Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Obama Official Portrait Color Errors?  (Read 4595 times)
mlmcasual
Newbie
*
Offline Offline

Posts: 25


« on: January 20, 2013, 05:05:03 PM »
ReplyReply

This is interesting..
I was looking at the newly released official White House Obama Portrait.
On my color calibrated monitor it looked "off" with a green cast...

I saved and opened in Photoshop- here is what I can found..  (you can do this yourself to verify)
In Histogram view, The Green and Blue channels look ok.. but the RED channel shows to have been cutoff at the top.. This missing red channel highlights is not naturally occurring and looks to have been done by a post processing method..






Originally, I thought this was the result of the photographer using the flag-white to white balance the picture via the white balance tool..
However doing that would not produce the missing red channel data above.  The only way I can find that produces that missing red channel highlights is by limiting the Levels Red Channel (or channels red channel) output.
You can easily verify with any picture this can be duplicated by cutting down the Red Channel Levels highlight output like this,







But there is no good reason to color correct doing that as you are throwing away highlight data and will cause other color cast  problems..

If you look at the flag white areas in the background, notice the whites are DEAD on neutral-
Which indicates the photo had been manipulated via limiting the red channel output until the flag set as the white neutral.



There are two problems with this.  One, is the background is lit with a different color temperature then the President (who was lit with 2 octobox strobes)
So by adjusting the entire picture for the flag white would result in the president with a green color cast (which it has)..
And second, the method of limiting the Red Channel Curve/level  is not the correct way to color adjust. I'ts never good to throwaway data.


Anyway, there really is no way to properly correct this after the fact since good data has been thrown away, but i did  mask out/isolate Obama and restore the red colorbalance and got this..
Just to give you an idea of what the skin tone should have looked like.














And to show them side by side.. though subtle...


     
Original..



Fixed..


« Last Edit: January 20, 2013, 05:12:04 PM by mlmcasual » Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6883


WWW
« Reply #1 on: January 20, 2013, 05:19:47 PM »
ReplyReply

Looking at this - I agree - the picture does appear to have a mild greenish cast; how interesting. BUT the red channel information in the first histogram does not mean there is missing red data. All it means is that the highlight end of the tonal range doesn't have red in it, which is not impossible even if the image were properly colour-balanced. If you are saying that the whole of the red channel is under-represented in the colour balance of the image, that is a possibility. I would be looking more at the concentration of red in the mid-tones rather than the highlights, especially as the all important skin tones are mid-tones. Colour-balancing images in Photoshop actually requires a channel-by-channel adjustment of black points and white points to really do it properly. It is more easily achievable in Lightroom, but we need to know what is really neutral in the image, and often we don't know this because lighting of different temperatures may hitting and reflecting off the surfaces, yielding unexpected hues relative to what one's brain says the colour should be. And getting brown skin tones correct in a photograph can be challenging because brown is a very "hybrid" mixture. I wonder whether the photographer who made this portrait used a WHIBAL, Gray Card or ColorChecker or not? One would think........
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1152



WWW
« Reply #2 on: January 20, 2013, 05:43:42 PM »
ReplyReply

You should see how many versions of this image I've found in a Google Image search having different skin tone color. Fortunately most have an embedded profile to insure intended results by who ever uploaded all those different versions.

I wonder if there are copyright issues editing this image without the photographer's permission.

BTW the edited final makes the President look way too red/orange. Here's one I found online that's a bit more realistic since I couldn't find the White House source link to the original.

http://www.eurweb.com/wp-content/uploads/2013/01/president-obama-portrait-819x1024.jpg
Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6883


WWW
« Reply #3 on: January 20, 2013, 05:55:17 PM »
ReplyReply

Yup - the one you linked looks pretty good.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
mlmcasual
Newbie
*
Offline Offline

Posts: 25


« Reply #4 on: January 20, 2013, 06:04:56 PM »
ReplyReply

No, the one you linked is untagged (no ICC profile) so can't really consider that.

As I mentioned the "fixed" one I placed will still be pushed a bit too red because the original the original highlight was thrown away (if my theory of how the photographer processed it is correct)

Quote
BUT the red channel information in the first histogram does not mean there is missing red data. All it means is that the highlight end of the tonal range doesn't have red in it, which is not impossible even if the image were properly colour-balanced.

I would have to disagree in this case.  Look at again that how just the Red Channel is missing highlights..


Where have you seen a Completely Flat Red Highlight channel like that? I never have. I would theorize that's throw away data from the red channel curve limit.
« Last Edit: January 20, 2013, 06:09:09 PM by mlmcasual » Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6883


WWW
« Reply #5 on: January 20, 2013, 06:58:02 PM »
ReplyReply

Plenty of times. Any channel can be missing highlight data if the highlight hue is not supposed to have that colour present. A properly colour-balanced image does not necessarily require the presence of data in all three channels right through the tonal range. The mid-tones are what matters most here. The essential point I think is that the image was not properly colour-balanced from the capture stage forward and that's pretty obvious without even thinking about the upmost 5% end of the tonal range in the red channel. Put otherwise, even if the image were properly colour-balanced from the get-go, it is conceivable that there could be an absence of red at the pinnacle of the tonal range. 
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
darlingm
Sr. Member
****
Offline Offline

Posts: 337


WWW
« Reply #6 on: January 20, 2013, 08:04:33 PM »
ReplyReply

I have to agree with Mark on the histogram analysis on the technical basis.  However, mlmcasual, still very nice catch, for the version of the image you're talking about!  Yours is much better/accurate.

I wonder if there are copyright issues editing this image without the photographer's permission.

Photo was taken by Pete Souza - the official chief White House Photographer.  The photo is considered public domain under US Code Title 17, Chapter 1, Section 105.  This section states: "Copyright protection under this title is not available for any work of the United States Government, but the United States Government is not precluded from receiving and holding copyrights transferred to it by asignment, bequest, or otherwise."  Section 101 defines the term "work of the United States Government" as: "a work prepared by an officer or employee of the United States Government as part of that person's official duties."  The general idea is that US government officers and employees are paid to do their work by the tax dollars from the citizens; therefore, the work belongs to the people.  If this photograph were instead taken by a private citizen, that person would then own copyright.
« Last Edit: January 20, 2013, 08:08:43 PM by darlingm » Logged

Mike • Westland Printworks
Fine Art Printing • Amazing Artwork Reproduction • Photography
http://www.westlandprintworks.com • (734) 255-9761
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1152



WWW
« Reply #7 on: January 21, 2013, 01:19:29 AM »
ReplyReply

The screengrab below is what I'm seeing of your final edit on my calibrated/profiled sRGB-ish gamut display. Those are some crazy Lab numbers for African American skin tones. It's Boehner that has the orange skin, not Obama.

Why you're pointing out technical issues with the Red channel as the cause for busted skin tones makes no sense seeing we don't know what processing was done upstream or what the original image looked like.

Do you have the official White House link to this photo?

And you're right about the lack of profile of the version I linked to but even assigning my monitor or sRGB to the image still made the skin tones look reasonably African American.
« Last Edit: January 21, 2013, 01:24:00 AM by tlooknbill » Logged
opgr
Sr. Member
****
Offline Offline

Posts: 1125


WWW
« Reply #8 on: January 21, 2013, 03:12:12 AM »
ReplyReply

It's interesting because the original screengrab posted shows the image on my monitor as very dull in color and does indeed look greenish. Whereas the standalone and "corrected" images both show very saturated colors and slightly redish in appearance. (on my monitor).

The eurweb link does indeed seem to have the most natural look (both with and without sRGB profile).

At the very least there is an issue with colormanagement in OP's screengrab?
Logged

Regards,
Oscar Rysdyk
theimagingfactory
Vladimirovich
Sr. Member
****
Offline Offline

Posts: 1320


« Reply #9 on: January 21, 2013, 01:39:25 PM »
ReplyReply

ok, birth certificate was not enough... now what ? shall I wait for Fox to start bitching about the red channel ?
Logged
Randy Carone
Sr. Member
****
Offline Offline

Posts: 539


« Reply #10 on: January 21, 2013, 05:36:49 PM »
ReplyReply

Vlad - that made me laugh out loud.
Logged

Randy Carone
Eric Myrvaagnes
Sr. Member
****
Offline Offline

Posts: 7874



WWW
« Reply #11 on: January 21, 2013, 07:14:49 PM »
ReplyReply

ok, birth certificate was not enough... now what ? shall I wait for Fox to start bitching about the red channel ?
I would have expected Fox to be trying to eliminate the Blue channel altogether.  Cheesy
Logged

-Eric Myrvaagnes

http://myrvaagnes.com  Visit my website. New images each season.
hugowolf
Sr. Member
****
Offline Offline

Posts: 612


« Reply #12 on: January 21, 2013, 09:10:39 PM »
ReplyReply

Do you have the official White House link to this photo?
At the White House Flickr account, there is an sRGB version
http://www.flickr.com/photos/whitehouse/8390033709/sizes/o/in/photostream/

Brian A
Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6883


WWW
« Reply #13 on: January 21, 2013, 09:16:44 PM »
ReplyReply

Looks fine. So how did the OPs version get so messed?
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
hugowolf
Sr. Member
****
Offline Offline

Posts: 612


« Reply #14 on: January 21, 2013, 09:27:02 PM »
ReplyReply

Looks fine. So how did the OPs version get so messed?
Looks like a conversion to AdobeRGB, stripped tag, then incorrect assignment of sRGB.

Brian A
Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6883


WWW
« Reply #15 on: January 21, 2013, 09:30:21 PM »
ReplyReply

Sounds plausible.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
digitaldog
Sr. Member
****
Offline Offline

Posts: 8782



WWW
« Reply #16 on: January 22, 2013, 09:31:44 AM »
ReplyReply

Looks like a conversion to AdobeRGB, stripped tag, then incorrect assignment of sRGB.

The Lab values for the skin at the original site look very good with a and bstar almost the same value. I think your explanation makes sense, compared to the Lab values shown earlier by Tim. For skin, Lab in sRGB, ProPhoto, Adobe RGB (1998) etc would all follow the same ratio so it looks like the ugly rendering was converted incorrectly at some point.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad