Tuesday, June 24, 2008

'Data's Head' image proved fraudulent

Background
=========


In "Dark Mission" and on his web site, Richard Hoagland makes much of a certain rock in the lunar crater called 'Shorty', which was Station 4 of EVA2, Apollo 17. Hoagland presents a retouched photograph of this rock, claiming that it is in fact a technical artifact of a lunar civilization that no longer exists. In the introduction to "Dark Mission" he writes that Apollo astronauts brought back samples of the technology of this defunct lunar civilization "for highly classified efforts at back engineering" [sic, he means reverse engineering]. While not specifically claiming that the Apollo 17 astronauts Cernan and Schmitt actually retrieved the rock from Shorty, he wrote that "they could have". I have refuted that claim previously on this blog, citing five reasons why they could not have. Hoagland's claim that it is an artifact will now be examined and tested.

Photo processing
=============


I was interested in Hoagland's photo processing steps because I had seen one of the original Apollo 17 Hasselblad frames and could not understand how the "Data's Head" jpg could have been derived from it. Since Mike Bara, on the darkmission blog, had stated that this was a process that "anybody can repeat" I asked twice for guidance on the steps necessary to get from the whole frame to the highly decorated detail image. On the first occasion Bara replied that he was "not prepared to do your homework for you", and on the second occasion he replied "you don't know anything about how images are processed." Mr. Bara is neither polite nor helpful.

The image is a 7 kb file, 361 x 291 px., containing 34,509 colors (as measured by Paint Shop Pro v7.02 on a 2 GHz Hewlett-Packard Pavilion PC)

This is the information Hoagland offered, in a December 2007 interview for 'Project Camelot':

Kerry: (laughing more) You found a robot head in the bottom of a photograph...

Hoagland: In 14 photographs. It was photographed again and again and again and again as part of the panoramic sequence. We are trying now to go - we've gotten two copies of film - not just the web but film, (which is really crappy copies that were sent to us), and what I was able to do was a computerized robot comparison with C3PO.

....

I need higher resolution, but I've done some playing around on the computer even with those frames, and you get very interesting results. The two frames that were actually film that we used, we were able to superimpose them and that's when the eyes popped out.

The round irises, the camera eyes, that tell me this is not a desiccated human being lying there on the moon, one of the lunar colonists that we were positing was there at one point. This is an artificial life form, a robot.


So the "Data's Head" image was generated from two separate prints that Hoagland himself describes as "really crappy".

For my follow-up analysis, I ordered a very high definition scan from Houston Photo Imaging, a company that is trusted by the NASA photo library and can scan direct from the negative, ensuring highest possible definition. Anything but "crappy".

I don't know where Hoagland gets his count of 14 separate Hasselblad frames. Here is the complete inventory of that camera magazine -- I see the rock only on frames 21000, 21001, 21005, 21006 and 21027. I ordered frame 21000.

The scan was shipped on CD and was in tiff format, 46.1 MB, 5190 x 6175 px., 136,589 colors. It was processed as follows:

At stage 1, a rough crop was performed to isolate the skull-like rock and the feature known as "the turkey". This yielded a tiff of 564 KB, 522 x 560 px., 26,898 colors.

stage 1 ref jpeg

Stage 2 reference jpg
At stage 2, the image was rotated to match the orientation of Hoagland's image. It was discovered that Hoagland had rotated his image 45° counter-clockwise. The crop was then matched as closely as possible. The stage 2 tiff was 40 KB 181 x 145 px., 9,179 colors.

At stage 3, the image was enlarged by a factor of 2 to produce an image with the exact pixel dimensions of Hoagland's. The file was 111 KB, 18,439 colors.

Stage 3 reference jpg

For stage 4, the image was given to a trained Photoshop operator (Hoagland himself does not say what training he has had in photo enhancement. The safest assumption is 'none'.) The operator was asked to use any available image enhancement techniques to approach the style of Hoagland's manipulations. Software version was Adobe Photoshop CS2, v 9.0.2, on a 3 GHz PC running Windows XP.

- Tonality, color and contrast were altered using the 'Curves' feature. Input: 167 Output: 91.
- Shadows, midtones, and highlights were adjusted using the 'Levels' feature. Levels: 69, 1.00, 238.

The operator commented as follows:

All other adjustments to the photo, from reshaping the eyes, to darkening the shadow under the rock (and making the edges much more defined), and including adding shadow around the nose area, plus the odd color additions were made by someone, and not part of the original image.


The reference jpg is shown below, with the Hoagland original for direct comparison:

reference jpg from photoshop
hoagland original

The photoshop file was 2,254 KB, 31,869 colors. The reference jpg is 60 KB, 18,907 colors. Note that this compares with the 7 kB of the Hoagland jpg, attesting to a high degree of compression (and consequently low fidelity). It's possible that a higher-quality master file was used for the book -- however, the overall quality of image repro in the book is so appalling it hardly matters.


Conclusion
=========

It is plain that Hoagland manipulated this image in ways that go far beyond what is considered legitimate image enhancement. Despite the monochrome general appearance of the moon's surface (and notwithstanding the famous orange soil that was discovered on the rim of this very crater), there is color in these images. However, the 34,509 colors of Hoagland's image bear witness to his artificial colorization. If you inspect a de-rotated version of the 'Data's Head' image, it becomes obvious that rectangular patches have been independently colorized and then re-superimposed on the frame.

de-rotated Hoagland cheat
It's not so obvious on the rotated version, and this may explain why he performed the rotation (which has no innocent motive that I can determine).

I do not see the "camera lenses" that he says he sees, but if they are there, they are far more likely to be the result of an imperfect superimposition of his two scans rather than anything real. Hoagland's superimposition technique is dangerous because any slight change in perspective from frame to frame can make true matching impossible.

The bright red jaw-piece that has been described as "anodized" is simply not there. It's an artifact of processing, whether by design or incompetence.
I can't help wondering whether Hoagland enlisted the services of the Minnesota design studio that designed his web site -- VA Graphics. This company offers expertise in photo restoration, especially "facial reconstruction" and addition of "special effects and fantasy art". The name under which this company operates that business? Glad you asked. It's FIXYOURPHOTOS.COM. How very, very apposite.

My conclusion is that Richard Hoagland's image is fraudulent, and I call on the publisher, Adam Parfrey, to delete it from future editions of the book.