CCIR601
---------------------------------------------------------------------------
* A) vhs; cable or satellite - will result in lighter color output
* B) footage take from TRUE DV cam - darker (or right) color output


w/ A) you're gonna have mixed results. Encode will be darker but colors
will be blotchy or poorer. Best to leave this option unchecked, under
these conditions.

w/ B) only issue will be your lighting condition ie, dark light leads to
poor results. So, you might want to uncheck this option under low light
footage from your TRUE DV'ing around.
Otherwise, definately leave it [x] checked w/ respectable lighting is
evident ie, outside footaging for intance.

I don't have a problem w/ this situation and my encodes look good weather
encoded w/ or w/out this option checked, when my source is TRUE DV.
There's more to it then just checking this option (see below) and that's
another key element to these fun Video Endeavors
But, when my source is vhs; cable or Satellite, and I use my ADVC-100, I
leave this option unchecked because of the colorspace already set in
the source

Of course, final output quality will vary from pc to pc, and from user
to user, and from OS setup to OS setup and people skills to people skills
and so on and so forth. So, do factor that into the equation because it
is very important when messing around in this area of topic and
especially if you're a newbie at it. You're gonna almost always have
some issues or questions or doubts.

From the Video Workstation of,
-vhelp 2133

[Spirogyra]

Thank you for your detailed response, vhelp. Just to clarify, since I didn't explain this in my post, although I am converting both "true" DV source, as you call it, and VHS analog source digitized by running it through the DV camcorder, the test I conducted was done on true DV source, not the VHS material, taken in a very well-lit and bright environment. I guess I might have to do some testing on that source as well.

[Spirogyra]

I guess I might have to do some testing on the VHS source as well, that is.

[lordsmurf]

I guess I might have to do some testing on the VHS source as well, that is.

All of my tests have shown it to be left unchecked on analog source (tape, VHS, broadcast). I tested again last week. Same conclusion. Leave it unchecked.

[Thorn]

While it's not the true test of usability, it might be easier ot tell the difference on a computer monitor. TVs and DVD players can have filters built it, so what looks ok on your TV may look different on someone else's.

[ronnylov]

You may get different results depending on what DV camera you use, what DV codec you use and how you open your DV files. So the correct method is to test both options and see what you get.

[andie41]

Very interesting topic: I have been curious about this option for a long time. As others have said, there are so many variables that will affect this and other options.

I tested a clip that contained snow, shadows, people with dark clothing - wide luminance range. My source was HI8 tape converted to DV with a Digital 8 camera. I took in the finished clip into VDub and captured a frame and took that into Photoshop. The test clip where I had checked the OutputYUV Data... box, had spead out the luminance values to include the full 0 to 255 range and pretty much evened out the entire range. When I viewed this on my NTSC S-video monitor, it at first appeard to be an improvement (snappier, more contrast), but upon closer inspection, it was apparent that the monitor could not actually show the detail in the shadows and highlights - the test clip without the option checked showed more detail. The increased luminance range looked good on my RGB, as it shows the entire range of values.

I also tested my DV avi, using different codecs to decompress, again into VDub, capture frame into Photoshop. The Microsoft codec exhibited more contrast, again with many pixels in the 0-12 /235-255 range, and the Sony codec showed almost no information in these ranges.

The mpeg I tested (as above) was encoded with the Sony DV codec installed, so it appears that by checking the Output YUV ... TMPGenc has forced values to the 0-12/235-255 ranges.

What does this mean? I don't know. As said previously, it depends on so many factors. For me, if it looks good on my output monitor, I am going to go with it. NTSC video is marginal quality anyway, and hides a lot of blemishes.

andie

[lordsmurf]

If it looks good on the monitor, it will be the wrong spec for a tv, and will be too dark and/or over-contrasted.

It should appear slightly muddy on a computer monitor. Not true white or black. The tv will handle that part for you, as it has a smaller range of colors and a different white/black value system.

[Perro Grande]

Great point, LordSmurf.

Darn near everything about a monitor and video card works fundamentally differently than an NTSC (Never Twice the Same Color) TV. You've got to be careful of contrast, color range, color selection (not all colors are "safe" for NTSC) and, of course, interlacing
One must be extremely careful not to optimize everything for computer screen viewing unless you're certain that it is your only target...

PS -- your site is excellent! I'm a Newbie to this forum, and this is the first I've seen it.

[andie41]

If it looks good on the monitor, it will be the wrong spec for a tv, and will be too dark and/or over-contrasted.

Precisely my point. My final (and really only important) test, is how it looks on my NTSC monitor.

andie

[Wilbert]

Like many before me, I have been struggling to understand the check box in TMPGenc which says “Output YUV data as Basic YCbCr not CCIR601.”

A while ago I performed some tests to see what this option does (last two posts of my in the following thread):

http://www.kvcd.net/forum/viewtopic.php?t=5202&postdays=0&postorder=asc&highlight=ccir601&start=48

"My conclusion is that all YUV values with Y<16 are clamped to (Y,U,V)=(16,128,128) when leaving "output YUV as Basic YCbCr and not CCIR601" unchecked."

"conclusion: the scaling (16,235)->(0,255) is performed when the option "output YUV as Basic YCbCr and not CCIR601" is checked."

So, the help balloon is not correct (it talks about [8,235], etc.)!

So, it should always be unchecked (as it is by default).

If for some reason your dv outputs [0,255], you should correct that in AviSynth by adding the line:

ColorYUV("TV->PC")

[LisaB]

A lot of talk about how the colorspace of TV's and computer monitors are different...but what about HDTV's?? If the output will be burned to DVD and viewed on an HDTV, does that enter into the argument anywhere. It seems that the people in this thread are saying that this setting is determined by the source and not the intended output device, but I just figured I'd ask anyway...