Replies

Let me preface all of this by saying that, again, I am NOT an expert ... I just have experience with this crap ... I hope people shoot me down if I get something wrong :-). Also, if you are familiar with this stuff, please don’t view it as a condescending explanation ... that is not my intent ... some people know nothing about this stuff while others will cringe at my explanation since I am trivializing a few things :-).

Also, this is really long ... sorry if it puts you to sleep :-) .

First, this photo has certainly been compressed. Just in case you aren’t familiar with compression, it is a scheme where you run an algorithm on small regions of the scene, compute some values for these regions, and effectively replace these “regions” with your computed values. You’re trading off image quality for an image that requires less space for storage.

These days, virtually everything you look at is compressed using some kind of JPEG compression ... JPEG works by dividing the picture into tiles ... the tiles are usually 16x16 pixels.

Let’s say a pixel is 32 bits of data. A 16x16 tiles requires 16x16x32 = 8192 bits. JPEG will take that 16x16 region, run what I like to call “scary math” on that tile (sometimes wavelets are used, other time discrete cosine transforms), and spit out some values. These values require far less memory than 8192 bits and are used in place of the 8192 bits. This process is called “encoding”.

When you open a JPEG file, all of those “values” that were spit out are run through the inverse of the algorithm that spit them out ... this creates the series of 16x16 tiles for the user to view. This process is repeated until the picture is reconstructed. This process is called decoding :-).

Now, the only reason I went into detail about this process is that many people confuse a lot of the “noise” in a compressed picture with “attempted photo hacking” :-). Recall that I said that the encoding process computed some values that represent the data in the tile. This means that we are discarding the original data. What you look at in a photo is only a representation of the original scene.

When we reconstruct the photo based on those computed values from the encoding process, we are not making a perfect recreation of that picture ... we are making something that looks a lot like the original, but will have some odd “noise” added to the picture since it is impossible to reproduce an image perfectly if the original data was discarded. This noise is commonly referred to as “compression artifacts”.

With all of that crap out of the way, let’s look at the pic of Slow Joe and Marine 1.

First, the compression is quite poor :-). If you zoom into the photo, you’ll notice the “tiles”. They used less “precision” (didn’t compute many values using scary math) to compress the photo ... it’s why those 16x16 tiles I mentioned look very pixely for lack of better terms. If you want to improve the image quality, you can increase the number of values the scary math part of the encoding process spits out ... however, you will require more space to store the results. This is why some JPEGs look better than others :-).

Next, some of these 16x16 tiles have a mix of Slow Joe pixels in them and some pixels from the background. We call these “edges”. Given the way the compression works, it is tough to represent an edge (or edges) with only a couple of values. When we run the image through the decoder, it will not do a great job reconstructing these edges since there isn’t enough information. Edges are the biggest source of compression artifacts in an image. They are also the most likely place you are going to see if someone was manipulating a photograph.

Now, let’s zoom into the area around Xiden’s feet. A lot of the area around his feet looks really odd ... the grass in some patches don’t seem to match up ... but a lot of that is due to compression artifacts. Grass has a lot of details/edges in it ... it will look poor once compressed and decoded. Nothing I see in that area looks out of the ordinary. His one foot looks like it is missing, but with the way the grass is standing in front of his “missing” foot, it looks more like that foot landed in a lower part of the ground and is angled a bit. Nothing looks ‘hacked’ in that region ... it looks noisy, yes, but you are going to get a lot of compression artifacts due to the edges of the grass, Xiden’s pants edges against the grass, etc.

If you focus on the edges of Biden in this picture, you will see lots of noise, but, again, that noise makes sense given the details along the edges. Look at the pedophile’s head ... you see the various little plugs of hair standing up and all over the place ... the edges of his white hair are up against a dark brown tree ... again, lots of edges + poor compression = lots of noise. Some of it looks synthetic/hacked, but it’s pretty much what one would expect to see in a low quality compression.

Overall, I don’t think this particular image is a hack at all. However, if they’re going to “hack” an image, they’re going to work with a raw source image that is not compressed, make the changes, THEN compress it. This process will obfuscate many fingerprints left behind that would indicate a manipulation, but there are still clues you might find if you look along the edges of the objects in the picture. Some of those clues are unnatural looking shapes (i.e. lots of straight lines kind of cut into the edge of a person & lots of “smearing” around an object are the first two things that you could look out for, but you also have to be aware of noise possibly causing the unnaturality :-) ). I can’t see anything like that anywhere in this image ... I think it is genuine.

Of course, none of this means that any theories are shot. I don’t think our military has much respect for the piece of trash. If this is all an act, they’re going to have to make things look real you know. I think this photo is a real one of Joe Biden getting out of “Marine One” (I put that in quotes on purpose) and is being saluted by a “US Marine”. That doesn’t mean that what others are thinking about the entire situation are wrong. I simply wouldn’t use this picture as evidence of that.