SUBMISSIONS
- Page 1 -






Bill Stanley (Yours Truely)
Belleair, Florida

July 29, 1998


Some of my recent efforts using
Paint Shop Pro 3.0

A fairly interesting "pile of rocks" wouldn't you say? Perhaps the "skull" is a prank perpetrated by the Pathfinder imaging team.










(New submission by...)



Woodlock


December 7, 1998














(New submission by...)



Bill Schlosser


December 16, 1998






(New submission by...)


Anders Gunnarsson

Mariestad, Sweden

January 31, 1999

Used Photoshop 5






(New submission by...)


Todd


January 27, 1999

Used Paintshop Pro 5.0








Clipped from PIA01000 in TIFF
Rotated 90º Right
Enlarged 400% using Bicubic Resample method
Brightness increased 3%
Contrast increased 3%
Converted to GIF for the Web

I can send you the tiff file of the same dimensions if you'd like.

I went back to the link you provided and clicked on the download options button. Apparently I only downloaded the reduced size image the first time (about 1 Mb) so I downloaded the full size 5 Mb image. While I was waiting I read the comments regarding how the image was created from several smaller pans, then they enlarged the image 500% (probably with some type of interpolation). That's a lot of enlargement!

Once I started processing the image I found that I had to enlarge it another 400% to produce a size simlar to the "skull" that you displayed on your Web page. That's a total 40,000% increase in enlargement from the raw image or 400 times as many pixels.

Each time the image is enlarged 100% (doubled) each pixel is now split into four duplicate pixels. If you enlarge it using resampling (interpolation) the software doesn't just duplicate each pixel four times, but it actually calculates the value of each pixel using a percentage of the information in each adjacent pixel. Different types of interpolation calculate the values in different ways and use different numbers of adjacent pixels. Mike Bara wrote a pretty understandable article on this topic called "Image Enhancement 101" (no connection to this site). You can find a link to Bara's article on this site in my "resources" section (suggested reading).

My problem with this method on this image is this. The image was enlarged 500% (5 times) presumably using some form of interpolation before we ever got it.

That takes every single individual pixel and multiplies it into 25 pixels. Then by using interpolation or resampling each 25 pixel group isn't copied as an exact duplicate of the original pixel. Each of the 25 pixel groups is an "average" of the original pixel and the pixel[s] nearest and adjacent to the original. It's synthetic data, created by averaging the original data as it's multiplied. It's a good process, but it has limits. I think NASA may have gone about as far as possible without creating illusions caused by spreading the image data too thin.

Now, I enlarged the already enlarged and resampled NASA image 400% (4 times) using Bicubic Resampling. The image now has 400 pixels (20 X 20)for each one individual pixel in the original raw image. That's some serious enlargement! My finished image clip of the "skull" is 250 pixels wide and 300 pixels high, 75,000 pixels total. However this 75,000 pixels represent the data taken from only 187.5 pixels from the original raw image, or an image that measured 12.5 pixels high and 15 pixels wide (about the size of the right eye in the skull clip). The balance of the 75,000 pixels in the new image are synthesized. They're artificial illusions.

To compound the problem, my Bicubic resample did not use original data to create the "averages" for the new pixels, but used the already synthesized-by-averaging pixels of the NASA processed image. In essence I created an enlarged and averaged copy of an enlarged and averaged copy. My conclussion is that the finished image of the "skull" is almost completely synthesized and is unreliable...sorry.

It's spooky looking and you have a keen eye for having noticed it, but it's just not real.

I tried adjustments like histogram stretching and brightness and contrast adjustments but this image was very sensitive to these steps and only allowed very small adjustments without noticably suffering. Sharpening, smoothing, blurring, and any other filter I put it through seemed to make it worse. The Bicubic resampling did produce a clearer image than the one you created, but the finished product is very similar.









(New submission by...)



Linda Goin


February 12, 1999

Used Photoshop






Lostboy1 is where I brought the original skultiff.jpeg into Photoshop and used the graphic pen filter to render the values (stroke set at 5, light/dark balance at 50% and stroke set at right diagonal).

Lostboy2 is taking the graphic pen rendering and using the smudge tool to blend in the shadows and bring up the highlights. This is where the mouse becomes unpredictable. The strokes were hard to manage. This is also where I start 'fleshing in' the skull - providing the structure to the face (and also where I realized how 'human' this skull was).

The final rendering was from touching up the watercolor rendering - adding details such as the eyes and highlights. I took the liberty of making the light source coming from the top right facing the picture - it still isn't structurally correct, but then again - it is an alien, right?

I'm an artist, (and a realist), so I tended to not get very 'alien'. In fact, it was rather eerie doing this, as the facial structure is similar to a young earth person - with the exception of a possible injury to the left side of the face (skull). Looks like he/she was bashed with something...at any rate, I found drawing with a mouse to be very awkward. A stylus would be on my gift list to myself with an art program.

Thanks for the practice and opportunity!

Linda Goin




(New submission by...)


RUBBERDOWN


February 27, 1999

Photo Magic 4









(New submission by...)



Silvio Sanchez


April 4, 1999

Miami, Fl.











Michael Lesnussa
Dordrecht
The Netherlands
August 14, 2000














Tony Scheuhammer
Ottawa, Ontario, Canada
I think he/she looks quite beautiful, actually.
Used PhotoShop
June 20, 2000















Urpi


February 21, 2001

I flip the image horizontal to see de lost part of the picture, I also work with de levels of light, an create a margin around the picture. Hope this work give us an idea of the face









Mark Zurbuchen


Doctoral Student Penn State Materials Science Currently at Argonne National Laboratory

December 19, 2000











Nathan Wong

Sydney, Australia

May 21, 2001











Melissa Woodington














(Your enhancements here)


Go To Submissions Page 2

Back