More on photograph from Mars

More detail is found in the NASA press release posted at http://www.spaceref.com/news/viewpr.html?pid=11583

-- Edward Tufte, May 25, 2003

Was ET hoaxed?

There are some indications that the image of the pale blue Earth was colorized. Anyone know details? Perhaps I have been fooled by the NASA pictures. I retract my "Powers of 10" remark until the published images are confirmed.

-- Edward Tufte, June 1, 2003

Are the Mars images a hoax?

The page http://www.msss.com/mars_images/moc/2003/05/22/ contains this note:

     A note about the coloring process: The MGS MOC high resolution camera only
     takes grayscale (black-and-white) images. To "colorize" the image, a Mariner
     10 Earth/Moon image taken in 1973 was used to color the MOC Earth and Moon
     picture, and a recent Cassini image acquired during its Jupiter flyby was
     used to color the MOC Jupiter picture. The procedure used was as follows: the
     Mariner 10 and Cassini color images were converted from 24-bit color to 8-bit
     color using a JPEG to GIF conversion program. These 8-bit color images were
     converted to 8-bit grayscale and an associated lookup table mapping each gray
     value of that image to a red-green-blue color triplet (RGB). Each color
     triplet was root-sum-squared (RSS), and sorted in increasing RSS value. These
     sorted lists were brightness-to-color maps for their respective images. Each
     brightness-to-color map was then used to convert the 8-bit grayscale MOC
     image to an 8-bit color image. This 8-bit color image was then converted to a
     24-bit color image. The color image was edited to return the background to
     black. Three separate color tables were used: one each for the Earth, Moon
     and Jupiter. Jupiter's Galilean Satellites were not colored.

The bottom of that page contains links to the unprocessed Earth/Moon and Jupiter images. This Earth/Moon picture was cut from that raw image and enlarged to 400%. The Earth appears to be less than 27 pixels in diameter. The Moon is the barely visible smudge at the bottom right corner.

-- David A. Nash (email), June 2, 2003

Where's the Moon?

A friend wrote that he had to clean his monitor and darken the room to find the Moon. In this version I used Photoshop's "replace color" command to lighten the background and reveal the Moon.

-- David A. Nash (email), June 2, 2003

Original vs published images

For direct comparison, here is what was photographed and what was published:

-- Edward Tufte, June 3, 2003

Messing with pixels

In enlarging the "what was photographed" images it appears that pixel interpolation was used to smooth out picture. There are a variety of interpolation methods (examples), each with advantages and disadvantages, but they all add extra shape and shade information that were not actually photographed. NASA also used interpolation in preparing the "published" images.

-- David A. Nash (email), June 4, 2003

ET small multiples and vertical comparisons

This is also a lesson in small multiples and showing comparisons. And scrolling down to find OJ made it the most powerful wordless editorial I have seen.

-- David A. Nash (email), June 17, 2003

Alternations to Mars photographs not mentioned in NASA public report

The main release of the photograph to the public shows only the constructed image and does not document any of the image manipulations:

http://www.nasa.gov/multimedia/imagegallery/image_feature_56.html

-- Edward Tufte, June 22, 2003

Excellent report on image altering in scientific publication

See Nicholas Wade in The New York Times on image fudging:

http://www.nytimes.com/2006/01/24/science/24frau.html?ex=1295758800&en=9b1edac4c317b9a4&ei=5090&partner=rssuserland&emc=rss

-- ET, January 23, 2006

NEJM problem

The New England Journal of Medicine this week expressed concern over two articles it published. Pictures supposedly from two patients were actually different magnifications of the same photograph. http://content.nejm.org/cgi/content/full/354/6/638

-- Prem Thomas (email), February 16, 2006

NEJM falsification

What I find most extraordinary about the example from the New England Journal of Medicine cited by Prem Thomas is how obvious the mislabelling is. The original paper was published in April 2001 and it is immediately obvious from the small size figures -- fully confirmed by looking at the enlargements -- that the supposedly different illustrations are actually the same one in two different magnifications. Yet it has apparently taken nearly five years before anyone noticed.

How does one explain this? Does no one actually look at the evidence advanced in medical journals in support of a conclusion? Or was the work considered to be of so little importance that no one bothered?

To me the fault seems so obvious that I would hesitate before thinking of fraud: it looks like several levels of incompetence -- careless preparation of the article, sloppy refereeing and editing, and inattentive examination by readers after publication.

-- Athel Cornish-Bowden (email), February 16, 2006

NEJM problem

The New England Journal image and its caption. If I had to speculate, someone else doing research on oral leukoplakia came across this. Why did it occur? The study included 150 people with disease and I'm welling to bet the authors kept the biopsy slides on more than two of them. However, the image is surely the worst reproduction I have ever seen in the New England Journal. Panel C is wrinkled and shows light dropoff near the top. I honestly don't know the details of the publication process, but is it possible the people who prepared the copy lost the third panel and masked their own error by changing magnification, and shifting and mutilating the second image?

Figure 3. Ploidy (Insets) and Histologic Findings in Two Patients with Moderate Dysplasia (Panels A and B) and One Patient with Mild Dysplasia (Panel C) (Hematoxylin and Eosin).

In each histogram, c denotes copy or copies, and the red columns to the left of the 2c (diploid) peak are internal controls. In the histogram shown in the inset in Panel B, "S phase to 4c" indicates cells in synthesis phase that are about to double their DNA content.

-- Niels Olson (email), February 16, 2006

Recognizing the commonalities in the slides

It is a long time indeed since I looked at stained slides down a microscope, but those of us with even the faintest flair for synthesising visual patterns see instantly that the "bird shape" (middle of slide B) is identical to the shape to the left of slide C.

-- Martin Ternouth (email), February 17, 2006

The Commissar Vanishes

It doesn't take high technology to alter photographs. The Soviets were masters of this technique, to further their political ends.

http://www.newseum.org/berlinwall/commissar_vanishes/vanishes.htm

On the topic of the thread. Nearly all NASA photos gathered from space or with telescopes are altered for color. Much of this is due to the limitations of the charge-coupled device (CCD), which is usually a black and white device. To make it color, either filters are used (which discard information and energy) or split by prism to multiple CCDs with different response curves (expensive and heavier).

-- Paul Parashak (email), February 23, 2006

Mars at 30X vertical

From Nature, an image with 30x vertical exaggeration:

Source: Baker, Victor R. 2005. Picturing a recently active Mars. Nature 434: 280-283.

Original image source: Head, J.W. et al. 2005. Tropical to mid-latitude snow and ice accumulation, flow and glaciation on Mars. Nature 434: 346-351.

The same page from Nature, reduced to 3x vertical exaggeration with Adobe Illustrator:

Detail of reduced 3x text:

-- Edward Tufte, April 13, 2007

Political video ad

As Time magazine photoshopped a still picture of OJ Simpson to make him blacker, so a Clinton campaign video has altered Obama's appearance.

Looking at before and after evokes for me the steps in Final Cut Studio 2 (a film video editing program) necessary to rescale the face and make the skin tone darker. Obama's shirt remains bright white in both the original and the Clinton videos. All this suggests horizontal stretching, desaturation, and contrast adjustments.

-- Edward Tufte, March 11, 2008

I agree that the skin tones of Obama are different in the two frames. But I think this observation (it's data, not a conclusion) begs two questions. Was it a deliberate attempt to make him look "blacker" as opposed to one of two options: an attempt to darken the tone of the presentation overall or an attempt to darken the mood around the Obama portion of the broadcast for persuasive purposes (i.e. something that would also be done if Obama were white). The second question is - Which is the more accurate representation of Obama's skin color? The comparison shows differences, but I could similarly ask - Is the CNN trying to make Obama more white? I seriously doubt it, but the representation of color on television screens is not a straightforward matter and is even more complicated than computer screens. Amongst television engineers a common joke on the NTSC format is that it stands for "Never The Same Color Twice". I think Obama's jacket looks washed out in the CNN clip. It's also too dark in the Hilary clip. The "truth" is probably somewhere in between. I cannot rule out that someone adjusted the color balance on the Obama segment for simple technical reasons or just to bring it more in line with the rest of the clip which is quite dark (even Hilary is darker in this clip than in the CNN interview).

A photograph is no more evidence than a note scribbled on paper. What makes a photograph evidence/data is the attempt of the photographer/editor to be objective. A scientists attempt at objectivity is what makes the note evidence. A photographer makes choices about what and when to shoot a photograph and then in the post-production process, chooses what to bring out (in days of emulsion it was hand dodging, in our digital days it could be red-eye reduction). Photographs can be evidence and they can be art. Image like words are used to persuade and convince. We see exaggeration and hyperbole used in newspapers and magazines constantly, it is not unexpected than that images may be used in a similar fashion.

The broader question for this theme then becomes; What is the aim of the photographer/editor in these cases. Is it accurate reproduction for data purposes or is it an artistic choice. To make someone look sinister we can raise the shadows around them. Dark moods are used in movies and artistic photographs all the time. Is this the reason, to make OJ and Obama look sinister? Are we too sensitive to making a black American blacker (Why is that bad? Is a "blacker" American somehow worse?). A similar darkening of tone to darken the mood around a white American would be plainly seen as an artistic choice. Given the vignetting around the OJ photo, I don't think many would believe that the photo represents OJ's skin color. Whether or not this artistic choice is appropriate for a news magazine is a very real question. Personally I don't think so. Whether or not it is appropriate for a campaign is also appropriate. Here I am not convinced that the difference was a deliberate attempt to make Obama look more black as opposed to keeping the tone of the ad more even.

The simple view that O.J. and Obama were made to look "blacker" is really just too simple. The images are darker, no doubt. Why they are darker is, like the images, rather murky.

John Walker

Disclosures: I like Obama and would probably vote for him were I an American citizen

-- John Walker (email), April 1, 2008