An analysis of Bellingcat's analysis – Part 2 by @WeBaGeMyghty

An analysis of the analysis – Part 2

On 3. Juni 2015 by Myghty

Due to the overhelming response to my last posts, I decided to postpone my damn exam preparations and continue with the analysis of picture 5.

I will go through it a little bit fast since the mistakes made are pretty similar to the first part.

The metadata analysis above shows that Picture 5 has been processed with Adobe Photoshop CS5. After processing, the image was scaled down to a size of 900×600 pixels and saved as a new image with a compression quality of 75%.

– Bellingcat Forensic Analysis of Satellite Images Released by the Russian Ministry of Defense Page 20

Yes, that is correct, but as I pointed out in the first part, this is nothing suspicious. Someone added the seen annotations and therefore he needs some kind of editing software. Well, in this case Adobe Photoshop.

Again, they state:

[…] all image content should present roughly the same error levels if the photo has not been altered.

– Bellingcat Forensic Analysis of Satellite Images Released by the Russian Ministry of Defense Page 23

Which is just wrong again. As written in the tutorial from Dr. Neal Krawetz on fotoforensics, similar structures and similar colors mostly have the same error level. But not everything. Just due to multiple resaves it can happen, thateverything appears to have the same EL.

Now we get to the, according to bellingcat, added vehicles:

By contrast, area C of Picture 5 shows soil structures with significantly lower error levels than the square area around the vehicles.\

– Bellingcat Forensic Analysis of Satellite Images Released by the Russian Ministry of Defense Page 23

and again: different structures different colors. On top of that, they compare an ELA from MoD picture 2 with an ELA of picture 5. Since in picture two, we have a contrast between the vehicles and the ground, but see no difference in the ELA, I assume, that this was just due to some resaves. As pointed out already, the Error level is going to get more equal, the more often I save the picture.

That said, the second portion of the ELA they show, shows no big difference betwen the small rectangle around the vehicles and the ground. According to their statement, that all image content  should present roughly the same error levels, this would show, that the rectangle was not added to the picture. But we can clearly state, that the rectangle was added later on. Or does anyone know a camera which adds rectangles automatically?

Continue reading .......