USA Today Doctored Condi Photo: Answer Here

A huge brouhaha started when Michelle Malkin brought to the attention an obviously doctored photo of Condoleezza Rice.

The images in question are as follows:
Beelzebub CondiBeautiful Condi

Drudge, a master of headlines, summarizes the situation well:
DEMON-EYES-ING CONDI: USA TODAY Caught Photoshopping Secretary of State…
The question is not whether the photographs photoshopped, as a half-brained monkey with glaucoma could tell you the answer to that. The question is: Was the photoshopping intentionally manipulated to portray Condi with evil looking eyes. An important side question, when did photoshop become an English word?

The best summarized knee-jerk expert reaction from Michelle Malkin’s site Reader Scott J. writes:
I’m a prepress manager with twelve years experience and this is my professional take on it. The ‘retraction’ claimed that they sharpened the image and adjusted the brightness, they did not. The eyes were penciled in at the pixel level by hand. (VERY sloppy, I might add.) Their ‘retraction’ is nothing short of complete bullshine.
Here are the two images, simply split down the middle and combined.


Note the forehead and color on the face. No change from left to right. Only
doctoring was done to the eyes.
However, I do not believe that expert answer is good enough. Some attempt to prove or disprove the assertions of USA Today can be done. The editor’s note did confirm that some editing was done, but the claim was it was minimal and restricted:
In this case, after sharpening the photo for clarity, the editor brightened a portion of Rice’s face, giving her eyes an unnatural appearance.
While the editor may have excluded some details when writing that comment in the interest of brevity, I will limit my attempt to reproduce the image to simple sharpening and lighting without any detailed pixel-level manipulation.

Unfortunately, the editor did not include a set of detailed steps to reproduce the exact image and frankly that’s a bit of a stretch to imagine that anyone would have written down the steps to reproduce. As such, I will have to examine the image for clues as to what types of transformations were done to the image.

First, I will resize both images to the same size. The original ‘real’ image is 450×354 pixels and the ‘evil’ image is at 180×142 pixels. Maintaining aspect ratio, I notice that the image has not been cropped. This is good as at least the details are not lost. Disappointingly, the original image was saved in a very lossy JPG format which means the blocking artifacts of JPG will be present and distracting from the analysis.

To do this, I will zoom in on the before and after images to unreasonable levels typically reserved for the late 80’s style awfully ultra-close-up TV commercials.

The first clue is the shoulders. The real Condi image looks like:





The evil Condi image looks like:






Three observations can be noted:

1. The dark blue at the edge of the shoulders is considerably darker in the evil image
2. The light blue above the shoulder is considerably lighter in the evil image
3. The amount of light and darkness adjustment is not uniform across colors of the same lightness but rather concentrates around the edges.

This type of editing is consistent with the “sharpness” that the editor notes. This does not prove or disprove the editor’s assertions that the eyes are unintentional but does give credibility that a sharpen filter was applied.

NOTE: The distortion on the evil image on the far left of the shoulder is a blocking artifact caused by JPG compression.

Further examples of a sharpen filter can be found in other areas, such as the left ear:
10-19-2017 3-44-40 PM





The lips:
10-19-2017 3-45-03 PM










The collar:
10-19-2017 3-45-20 PM










… and various other locations: the other ear, the hair, the neck and yes, even those dreadfully wicked eyes.

So step 1: Apply a global sharpen across the entire original image. Why? Obviously, every edge of the picture has been affected by a sharpening effect. Doing two forms of sharpening, a normal and a “more”. I discover the normal sharpen closely matches the sharpening applied to the evil image in every area, except those devil eyes. Actually, I think they rather look like the eyes from Star Gate SG1.








I often note that finding anomalies can be best done by breaking the images down to their various channels. In this case, I break down the image into lightness, hue, and saturation which correctly shows the eye brightness difference and the similarity in edges between the two images for their saturation.

Original, lightness, hue and saturation broke down as follows:
10-19-2017 3-44-15 PM










As you can see the image is identical except for the eyes. This can only mean one thing remains, the eyes must have been separately altered from the rest of the image. However, the type of altering could have been with a fine-toothed pixel brush (i.e. intentional altering) or with a broader “I want to enhance the eyes” brush.

10-19-2017 3-43-43 PM

Again, when looking in detail the eyes yield some of the techniques used to enhance them:
1. The dark black at the edge of the eyes is considerably darker in the evil image
2. The light of the eyes is considerably lighter in the evil image
3. The amount of light and darkness adjustment is not uniform across colors of the same lightness but rather concentrates around the edges.
4. The eyebrows post-sharpen are identical thus no additional sharpening was applied to the surrounding area.

So the conclusion: Likely a second sharpen filter was applied to only the area of the eyes. The question again becomes how much sharpening was applied? Without great precision, I applied 3 times the sharpness as the original sharpen, except only to the eye area and voila – the image now looks like the evil image. For your consideration, which of the following is the original evil image [don’t cheat]:


I’ll admit this isn’t anything of a precision science; and further, I was using Paint Shop Pro and not Adobe Photoshop as the original “evil” image, so I can’t expect exact results. However, the results are very telling on their own.

My conclusion:
1. There is no doubt the eye area has special attention in photoshop.
2. A pixel level editing level did not have to be done to obtain the evil image and frankly would have been tedious to get the correct balance between lighter and darker areas.

Was this intentional? Absolutely! Was it with intent to make the eyes stand out? Yes, for sure. Was it to make Condi look evil? Doubtful. This looks like more of an attempt to airbrush Condi by making her appear to have strong eyeliner and bright eyes, and a more refined presence from the blurred background.

This type of editing is fairly common, although I personally disdain it. If someone wants to sharpen contrast or lighten or darken an image, try and do so on a picture level rather than doing so on individual areas – it rarely looks ‘human’ and gets you into trouble.

For anyone thinking I’m defending the media, please note the Moron of the Week #5 article, and for anyone who thinks I have a thing for Bush and company, please read Why this Canadian supports the War in Iraq.

For those new to this site, while you are here, why not check out more of Samantha Burns. She’s a fantastically witty blogger and I’m the equivalent of the techie Igor around here.

Update: Another doctored photo has been found by Michelle Malkin and the original story found at Little Green Footballs.

For further reading on this topic and various opinions see:

Rhymes With Right
Mike’s Noise
Confederate Yankee: Photo Ethics Eludes USA Today
Steel City Cowboy: USA Today Uses the Sharpen Filter!
Independent Sources
The Anchoress
T. Longren
GM’s Corner
Viewpoint Journal
Little Green Footballs
Holstein Grove

Leave a Comment