Part of Viking image 70a13 showing “Face” at
Cydonia. Contrast was adjusted separately on sunlit and
shadowed (outlined) sides to bring out details of both at
comparable lighting levels. Bright border of outline is an
artefact of brightening everything inside the outline
The MGS spacecraft took a high-resolution photo of the “Face on
Mars” in April, 1998. That image suffered from four handicaps: a low
viewing angle; a low Sun angle from the direction under the “chin”; an
almost complete lack of contrast; and enough cloudiness to scatter most of
the light and eliminate shadows. To add to these difficult circumstances,
JPL-MIPL personnel, apparently judging that the controversy over
artificiality would not be ended when the actual photo was released,
processed the image through two filters having the effect of flattening
and suppressing image details. This step is documented at a JPL web site.
Here we do image processing correctly and present the results of computer
corrections to compensate for the poor lighting and low viewing angle. The
actual image shows clearly the impropriety of the JPL-MIPL actions because
the visual impression of artificiality persists. However, appearances
after a discovery are not a valid basis for drawing conclusions, but only
for forming hypotheses for further testing. This is called the a priori
principle of scientific method. The 1976 Viking imagery allowed the
formation of competing hypotheses, natural vs. artificial origin, and
tests to distinguish them. When applied to the high-resolution MGS
image of the Face, all artificiality predictions were fulfilled despite a
lack of background noise. The combined a priori odds against a
natural origin of the Face on Mars are 1021 to 1.
The “Face” at Cydonia on Mars is shown in 1976 medium
resolution Viking spacecraft image 70a13 in Figure
1, and in the Mars Global Surveyor (MGS) spacecraft
strip-image SP1-22003 in Figure
2. The mesa is about 2.5 km tall by 2 km wide, and extends several
hundred meters above ground level. The appearance is much less face-like
in the high-resolution MGS image in Figure
2 than in the original Viking image in Figure 1 for the
(1) The MGS spacecraft took its image from a low-perspective
angle well to the west, rather than from nearly overhead as in the Viking
spacecraft view. Mainly the western half of the “Face” is seen in Figure 2, with the
eastern half largely hidden behind the nose ridge.
(2) Sunlight shines on the Face mesa from the low west in the
image, but from the low southeast in the MGS image. The latter
tends to distort facial features, much like a flashlight held under the
(3) The Viking image had a normal variation of grayscale levels
to provide contrast between adjacent features. The range of grayscale
levels in the MGS image was inadequate to provide the amount of
contrast normally utilized by the human eye.
(4) Following analysis, it became apparent that the major face-like
features on the mesa have the characteristic that they cast shadows that
enhance the face-like appearance at almost any Sun-angle. For example, the
eye socket is a depression that contains the shadow of its walls while the
Sun is anywhere but overhead. It is similar for the mouth feature, which
casts a shadow into the ravine between the lips at most times of day. The
facial appearance is enhanced by such shadows, but is difficult to
separate from the background when the shadows are absent. By bad luck, the
sunlight was so scattered by thin cloud cover that light on the Face was
mainly ambient (omni-directional, shadow-free) light. This partially
ameliorates difficulty (2), but creates a greater problem by removing one
element important to the perceived appearance of the mesa.
2. Part of MGS image SP1-22003 showing “Face” at
Cydonia. Inset locates “facial” features. Contrast is adjusted
separately on the two sides, with the sunlit portions outlined.
Dark border of outline is an artefact caused by darkening
everything inside the outline to bring out its details.
Photographs of actual human faces and of face sculptures taken
under similar viewing perspective and lighting conditions as prevailed for
Figure 2 are commonly no
longer recognizable as faces. The image in Figure
2 initially leaves the question of the degree to which the mesa
resembles a face unresolved. Various features of Figure 2 can be cited on
both sides of the issue.
for the objectivity that scientists are supposed to maintain, the Jet
Propulsion Laboratory (JPL) apparently was unhappy that the
high-resolution image received by its spacecraft did not immediately
settle the artificiality controversy. Strong public statements ridiculing
the “Face” and the serious scientific investigation thereof had
previously been issued by certain scientists working for JPL, Caltech
(which owns JPL), and JPL contractors, and by other supporters of robotic
space exploration (managed and controlled almost exclusively by JPL) over
manned space exploration (for which little science or funding goes to JPL).
Indeed, the laboratory and MSSS, its contractor for the MGS imaging
mission, initially refused to take the high-resolution images of the
“Face” on the stated grounds that it would be a waste of public funds
and a slap at the integrity of the scientists in the program. They were
ordered to take them anyway by NASA Headquarters.
the first picture arrived at JPL, its Mission Image Processing Laboratory
(MIPL) passed the image through two filters, a low-pass filter and a
high-pass filter. It is difficult to see how usage of these filters on
this image before release to the media could be scientifically justified.
Indeed, usage of the high-pass filter gave an especially damaging
impression. From Adobe’s Photoshop software, we find the following
description of the function and purpose of this filter:
Pass Filter: Retains edge details … and suppresses the rest of the
image. … The filter removes low-frequency detail in an image … The
filter is useful for extracting line art and large black-and-white areas
from scanned images.”
usage of these filters on the “Face” image is documented on the JPL
web site <http://mpfwww.jpl.nasa. gov/mgs/target/CYD1/index.html>.
The same day that the raw spacecraft image data was received at MSSS and
posted to the Internet, the JPL Public Information Office (PIO) released
the MIPL-created, filtered image shown in Figure
3 to the world media.
High-pass-filtered “Face” image released by JPL to the world
a direct consequence of this act, it has become extraordinarily difficult
to get material on this subject considered in the scientific community.
For example, a technical abstract on the subject of Cydonia submitted by
this author in the summer of 1998 for oral presentation to the Division of
Planetary Science (DPS) of the American Astronomical Society was rejected.
This was the only rejection of an abstract by a member in good standing at
this meeting, with over 600 other abstracts accepted. Rejection of a
member-submitted abstract is a rare event (unprecedented for this author)
because presentation of papers before peers is the primary means of
getting feedback before submitting written versions of papers to journals
for peer review, and because justification of conclusions is not normally
provided in an abstract. The DPS abstract review committee based its
decision on the evidence they had seen with their own eyes in the image
released by JPL-PIO to the media. On appeal, they reversed their decision
and accepted the abstract for a late poster paper; but the damage had
already been done. The subject matter of Cydonia and the “Face” on
Mars was by then on a list of topics not suitable for consideration by
certain mainstream technical journals such as Nature magazine. By
editorial policy, papers on the subject of the “Face” can no longer
receive peer review at that magazine.
your opinion about the artificiality of the “Face” may be, and
whatever the actual merits of the issue may be, it seems beyond dispute
that allowing world opinion to be based on the image in Figure
3 was scientifically inappropriate. When considering why this
happened, we appear to be left with an unhappy choice between dishonesty
Correcting the Photographic Shortcomings of the
what would the Face mesa have looked like if the image had been taken
under better lighting conditions from an overhead perspective? Modern
computer image enhancement techniques can do an excellent job of
simulating different lighting and perspectives without significant
alteration or distortion of the image content. The results presented here
are the combined efforts of three professionals skilled in computer
graphics and image enhancement. Boris Starosta noted that, because of the
abnormal lighting conditions, the negative of the April 1998 MGS
Face image looked more like the 1976 Viking image than its
positive. Boris began with the MGS negative (shown in the left
panel of Figure
4), and switched the lighting so that the source of illumination
was northwest (upper left) of the Face and creating shadows accordingly.
This view is shown on the cover of this issue and in the center panel of Figure 4. Mark Carlotto
had previously mapped heights on the Face using shape-from-shading and
triangulation techniques, allowing him to change viewing angles, for
example, to overhead. This process is called “orthorectification”.
That view is shown in the right panel of Figure
4. Mark Kelly optimized the brightness and contrast for the
purpose, then put the transition between these steps into an animation,
available on our web site at <http://metaresearch.org>. The
starting, middle, and end images from Kelly’s animation are shown in Figure
Left: negative of the Face as seen by the MGS spacecraft in
April, 1998. Center: Lighting source switched from SE to NW.
Right: Viewing angle switched from 45° west to overhead. Click on
above image to view full
animation by Mark Kelly, whose web site is <www.electrobus.com>.
Need an animation viewer? Click here
and see link at end of page.
I studied image processing myself, I worried that the biases of the person
doing the processing might contribute significantly to the image seen. Now
that I am more familiar with the process, I can see that it uses
objective, standardized computer techniques, and does not add features to
an image that are not present in the original. The techniques used are
more like focusing a camera – they change the camera’s view to one
more like what the human eye would see if viewing directly. The exception
is the left portion of the east (right-side) eye, which was hidden behind
the nose ridge, and for which no data exists other than that in Figure
1. It was therefore filled out artistically by assuming symmetry
with the other eye socket.
** [See footnote at end.] **
JPL personnel who decided to release Figure
3 to the media were right about one thing. If they had released the
unfiltered spacecraft image to the press, the controversy over
artificiality of the Face would not have been settled in the minds of many
Proof that the “Face” is Artificial
The Viking images presented us with a mesa and a reason to
suspect artificiality – its seemingly improbable humanoid-face-like
appearance. However, as is well known, face-like images sometimes appear
in clouds, profiles of mountains, and various other random or visually
noisy scenes. Moreover, humans have a tendency to perceive order, even in
the midst of chaos. A scientific principle known as the a priori
principle, a part of scientific method, teaches us that judgments of the
significance of unexpected findings in random data have ambiguous
significance at best, and are generally not significant. This is simply
because remarkably regular patterns frequently arise by chance, even when
the a priori odds are billions to one against that happening.
example, every deal of 13 playing cards gets the player 13 unique cards.
The odds against being dealt those specific 13 cards are 635-billion-to-1.
Yet every hand dealt yields 13 unique cards with the same long odds
against that particular deal happening by chance. Whether the player
receives 13 random cards or 13 spades, the odds against the unique result
of the deal occurring by chance are the same.
By contrast, if someone predicts before the deal takes place that
he/she will be dealt 13 spades (or any 13 unique cards), and that
prediction turns out to be correct, we may be certain, at odds of 635
billion to 1, that something other than chance was responsible for this
successful prediction. Both scenarios involved the same event – a hand
of 13 spades dealt to a particular player from a deck of 52 cards. In one
case, no correct prediction was made in advance, nor was one possible. In
the other case, a correct prediction in advance of the deal was made. The
latter is an a priori prediction, meaning one made before the
result is known. The expression a priori means “proceeding from a
known or assumed cause to a necessarily related effect; deductive; based
on a hypothesis or theory rather than on experiment or experience; made
before or without examination; not [yet] supported by factual study.”
an a priori prediction exists, the results then become a test of
the hypothesis on which the prediction is based. A successful prediction
tends to support the hypothesis that generated it, and a failed prediction
tends to falsify the hypothesis that generated it. The degree of support
or falsification depends on how probable or improbable it was that the
prediction would happen or fail to happen by chance alone. A priori
predictions are a valid basis for testing scientific hypotheses. A
posteriori findings (made after the results are known, but still
sometimes called “predictions”) are generally not a valid basis for
drawing conclusions because their significance is, at best, ambiguous or
indeterminate. The number of possible ways an a posteriori finding
might have arisen by chance is usually vast and impossible to estimate in
an unbiased way. Ignoring the results of an a priori prediction is
no more valid scientifically than is drawing conclusions from the results
of an a posteriori finding.
As this applies to the “Face”, all Viking images were a
posteriori, so no reliable conclusions could be drawn from the data
initially available. However, the images did allow formulation of specific
hypotheses for further testing. The competing models were:
- The “Face” is an artificial structure built by an intelligent
species (indigenous or visiting) and intended to depict the face of a
member of a humanoid-like species, whether their own, ours, or some
- The “Face” is of natural origin, resembling a humanoid face
entirely by accidental chance combined with our predilection to see
familiar patterns in otherwise non-ordered data.
With regard to some of the new data provided by the MGS
images taken in 1998 or later with ten times greater resolution than the
older (1976) Viking images, these competing hypotheses and their
consequent predictions have a priori status. Scientific method
attaches significance to the test results of predictions having a
priori status. Disputing or ignoring the results of tests of a
priori predictions, whichever way they go, is itself a form of a
posteriori reasoning, generally of questionable validity because it
violates the controls against bias imposed by scientific method.
example, the artificiality hypothesis predicts that an image intended to
portray a humanoid face should have more than the primary facial features
(eyes, nose, mouth) seen in the Viking images. At higher
resolution, we ought to see secondary facial features such as eyebrows,
pupils, nostrils, and lips, for which the resolution of the original Viking
images was insufficient. The presence of such features in the MGS
images would be significant new indicators of artificiality. Their
existence by chance is highly improbable. And the prediction of their
existence by the artificiality hypothesis is completely a priori.
contrast, the natural-origin hypothesis predicts that the “Face” will
look more fractal (e.g., more natural) at higher resolution. Any feature
that resembled secondary facial features could do so only by chance, and
would be expected to have poor correspondence with the expected size,
shape, location, and orientation of real secondary facial features. Any
such chance feature might also be expected to be part of a background
containing many similar chance features.
2, it is possible to see details in the image (once the right
correspondence to the Viking image is recognized) that might have
been intended to portray each secondary facial feature – eyebrow, pupil,
nostrils, and lips. These are more plainly visible in higher-magnification
views with brightness and contrast adjusted for each area because of the
limited contrast in the image. Such views may be inspected at
<http://metaresearch.org> in the Cydonia section. Detailed study
with image processing software shows that these secondary facial features
exist where expected by the artificiality hypothesis, but nowhere else on
the mesa. This rules out a background of many similar features from which
we might pick out just ones that fulfill our expectations. Moreover, each
feature is present at the expected location, having the expected size,
shape, and orientation. The odds are against any of these features arising
by chance, and against each feature having any of the four listed
characteristics. Each of these probabilities has been carefully and
conservatively estimated in a fuller treatment of this topic.  The
combined odds against all of these features being present and having all
expected characteristics to the degree actually present, when taken
together with the absence of similar features in the background, exceed a
thousand billion billion to one (1021 to 1).
Strictly speaking, science does not prove physical hypotheses; it
disproves them. In that sense, all we have done, technically, is rule out
the natural origin hypothesis at the cited odds. However, unless we can
formulate some other hypothesis competing with artificiality that makes
similar a priori predictions, we are compelled to accept
artificiality as the most reasonable explanation consistent with the a
priori principle of scientific method.
 T. Van Flandern, H. Crater, J. Erjavec, L. Fleming
& H. Moore (2000), Evidence of Planetary Artifacts, preprint
available from Meta Research (may eventually be posted to
** Footnote added 00/08/14 **
I have since learned that the last two sentences of this paragraph
(immediately following Figure 4) are incorrect. Mark Kelly filled in the
small, missing information using low-resolution Viking imagery, not
artistically. This was based on Mark Carlotto’s shape-from-shading models
that combine both Viking and MGS images (<http://www.newfrontiersinscience.com/martianenigmas/>).
This answers concerns expressed about the reality of the extension of the
mouth to the right (east) side of the Face, and the presence of a complete
eye socket on that side. Both are real features, not artist’s concepts.
Lan Fleming has done a critique of Kelly’s final image, available at <http://www.vgl.org/webfiles/mars/face/newface.htm>.
He says, “While I pointed out what I see as a few flaws in [Kelly’s]
enhancement, I think the overall work is valid and the result is powerful,
which is why I went to the trouble of doing a critique of it.” If you have
not already viewed the animated transition of the Face from the MGS lighting
and viewing angle to the Viking lighting and viewing angle, it is well worth
the effort. [animation]