It has long been assumed by most people familiar with JPL's Catbox "enhancemenet" of the MGS Face image that its misleading appearance was primarily the result of excessive high pass filtering and failure to remove the vertical streaks caused by the nonuniformity of the CCD elements in the camera's sensor array. (JPL's own web page says that they used high-pass filtering on the Face and other Cydonia images.) However, my past attempts to reproduce the Catbox using available commercial software (Photoshop) have been less than satisfactory. Figure 1 below shows a normally processed version of the MGS Face image and the same version subjected to high-pass filtering.
Figure 1. Left: Face image processed by standard techniques (contrast adjustment). Right: Same image after high-pass filtering.
While the filtering does remove most of the tonal variation from the image, the Face still is unmistakable as an unusually symmetric, smoothly contoured, and raised landform surmounted by features bearing a resemblance to a grotesque face. It does not produce the rough, jumbled mess surrounded by a vague "footprint" that is the Catbox. I now believe I have found the missing steps in JPL's processing procedure, the application of two additional filters: a noise filter and an embossing filter.
My "enhancement" is shown next to the original Catbox in Figure 2. Note that both have a grainy appearance that was lacking in the version that had only been high-pass filtered.In my copy-catbox, the graininess is due tto the application of the noise filter. Of greater significance are the effects of the embossing filter. Embossing is a technique used in computer graphics that makes flat features, usually text, appear to be raised or depressed (depending on which direction the observer wishes to think the lighting is coming from). The Catbox replica is not exact. For one thing, it lacks the broad vertical band running down the middle of the Catbox caused by the nonuniformity of the camera CCD response. But notice the"wrinkle ridges" where indicated in Figure 2 and elsewhere, covering the entire landscape surrounding the Face. Notice also the "shadow" cast by the bright feature on the left in the attempted replica. The pattern of "ridges" and the "shadow" are virtually duplicates of features in the original Catbox.
And they are false.
Figure 2. Left: JPL's Catbox. Right: Attempted replication of the Catbox using Photoshop noise and emboss filters after running the image through a high-pass filter.
These "ridges" and "shadows" are at the boundaries between areas of differing albedo in the original image that do not appear to be ridges at all in properly enhanced versions of the image. The "shadow" cast by the bright feature does not exist in the real MGS image. It is purely the result of the embossing filter. No shadow should be pointing straight up toward the 12 o'clock position, implying illumination from the 6 o'clock direction. The actual direction of the sun is at approximately the 4 o'clock position -- almost directly below the "chin" region of the Face. This lighting is in large part responsible for the grotesque appearance of the Face, which becomes much more face-like from the more normal lighting angle for viewing a face in the Viking images and as simulated in the Kelly enhancement of the MGS image.
In Photoshop, the direction of illumination can be chosen to make false shadows appear at any position around a feature. Illumination from the six o'clock position (directly below the image) gives the closest match to the Catbox. The effects of embossing on text characters is shown in Figure 3.
Figure 3 The effects of embossing on a string of plain text, with false illumination at the 6 o'clock position.
The impression that the characters are raised above the surrounding surface is, of course, an illusion. As is the Catbox.
After JPL removed most of the tonal variation in the original image that gives the observer the visual cues to the real three-dimensional shape of the object, they added false visual cues to give the object its rough, jumbled appearance, inadvertently falsifying the appearance of the surrounding terrain as well.Whether JPL used a commercial product like Photoshop and its noise and embossing filters or simply abused their in-house Vicar image-processing software is not known. But the results are much the same.
The Catbox is not a "poor" enhancement, as it is often called; it is a crude but very effective fraud perpetrated by employees or contractors to the United States government. Even if the Face is proven to be completely natural, this is inexcusable misconduct and a gross abuse of power. If the Face ultimately is proven to be artificial, the Catbox will certainly come to be regarded as the greatest, most malicious, and most destructive scientific hoax since the Piltdown Man, and perhaps of all time.
--- Lan Fleming
JPL's Catbox (1998 Article)
Evaluation of Mark Kelly's Enhancement of the Face
VGL Home Page