ANALYSIS OF "FAKE" IMAGES

BY DAVID RATH

The Art Bell vs. Courtney Brown Controversy


(Last update 1-21-97)

This is an Email sent to me that I find challenges the recent controversy over images and suggest that perhaps there is more than one answer to this argument. HERE is the response from Dr. Courtney Brown concerning the accusations. I will let you draw your own conclusions.

Thank You,

gary d. goodwin



 
From: David Rath

Hello.

For those who may not know, the following document and attached images are being sent to Gary Goodwin as well as Art Bell.

These images represent a small amount of work to debunk the debunkers, as it were. For a little background history, let me fill you in on the chain of events thus far.

On a tip, I decided to check out the controversy surrounding Hale-Bopp and the Companion (Hail-Mary,) herein reffered to as HB and HM, respectively. What I found was a lot of people working hard, even to the point of lying, to debunk the notion that HM even EXISTS despite solid proof (some released by government agencies themselves) that shows Hale-Bopp with Companion in tow indisputably.

Hearing the Limbaugh-like (much bark, little substance) ranting and raving and finger-pointing on the side of the debunkers, I noticed a few things: first, that nobody WANTS the public to know or even suspect that HM might be real, and second, that they throw a lot of "fluff," namely information that would be unverifiable to the average citizen.

I am not the average citizen. I am an Artist well-versed in digital mediae, as the computer is used in almost all of my imagery, and have a rather intimate knowledge of rendering capabilities and tendancies. I have an open mind, and would be willing to believe whichever side was believable.

What I would like to address here are the two images labelled as "REAL" and "FRAUD," both showing very similar images of HB before a cluster of bright stars, more in the background. These images are, at first look, virtually identical with one major feature, that being that "FRAUD" shows the image of HM above and to the left (or NorthWest) of HB, in the midst of the cluster of other bodies.

Both images were touted as being from the same source photo by the debunkers, on the following criterion:

1] THEY ARE EACH EQUAL IN PIXEL SIZE.
What they mean by this is roughly a mystery, but I have come up with three possible interpretations. First, that there is somehow an actual difference in the size of their pixels, which is in itself a joke. Second, that their vertical and horizontal size is the same in pixels, which would be irrelevant because any image can be scaled or cropped to any size. They are, in fact, the same size: both 250 pixels by 190. Third, that the size translation for printing is the same. For example, a picture is defined as having 72 pixels per inch. That would make the PRINTED pixel size 1/72nd of an inch. As well, if it had a size of 120ppi, the pixel size would be 1/120th of an inch. Simple enough, but also irrelevant because MANY imaging programs can adjust this measurement and it has NO BEARING on the origin of the picture - unless you consider that most of these images are in GIF format, which HAS NO SPECIFICATIONS FOR PRINTER SETTINGS ANYWAYS. In other words, 3 points, all moot.

2] THE IMAGES ARE IDENTICAL.
This notion is flatly false as well. I have not uploaded the original images, as redundancy is memory-waste but these can be rechecked any time. To display this, I have made two overlays of the original picture that are VERY revealing. This I did by tinting the "REAL" image as red, and the "FRAUD" image as blue, then compositing the image to show their true relationships.
The first image,
simply shows a composite of one image laid directly over the second. Lo and behold, and what do you know? The images are shifted, as it were - THEY DO NOT MATCH IDENTICALLY. Almost all bodies are out of position for where they would be in identical images. This I noticed while netsurfing, but I made the composite to doublecheck and my suspicion was validated. Then came the second composite,
which I created to determine the relationship of the shifted objects. This I did by overlaying the red/blue images as I did before, but this time I centered the prominent objects as best I could - HB and the other prominent central stars. This revealed that there had been some scaling going on. Namely, the "REAL" image was vertically stretched (in relation to "FALSE") and just as well, "FALSE" was horizontally stretched, in relation to "REAL." I don't profess to know much about CCD imaging, but in my estimation, such variances in high-precision digital imagers would be common. This in itself would suggest that the images had two separate sources, but on the alternatives that images were carelessly scaled to resemble one another, or to hide certain telltale signs of digital doctoring. More, that the "REAL" image displayed a SLIGHTLY different area that the "FALSE," as it contains extra material not shown in the other. These signs may also suggest that the images were taken from separate locations, as once again, they were out of equal scale. A lot of inconsistancies for two supposedly identical pictures.

3] The "FALSE" photo is an astronomically-acceptable image.
This was said by the debunking committee themselves. What a dangerous thing to say, especially if your own image is not of that same standard. Once again, I'm a nitpicky Virgo. Just as I noticed by sight that the images were out of line, I noticed something else: a barely-discernable fingernail crescent in the midst of the star mass of "REAL," just above and to the left of HB. Guess what? That slight fingernail crescent, for those of you attentive enough to discern it (come on, I did it with ONE EYE) most EXACTLY outlines the image of the Companion, HM, in the "FALSE" photo. As I said, this crescent outline is barely discernible, so as an Artist, I made it my duty to make it more visible for comparison in case you want to look for yourself - and I hope you do. To do this, I focused in on the cluster itself. From each picture, I extracted the cluster. The sizes of the pics I took were identical, and as best centered on the cluster as possible. To make the pics compatible and comparable, I converted them both to grayscale, since the "FALSE" image effectively was already. This alone helped discern the fingernail crescent, by removing contrasting colors that broke it up. From there, I adjusted the intensity levels of the "REAL" fragment to diminish the background and enhance the crescent. Unfortunately, I could only do a limited job of this - it is a somewhat disperse and grainy pic that to go any further would destroy detail. As I suspected, the crescent showed up.
This image shows the crescent amid the cluster, while-
this image shows the cluster with HM. This was a lot, but not enough as the crescent could be explained away as a twist or flare in the coma of HB. What I needed was to prove their relationship. From here I effectively faded one picture into the next, as shown in this image-
As you can see, the crescent DOES IN FACT DEFINITIVELY OUTLINE the right side of the Companion!!!

So - how does this in fact prove that the "REAL" image is in fact the fraud? It doesn't - if you believe in a coincidence so far beyond the realm of likelihood that impossibility is your nearest neighbor. All I can do is try to instill in you some understanding of photoimaging.

One of the most difficult things to do is to hide a seam. For one thing, the process of selecting a portion of an image is not quite perfect. While selections can be "Feathered" (making the edges gradually less transparent for better blending) this has it's own flaws. For one thing, the mathematics a computer uses are rather rigid, often resulting in one edge getting better attention that it's opposite. This may result in a left/top border being more/less attended than the bottom/right. This may have caused the seam. Another factor to take into account is contrast. A flatly-toned image pasted into an image of varying tones will show edges more sharply in areas of higher contrast, less so in areas of less contrast. For another, in the example of "REAL" and "FALSE," PIXELS COUNT! These are relatively low-resolution pics, so a few overlooked pixels (such as those in the border of a poorly-selected area) can make a VAST difference.

In this case, it took the eye of an artist who knows his medium to spot the difference, and it is there. I hope that after reviewing the materials, you can spot the difference as well, even in that many copies of the original sprinkled about the net. I've done all I can to help you on this one (for now) but I'll be keeping my eye on any future releases of other images, and am even creating some conceptual art based on the whole situation.

D.A.Rath
 


  Why not provide the raw data and let each person come to their own conclusion? 
 
 
copyright 1996 gary d. goodwin