Stan – Dr.Ciani contacted me by email and offered to evaluate the originalÂ digital copy. At that time I obtained from Phillips – Bilby an untouchedÂ copy. Below is his analysis of the Gamecam Photo, originally postedÂ here on the 19th of April, 2010. And then below that is Dr.Ciani’s analysisÂ of the untouched copy.
Anthony Ciani, PhD (firstname.lastname@example.org)
Condensed Matter Physicist
(Dr.Ciani’s thoughts on the posted photo of Gamecam Photo
of 19th of April, 2010.)
After reading this and seeing the original scan, it became clear what thatÂ person did. He tried to apply a sharpen or deconvolution filter to aÂ heavily noisy image, which was already in focus. The result is that itÂ creates a moire pattern, which hides detail. He then “cut” around theÂ figure, and applied some type of level adjustment (probably a gammaÂ filter) to attempt to brighten the subject. The end result was a lowÂ quality, distorted image, which appeared to be doctored.
[Stan – the original non-enhanced photo. For large scale image click link]
(Dr.Ciani was supplied with the above untouched original photo whichÂ he then enhanced himself. What follows is Dr.Ciani’s analysis of thatÂ original photo of which Dr.Ciani enhanced himself and is posted below inÂ both small and full scale.)
My analysis of the photo:
The digital image purports to be a scan from a developed photographicÂ film, used in a game camera. The type of film is unidentified. The imageÂ contains EXIF data indicating it was made by a Noritsu Koki QSS-32_33, andÂ saved by Noritsu’s image software. EXIF data can be faked; however, thereÂ are no indications that the image has been altered in any way, post scan.Â The Noritsu Koki QSS-32_33 is a low volume film processor, and the imageÂ was probably made directly from the negative; although the QSS-32_33 canÂ scan images from prints. For simplicity, the term “film” will be used toÂ refer to the scanned original, and “image” will be used to refer to theÂ digital data.
The game camera (model unknown) is a flash photography model. TheÂ object of interest, hence forth referred to as unidentified humanoid shapeÂ (UHS), is just within the range of the flash, and is approximately asÂ dark as the surrounding vegetation, and similarly hued. The intensityÂ values in the image span almost the entire range, but there is no trueÂ black present in the image, which reduces the likelihood that there hasÂ been alteration of the image. UHS is within a darker region of theÂ picture, and being dark itself, there is a substantial amount of ditheringÂ over UHS. This is typical when scanning dark areas, as noise in theÂ electronic sensors becomes equivalent to the signal.
UHS appears to be a real object in the film. The camera flash and lensÂ are not coaxial, which allows the lens to see shadows caused by the flash.Â These shadows should be offset in the same direction which the lens isÂ offset from the flash. In this case, the lens would be below the flash.Â The distance between the shadow and the casting object should decrease asÂ the object rises above the horizon. The shadows being cast on UHSÂ appear to be consistent with this. UHS is also clearly in front ofÂ some limbs and trees in the background, which pass behind UHS. There isÂ no indication of UHS being anything other than a real object in the scene,Â and there is no indication that any part of the scene is a compositeÂ image.
[Stan – Dr.Ciani’s enhancement – For large scale image click link]
The above enhanced image has undergone several processing steps to removeÂ noise (which adds blur), and then enhance the edges. This process wasÂ performed in the following manner. The image was copied into a newÂ layer, and the color space was reduced to gray scale. A filter was appliedÂ to erode dark pixels. This filter increases the intensity of a pixel ifÂ it is of lower intensity than its neighbors. The effect is to removeÂ noise over the lighter areas, identifying solid objects. The filter wasÂ applied with a 25% level (75% original gray scale, 25% filtered gray scale).Â A second gray scale layer was then enhanced using a difference ofÂ Guassians, to bring out edge detail, which created a light map, usedÂ to highlight continuous regions. The edge enhanced layer was thenÂ combined with the filtered layer to produce a light map which highlightedÂ the branches and UHS. This light map was then applied to the originalÂ layer, to brighten the key objects. The result brings out the form of UHSÂ and the trees, but removes the appearance of texture from those objects.In reality, it was the noise which destroyed the texture.
There is a caveat as well. If the noise is strong enough, this processÂ can highlight objects which never existed. The only region with strongÂ enough noise for this to occur is the area around or above the head ofÂ UHS. In the enhanced image, it appears UHS may have a highly domed head.Â This may be a result of the filter highlighting a random “pattern” in theÂ noise, essentially, digitally enhancing pareidolia. The general form ofÂ UHS is certain. The noise is too low for pareidolia.
A scan of the film using higher exposure settings may provide additionalÂ detail for the dark regions of the film, while destroying detail in theÂ empty area illuminated by the flash. Some of the noise may also beÂ due to film grain, depending on the ASA speed and model of film. TheÂ limiting information seems to be pixelation, more than noise or filmÂ grain. The film should contain more information than this, and a higherÂ resolution scan should yield significantly better detail. Even better ifÂ multiple, high resolution scans are averaged together, to remove noise.
Stan – Original untouched photo on left, Dr.Ciani’s enhancement on right.
The UHS object itself is interesting. There are three bright points onÂ its figure. Two near the knees, and one between the left elbow and body.Â These might be reflective materials (e.g. mica), bare metal, or brightlyÂ colored objects. The two lower spots are very likely flecks of material.Â The object under the arm is different, and clearly being carried, withÂ the left hand wrapped around it. It is also not a single bright point,Â and some details can be identified. The object is oval in general shape,Â darker than UHS, and may have color, perhaps blue or green. The brightÂ spot is roughly in the center of the object, and is not perfectly round.Â The bright spot seems to have a darker spot in its middle. In this case,Â pixelation is the limiting factor, the noise is not substantial.Â Overall, the image appears to be an unaltered scan from a film processingÂ unit. The film is likely a picture of a real scene. The limiting factorsÂ pertaining to detail in the image are the scan resolution, followed by theÂ exposure level. It is suggested that the film be taken to a professionalÂ photographic shop, where it can be scanned using a higher end filmÂ scanner, perhaps with an optical resolution of 4000 dpi or higher. FilmÂ scanners with optical resolutions of 7200 dpi exist, and would be best.Â Ideally, a scanner with an adjustable exposure level or high dynamicÂ range should be used to intensify the dark areas. Film is analogue, andÂ so, such an optical intensification should produce a better scan.