Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > MPEG-2 Encoding

Reply
 
Thread Tools Search this Thread Display Modes
Old 9th July 2002, 13:52   #1  |  Link
Pyre8
Registered User
 
Join Date: Feb 2002
Location: Canada
Posts: 62
Luminance levels illumination please

I was just wondering what the luminance level controls in CCE do. Most folks here seem to recommend 16 - 235, I have used the 0-255 without seeing any problems (but I have not tried the 16-235). Anyone has some explanation on this control and how it should really be used?
Pyre8 is offline   Reply With Quote
Old 10th July 2002, 07:33   #2  |  Link
wmansir
Moderator
 
wmansir's Avatar
 
Join Date: Oct 2001
Location: USA
Posts: 1,919
I'm not an expert, but from what I recall the 16-235 range is for material that is ment for TV, since they can have problems displaying the extreme ends of the brightness spectrum.
wmansir is offline   Reply With Quote
Old 14th July 2002, 00:51   #3  |  Link
daveidmx
Just One Calorie
 
daveidmx's Avatar
 
Join Date: Oct 2001
Posts: 112
in NTSC-Land USA a luminance value of 0 corresponds to a video voltage of 0 IRE. A value of 16 corresponds to 7.5 IRE. 7.5 IRE is considered "true black". The NTSC system reserves 0 IRE for "blanking" and calls 7.5 IRE "setup" -- two of the video signal timing components. Luminance values less than 7.5 are "blacker than black" and may get lost on many video systems, resulting in muddy image. at worst they could be confused for video timing signals, though i've never seen it happen.

Ideally, your final MPEG should have a black level of approximately 16 (out of 256). Two ways this can happen. if you used "PC Scale" in DVD2AVI, then your decoded video will have a black level of 0. you should then use "16 to 235" in the encoder to compress the luma back to spec. If you used "TV Scale" then your video will already be 16-235 and you should select "0 to 255" to prevent further luma compression.

If your source was from somewhere other than DVD2AVI you can use virtualdub to do a histogram sample of the whole movie and look at the black level and see if the graph rolls off below 16 or not.

unfortunately, this is all by the "standard". DVD players are _notorious_ for failing to conform to the somewhat vague standards. Some DVD players will expand the luma range pushing the black level to 0 IRE believing that it will make their DVD player look better on uncalibrated systems. Some DVD videos have been mastered with blacks extending below 16 luma, again attempting to produce "richer" video. So, despite the fact that there _should_ be an absolute "right" and "wrong" way to select the luma range, i don't think there is. i try to do all of mine "properly" but you may have better visual results by fiddling.

PS if you do any video brightness/contrast/gamma correction then calibration is a moot point and you should select the box based on fitting the luma range where you want it.

PPS NTSC in Japan does not use 7.5 IRE Setup and defines black level as 0 IRE, so this is all different there.
daveidmx is offline   Reply With Quote
Old 14th July 2002, 23:56   #4  |  Link
Pyre8
Registered User
 
Join Date: Feb 2002
Location: Canada
Posts: 62
Thanks, daveidmx, for a thorough informative response. I do use the TV scale in DVD2AVI and use 0-255 in CCE (without understanding it) so I was okay. The last one I used 16-235 and yes, it looked a little washed out. Thanks for the 255 IRE illuminance
Pyre8 is offline   Reply With Quote
Old 17th April 2003, 20:53   #5  |  Link
Jeff D
Registered User
 
Join Date: Feb 2003
Posts: 152
Diggin' up this old thread because I need clarification...

I've captured an indy movie with vdubVCR and did nothing special with the levels during the capture. All default values of 128 for brightness, contrast, color, etc.

The screenshots in the attachment are noise filtering tests, a composite of 4 different methods, no filter, stmedian, convolution3d and stmedian+convolution3d.
The soure has been modifed with the following AVS parts:
telecide
decimate
crop
Tweak (sat down and brightness down)
BicubicResize
AddBorders
AssumeFPS
ConvertToRGB

Screen shot 1: AVSFiltered.jpg
The output of the AVS script

Screen shot 2: CCEncoded.jpg
The image from after run through CCE.

The difference between the two image is that CCE screenshot appears to be brighter than source screenshot. If you look a the color values of the letterboxing with Photoshop's dropper tool, you'll see source screenshot (R:00, G:00, B:00) and that CCE screenshot has (R:16, G:16, B:16). These numbers seem to reflect the settings in CCE for Luminance Level (16 to 235).

I used Luminance Level of 16-235 becauase of the guides suggesting this. Is this not the correct setting?

Since my source was NTSC (and not computer source, except for the letterboxing level (0,0,0)) I can't see why I should clamp the Luminance Levels to 16-235, they should already be in this range because of the source being NTSC capture.

My other concern is the tweak step in the AVS file. I need to tweak to get the levels closer to "normal" values. I could see this having a negaitve effect on the "blacker than black" level.
Attached Files
File Type: zip images.zip (80.6 KB, 107 views)

Last edited by Jeff D; 5th May 2003 at 20:18.
Jeff D is offline   Reply With Quote
Old 18th April 2003, 04:57   #6  |  Link
Jeff D
Registered User
 
Join Date: Feb 2003
Posts: 152
(I hope the attachment is released...)

I did another check, this time I took 4 screen shots:
1) AVS Script addborders of (r0,g0,b0) without CCE encoding (file AVSfiltered.jpg in prev attachment)
2) AVS Script addborders of (r0,g0,b0) with CCE encoding Lum Level 16-235 (file CCEencoded.jpg in prev attachment)
3) AVS Script addborders of (r16,g16,b16) with CCE encoding Lum Level 16-235
4) AVS Script addborders of (r16,g16,b16) with CCE encoding Lum Level 0-255

Results were checked by comparing a single pixel in a single frame in all four clips. Results are RGB values from photoshop. I actually compared pixels in two locations, the pixel are the columns

Pixel 1 Pixel 2
1) 39,35,34 41,43,40
2) 49,50,45 59,59,59
3) 37,40,39 43,45,47
4) 25,28,22 38,41,38

Now, I understand compression is going to play a factor, but looking at the general level of the pixel you can see the change.
1) call this the control
2) Way brighter
3) Slighlty brighter
4) Slighlty darker

Another interesting point, the border color was set to 16,16,16 in clip 4, yet the CCE encoded screen shot has the borders at 0,0,0. CCE has made a big adjustment to the overall brightness or contrast of the clip.

Do these values make sense? And if so, how should I set up my avs script for processing to keep the level about the same? Option 3 looks the closest.

Last edited by Jeff D; 18th April 2003 at 05:00.
Jeff D is offline   Reply With Quote
Old 19th April 2003, 01:04   #7  |  Link
daveidmx
Just One Calorie
 
daveidmx's Avatar
 
Join Date: Oct 2001
Posts: 112
What tool are you using to grab stills from the mpegs to import to photoshop?

I think that pretty much makes sense. Unfortunately, I can't see your attachments to verify.

Question, though. When you say "brighter" and "darker" do you refer only to black levels or to white levels also? Could you take simultaneous samples of high-brightness points, also?

Also, do a histogram of your video with all letterboxing removed so its just the image. Where is the low peak?

What is your desired output? DVD to play on set-top I assume?
daveidmx is offline   Reply With Quote
Old 19th April 2003, 02:15   #8  |  Link
Jeff D
Registered User
 
Join Date: Feb 2003
Posts: 152
Tool used to grab stills is virtualdub, save image sequence as tga files. I believe these are uncompressed targa files, but I'm not sure.

I'll see if I can find someone to approve the attachment for downloading.

I have not done a histogram on the files, that's an interesting idea to try.

Brighter and darker, black level, white level, etc. it's an over-all image brightness.

what would simutaneous samples of points be, and done how?

Desired output is DVD for set-top playback, not PC playback.

Thanks.
Jeff D is offline   Reply With Quote
Old 19th April 2003, 11:39   #9  |  Link
daveidmx
Just One Calorie
 
daveidmx's Avatar
 
Join Date: Oct 2001
Posts: 112
what i meant by "simultaneous" was really just taking white-level measurements from the same images as you took black-level ones from.

strange, i'd expect these settings to affect contrast, not brightness, but oh well.

So you're doing DVD streams so they're MPEG2, right? i thought VDub could only decode mpeg1. how are you importing the streams?
daveidmx is offline   Reply With Quote
Old 19th April 2003, 17:16   #10  |  Link
Jeff D
Registered User
 
Join Date: Feb 2003
Posts: 152
MPEG into Vdub with dext's vdub mod
http://es.geocities.com/dextstuff/#VirtualDub

I guess attachements aren't working at the moment, i'll try to PM you the zip later.
Jeff D is offline   Reply With Quote
Old 28th April 2003, 07:04   #11  |  Link
adam
Registered User
 
Join Date: Oct 2001
Posts: 450
So no one else is interested in this eh? I've puzzled over this old post in the past, and have done quite a bit of reading on the subject and have concluded that basically every DVD Ripping guide on the net is incorrect when telling you to use 16-235 during encoding.

The source is already 16-235 on the DVD. With DVD2AVI set to either YUV decoding or RGB at TV scale it keeps this luminance range. When you set CCE to 16-235 you are not setting the output to 16-235, you are further compressing the luminance values. The result is less than true black and less than true white. All you are doing is decreasing the dynamic range of your colors. You'd only want to do this with your own, non-DV, captures which have captured "blacker than black" and "whiter than white" colors, or if you stretched the luminance scale by using pc scale in dvd2avi.

So when encoding DVD sources in CCE you should really be setting the luminance level to 0-255 to prevent further compression of the luminance scale, because its already correct as is.

Jeff D, as far as your situation, I don't capture at all but I think your findings are exactly the same as mine. If your source is 16-235 you want to encode at 0-255, and vice versa.

Are there anymore people who can verify this? If this is correct, than the guides and programs like DoCCE4U need to be changed.

http://web.archive.org/web/200202201...anges0400.html

Here is a good article on the subject. It definitely seems to suggest that a source encoded with 16-235 range should be compressed with the 0-255 range, not the 16-235 range.

Last edited by adam; 28th April 2003 at 07:11.
adam is offline   Reply With Quote
Old 28th April 2003, 15:13   #12  |  Link
digitalman
Registered User
 
Join Date: Feb 2003
Posts: 192
Thanks for the info everyone!!! Changing the luniminance level made a noticeable difference in my encodes!!!
digitalman is offline   Reply With Quote
Old 28th April 2003, 17:43   #13  |  Link
Jeff D
Registered User
 
Join Date: Feb 2003
Posts: 152
I gotta agree with adam's statements above.

It really makes sense if something is already configured for NTSC playback you shouldn't need to change it.

The key for my was adam's description of compressing the luma level. That's what's going on, but I never thought of it that way. Why take a full NTSC range and compress to 14/16 of what it was?

Here's a fun test...
take a small clip run it through CCE with the luma level of 16-240.
Use the output clip to run through CCE again and again pick 16-240.
Repeat this a few times, I'm sure the picture will end up at solid middle grey after a few passes, 8 at max.
Jeff D is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

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

Forum Jump


All times are GMT +1. The time now is 21:31.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.