Seriously Robert, if you want replication you will need:
The exact MAC the scan was performed on.
The exact scanner used.
All of the exact software used, in matching versions.
Sample stock of the security paper.
The research to find all of those.
Alternatively, you could go ahead and play with Photoshop sliders and get whatever result you want.
Keep in mind he's changing his horses faster than a Pony Express rider. At first he wanted his critics to exactly replicate the optimized PDF, right down to the number of objects. Now he wants his critics to replicate the scan of a photocopy, right down to auto color correction. Two separate requests -- none of them relevant, all of them shifting the burden of proof.
In order to replicate the PDF you'd not only have to have everything you mention above, you'd also have to place the document on the scanner bed
exactly as it was originally, to within 0.0015 inch (the Nyquist sampling tolerance for a 300 dpi scan) of the original position and to an incredibly tight angular tolerance of 0.00008 degree of the original document orientation (the same Nyquist restriction over 8 inches of document extent). Then
and only then will you get the same raster image from the scanner as the original worker did.
Why such tight tolerances? Because any material difference in the captured raster affects the geometric pattern matching of the PDF optimizer and will produce different objects. It also affects the skew factor, which determines which elements of the picture will require greater pixel depth to represent them. A perfectly aligned document, for example, can have its horizontal and vertical lines converted to bitmasks. A slighly askew document, without deskewing, requires greater pixel depth to accommodate antialised lines.
The stark reality of what it would take to exactly reproduce the Obama PDF in optimized form is why the Birthers get laughed out of court when the real experts show up. Real experts like me can show that (a) optimization produces a multitude of objects, which the Birthers have mistaken for "layers"; (b) the nature of any one layer is completely consistent with PDF optimization; (c) the overall distribution of image data among the objects is entirely consistent with PDF optimization; and (d) there is no forgery claim yet made on the basis of the digital image that cannot be better explained by ordinary PDF optimization.
Sure, let WND continue to dazzle their slack-jawed readers with pixels here and pixels there. Let Mara "I make pretty pictures" Zebest pretend she's got any clue when it comes to photographic analysis and document evaluation. Let that wacky sheriff pretend he knows how to run a real investigation. Pretty much everyone can see it's all just political tripe.
Any luck printing Pantone 877? No?
I'd love to see him even reliably
scan Pantone 877. My guess is that he's finally gotten some practical understanding of scanner, copier, and printer color gamuts and he's hoping this whole thing will blow over without him having to demonstrate a darn thing.
How about that elusive multicoloured pixel? No?
I'm still giggling over this one.
"Pixel" is short for
picture element, the smallest elementary constituent of a digital image. "Elementary" means it cannot be broken down into smaller elements that have useful properties. What are the properties of a pixel?
It has
location, meaning that it has unique and deterministic coordinates belonging to the dimensionality of the image. "Elm Street and 4th Avenue" should uniquely identify a location in a city. Some other corner would have some other unique address. In digital images we define this to be the case.
It has
extent, meaning that it completely and atomically represents a measurable portion along that same dimensionality. "The 400 block of Elm Street" identifies a region of the city. While the underlying reality incorporates bushes, houses, dogs, and sidewalks, the designation "400 block of Elm" contains them all. In digital images we define the pixel to be so quantized.
It has
photometric value, meaning it atomically represents photometry data for that location and extent in the image. The dimensionality of the pixel value depends on the dimensionality of the underlying model. IR imaging from the sun, for example, could have a single value. CYMK production print pixels, for example, have a tuple of 4 values. However, each pixel can have only one tuple by definition. You can't have a multicolored pixel any more than you can build two houses in the same place.
