After the pictures of the alleged assailants were published by the Met Polis it was soon noticed that the images of them supposedly arriving at Gatwick were suspicious as they both showed them apparently going through the same spot at the same second – clearly impossible. The response of official sources was to say that they had walked down two parallel corridors in exact synchrony which was why the timestamps were the same.

I downloaded the original images from the metropolitan police site, the metadata on both indicates they were edited at different times on May 3rd, so there is no doubt that they have been altered. Note this is two months after the date shown. The pixel dimensions of the two pictures are different, another telltale sign.

Next, there is the fact that if you mirror image and rotate one image by 11 degrees then the walls, barriers and doors come into alignment. The purported entry channels all have centrally mounted pan-tilt camera units in hemispheres about 10 cm across directly above their exits. Whilst pan and tilt can produce shifts in x and y directions, pan and tilt in azimuth and elevation cannot produce rotation or mirror imaging. Try it with the camera in your phone. To produce rotation of the image you have to rotate the camera around its focal axis. The cameras look similar to the ones below. To produce an 11-degree tilt in the view from one camera you would have to have it hanging down by about an inch and half on the left-hand side of one of the cameras, but the published photos of the cameras show them mounted flush to the ceiling.pan-tilt-security-cameras-4x-zoom-elite-2mp-vandal-61027big

Thirdly I have examined the timestamps at high magnification, and the shapes are pixel exact between the two timestamps. That is ok if they were generated entirely digitally, but there are several factors which militate against this:

  • The digital shape is anti-aliased, which is unlikely for a simple white message applied via overwriting but is just possible. Normally you would generate an overlay for a camera system with a simple binary pattern, not an antialiased one.
  • Anti-aliasing is a technique of using intermediate shades of grey to make it look as if the letters are of higher resolution than they really are. It is used in high-quality text rendering on computer workstations but requires either a GPU or appreciable processing resources which would normally not be necessary for a timestamp in a surveillance system. But just let us suppose that there was a sophisticated text rendering system in the pipeline.
  • If the letters were applied by the hardware of the capture system both the shape and the brightness of the pixel values within the timestamp would be the same. But in fact, corresponding pixels in the two timestamps, whilst having identical shapes differ in brightness by up to 10%.
  • Simple boolean operations in the capture hardware would never produce this result. You could easily produce it if you overlaid a stencil on two pictures and then subsequently adjust the gamma, of the whole image, including the stencilled area. This is easy to do in a photo editor, which the metadata indicates has been used.

Channels.gif

The GIF, which was produced by a reader of Craig Murray’s site, shows how if you mirror image and rotate one of the images you can bring the walls, doors, barriers and the rest of the background into alignment.  As argued above, this is not a transform that would occur by panning and tilting a camera.

Similarly whilst you cannot produce rotational or mirroring transforms by camera pan or tilt, all of these are easy in photoshop. So yes, the images have been produced using photo editing software, how extensively they have been altered is open to question.

 

Other people have pointed out clear photoshopped artefacts in photos of the suspects in Salisbury, areas where the suspects are surrounded by a glow consistent with the application of a Gaussian blur after applying a stencilled image.

40950764_892203534318193_5091910117134172160_n

40647351_892201810985032_494083142127190016_n