Tuesday, May 30, 2006
More Heflin fun...
On the left above is a cropped section of Heflin photo #2. It is cropped from a 600 dpi resample of the large scans from the Druffel, et. al., re-analysis posted here.
The image on the right is the same cropped section with Adobe Photoshop's "Emboss" filter applied.
When a photograph is blurred, or sometimes when over-exposed, detail can often be recovered using an emboss filter. While it can also result in false information, it can be useful. As with the sharpen filter, when the emboss filter is applied in an "excessive" amount, subtle details which were barely (or not at all) apparent in the original image can emerge.
If you look at the image on the right above, I think there is an apparent straight-line feature running from the object to the door frame. Now, this image is oriented as it was in the original photo, so there is geometry that can emerge as a result of the "grid" pattern of pixels. In other words, the apparent filament in the image could easily be a by-product of the pixelation of the scan, and could be misleadingly enhanced by the embossing filter.
BUT, I think it is noteworthy that we are assuming (for sake of argument) a hoaxed image using a model and a suspension line, and under this type of filtering we see a vertical straight-line filament appear...just where it would need to be. Note that there is not a similar feature anywhere else in the image. Note also that while the filament is not exactly lined up with the center of the object, the distortion of the filtering involves two copies of the image being offset, so that the filament might not be located exactly, but its "impression" can still be accurate.
I will add that I did nothing more to the image than what I described. Just cropped the resampled image, and applied the emboss filter at near maximum level.
Food for thought? Or just empty carbs? You be the judge.
The line is not at a rational balance point for the object, is my perception.
AVG Blog -- http://alienviewgroup.blogspot.com/
True...as noted in graf 5!
The filter used creates two copies of the source image...one biased to the bright and another biased to the dark. The amount of "offset" between the two versions creates the illusion of depth that defines the term "emboss". The filament does not line up with the center of the object, but that does not necessarily imply that a filament (if present) would be in THAT location...because the two biased images are not in their initial locations either.
The process is used to find a "signal" from the "noise". While this is by no means definitive, it is rather compelling that by simply applying a filter...with NO preconceived desire to achieve anything, we see a suspicious anomaly in the photo.
As I said, I don't know of the Heflin photos are faked or not, but these analyses seem to indicate that SOMETHING weird was going on.
But I agree with you that the apparent filament is not where it should be to provide definitive proof of a hoax.
Thanks for the comments!