Different color in PP CS5. Why different than Vegas and players?

F

f/0.0

Member
I just noticed that PP CS5 shows different colors than Vegas 10 and other media players.
At first, I thought it's the issue of 0-255 and 16-235 but that apparently isn't what's causing it.

The difference was much noticible when I was working on Asian lady's skin tone. On PP CS5, her skin has pinkish tone whereas it's yellowish on Vegas and other media players.

Can someone explain what the problem is?

(BTW, my monitor isn't calibrated using calibration tools. It is linked to sRGB color profile)
 
What camera is the footage shot with? Is it one of those canon h264 cameras? If so - CS5 should be treat as the "correct" version. Are for other footage, I couldn't tell ya...
 
Yes, it was shot with 5d mk2.
What I've noticed is that it's only CS5 that is showing a bit different color. When I viewed the file using Vegas, various media players, and Canon's own software (ZoomBrower), they all showed the same color.
The thing is that this difference is very subtle to the point I never noticed it until now.
 
f/0.0 said:
Yes, it was shot with 5d mk2.
What I've noticed is that it's only CS5 that is showing a bit different color. When I viewed the file using Vegas, various media players, and Canon's own software (ZoomBrower), they all showed the same color.
The thing is that this difference is very subtle to the point I never noticed it until now.
Click to expand...

PproCS5 has its own code written to access the h264 data without going through the quicktime engine as other apps do. They made sure to use and display the entire color range of the files with no clipping. Therefore treat the PproCS5 version as proper.
 
Thanks again for your input.

Yes, I do know that the new PproCS5 doesn't need quicktime engine. Same goes for Vegas 10 and these four media players that I'm using.
In fact, I uninstalled quicktime from my system to see if it'd change the situation and it's the same thing.
 
Again, here's some information about my system's settings:

Any sort of video enhancement option found in nVidia control panel: off.
Color Management setting found in Win7's control panel: sRGB IEC61966-2.1.
Codecs installed: Nothing except those that come with Premiere, Vegas, Media players, and Canon's ZoomBrowser. Even Quicktime is uninstalled.
 
Does the picture on the right look close?

659w1d.png



This was a "fake" BT.709 => BT.601 conversion (because it was done on RGB screenshot source, not Y'CbCr video)

Any sort of video enhancement option found in nVidia control panel: off.
Color Management setting found in Win7's control panel: sRGB IEC61966-2.1.
Codecs installed: Nothing except those that come with Premiere, Vegas, Media players, and Canon's ZoomBrowser. Even Quicktime is uninstalled.
Click to expand...

PS: color management within win7 's controls has nothing to do with how programs manage color. Various programs have their own management pathways that differ. For example media players will depend on both how the decoder is set, and how the renderer is set. If it's using a graphics overlay, then also the graphics card overlay settings affect the render. Most NLE's don't have color management options (in the sense of colorspace conversions or LUTS) they have a fixed pathway.

Most HD workflows are supposed to adhere to BT.709 workflow , but IIRC the 5D uses BT.601 for it's matrix coefficients (stream is flagged with BT.601). Whether or not a particular program reads the flag is another topic
 
Last edited:
Thank you so much for the explanation but I must admit I'm still very lost.

One thing I've gathered is that there's nothing I can do to fix it. Am I right on this one?
If so, how come NLE programs such as PproCS5 and Vegas show very different color? How can I start a grading process knowing what I see here isn't really what I'm going to see there? Which one should be considered as the correct one?

Since all of my videos are for computer viewing only, what would you recommand?
 
Last edited:
f/0.0 said:
Thank you so much for the explanation but I must admit I'm still very lost.
Click to expand...

Basically, virtually all cameras sensor output is in RGB, then converted to a Y'CbCr recording format. In 5d's case it's h.264 in mov. That colorspace conversion can be done in different ways, but there are industry standards. ITU BT.709 is the standard for HD. They specify coefficients for conversions between RGB<=>Y'CbCr. If you do the conversion with ITU BT.601, it will look different as shown above in the screenshots
http://en.wikipedia.org/wiki/Rec._709

Within some streams, there is metadata that tells applications how to do the conversion (the h.264 is in Y'CbCr format, and needs to be converted to RGB for display and most NLE's use RGB internally also). Wheter or not programs actually read this metadata flag is another topic (many ignore it)

The 5d streams have such metadata but notice it specifies a different matrix coefficient than color primary (ordinarily you would use the same for both), that may cause confusion in programs that do abide by the flags

Color primaries : BT.709-5, BT.1361, IEC 61966-2-4, SMPTE RP177
Transfer characteristics : BT.709-5, BT.1361
Matrix coefficients : BT.601-6 525, BT.1358 525, BT.1700 NTSC, SMPTE 170M


One thing I've gathered is that there's nothing I can do to fix it. Am I right on this one?
If so, how come NLE programs such as PproCS5 and Vegas show very different color?
Click to expand...

If you look at the screenshots, it suggests that premiere CS5 is doing the conversion using BT.709, the others are doing BT.601 (because when I did the "fake" BT.601 conversion on the premiere screenshot it matched up closely to the vegas screenshot). Note, if you decoded through quicktime API, it would look even more different (there is a gamma shift as well, not just color primary differences)

AFAIK, neither vegas or premiere have color management options to do the conversion differently (ie. you're stuck with how they do the RGB conversion) . The only way I can think around it is to do the RGB conversion in another program that allows more control and specify the matrix conversion.

How can I start a grading process knowing what I see here isn't really what I'm going to see there? Which one should be considered as the correct one?
Click to expand...

There is no such thing as "correct" or "incorrect" ; it's a bit more complicated than that because you are using different colorspaces. And there might be some who argue BT601 is correct vs. BT709 in the case of the 5D

Furthermore, when you export, unless it's in RGB, then you will be using a Y'CbCr format (all distribution formats use Y'CbCr -e.g. h.264, MPEG2 , so for flash, blu-ray, dvd everything) , and how that gets converted to RGB for display can lead to much confusion again! For example Youtube will use BT.709 (even for SD), but many media players will decode using BT.601

Since all of my videos are for computer viewing only, what would you recommand?
Click to expand...
Be more specific, on your computer, or other computers ? How other people have their decoders, renderers setup can affect how it looks. Consistency across various platforms can be difficult to achieve...

BTW, Color management is a huge topic and it's far too much to go into in 1 post... but to summarize, all the confusion usually occurs when colorspace conversions between Y'CbCr and RGB occur. How various programs do that conversion is what causes the inconsistencies in appearance
 
Last edited:
Thank you again for the great answer: I'm getting a hang of it.
As I have my background in photograhy, I thought it was a simple matter of setting and matching color profiles but now I know I was dead wrong.

At least, as I intend to use my videos for web viewing purposes only, I think such conflicts may be easy to deal with.
For that matter, I was wondering what would be the best solution for preparing videos for Youtube and Vimeo. One thing I noticed is that same video I rendered using Ppro and Vegas look different though they used same setting and codecs: Vegas generates contrasty videos unlike Ppro and the subtle difference in color remains.

On the issue of contrast, I have one media player set to use 16-235 output range because what I see in that media player matches with Youtube's way of displaying my video. In other words, I can match what I see to what others will see on Youtbe. However, based on what you've said, I guess there's no such trick to resolve this color conflict, right?
 
Last edited:
f/0.0 said:
Thank you again for the great answer: I'm getting a hang of it.
As I have my background in photograhy, I thought it was a simple matter of setting and matching color profiles but now I know I was dead wrong.

At least, as I intend to use my videos for web viewing purposes only, I think such conflicts may be easy to deal with.
For that matter, I was wondering what would be the best solution for preparing videos for Youtube and Vimeo. One thing I noticed is that same video I rendered using Ppro and Vegas look different though they used same setting and codecs: Vegas generates contrasty videos unlike Ppro and the subtle difference in color remains.
Click to expand...

How vegas handles colorspaces is another huge topic... :( Again the culprit is Y'CbCr <=> RGB conversions and how they are done . When you do colorspace conversions, how Y'CbCr is "mapped" to RGB can be done "TV levels" or "Full Range". The contrast issue in vegas is most commonly a levels issue


Basically it depends on the input format, the settings (8-bit vs. 32-bit mode, video levels vs. full range), and export format. Vegas handles different input and export formats in a different fashion (I haven't upgraded to v10 , and I probably won't, so I don't know if it holds true for v10, for example you mentioned that vegas 10 no longer decodes through quicktime?)

You can start reading these articles , and there were quite a few posts in this forum and others that deal with vegas' issues specifically (PP has it's own issues as well)
http://www.glennchan.info/articles/vegas/colorspaces/colorspaces.html
http://www.glennchan.info/articles/vegas/v8color/vegas-9-levels.htm

Also, even when you've exported a video , how youtube and vimeo handle that video depends as well...

I think the best advice is to do some small tests for your given workflow, settings, input and output formats, to see how youtube and vimeo react

I prefer more control so while I still use premiere (sometime vegas) for the actual cutting, I export uncompressed and use other programs for color management and grading
 
Last edited:
Thank you again for your great explanation! I'm really and really thankful for your support. It really helped me to understand this strange problem.
It amazes me how video people lived with and still live with such problems. I guess I'll just have to get used to it.
 
f/0.0 said:
On the issue of contrast, I have one media player set to use 16-235 output range because what I see in that media player matches with Youtube's way of displaying my video. In other words, I can match what I see to what others will see on Youtbe. However, based on what you've said, I guess there's no such trick to resolve this color conflict, right?
Click to expand...

Are you referring to approximate levels only ? It's a bit more complicated than that, because there are other settings that affect the display in terms of media players. The renderer choice plays a HUGE role and can make your video look completely different. Furthermore, you need to know input and output levels. ie.. is this a [0,255]YCbCr scaled video or is it a [16,235]YCbCr scaled video? You need a luminance waveform analysis that functions in YCbCr colorspace or YCbCr histogram to show this


f/0.0 said:
Thank you again for your great explanation! I'm really and really thankful for your support. It really helped me to understand this strange problem.
It amazes me how video people lived with and still live with such problems. I guess I'll just have to get used to it.
Click to expand...

If you're still referring to the BT709/BT601 issue for either Vegas or PP that I know of is to use other programs to convert to RGB first with the specified matrix
 
Last edited:
You must log in or register to reply here.
Back
Top