Turning onto Ukrainian problems I even started calling back my Russian lang (instead of improving my bad German, shame of me!) Thus exercising at Russian blogs I found a large assay by a popular blogger,
In it he gives his evaluation of the Forensic Analysis of Satellite Images Released by the Russian Ministry of Defence, which was published at the popular website for citizen journalism Bellingcat.
I dared to make approximate translation blogger ntv’s exploratory studies. That was a lingering and painful process but I like the road less traveled. I displaced some fragments. However still cannot call it Easy Reading. :) Thus,
Exposing Exposers of MH17
The Bellingcat investigative journalism group issued a report. It seems that the only aim of this report is disproof authenticity of satellite image published by the Russian Ministry of Defense (MoD) at a special briefing with media dedicated to the MH17 disaster.
I downloaded this report as soon as it was published and I bothered to read it from cover to cover. And if you like, now I submit “the investigation of the investigation” to approval of public.
From the beginning of the report the Bellingcat group asserts that photos published by Russian MoD had been doctored by using Photoshop program. Hereafter those journalists try hard to prove this unsophisticated speculation. For some obscure reason the group keeps under review just two in six images. For sure such selectiveness looks rather funny, but let it remain on the AA conscience.
The analysis itself utilized the following methods:
1. Metadata analysis.
2. Error level analysis (ELA) of JPEG-coding. In theory this method helps to reveal modified areas of an image.
3. Comparing dates according to indirect indicators.
It especially looks extremely solid for a layman. Right off the bat, the group makes a convenient discovery that Russian MoD used Adobe Photoshop CS5 software. How can we cop out after such an argument :)
Next step is ELA. The group marks on the photo areas with significantly different error levels because of crunching. On this basis they conclude that photo was falsified. They use http://fotoforensics.com site as a software tool for this operation. There are five such areas marked at Pic 4, for instance.
In substantiation of their arguments AA come up with Google Earth photo. There is also difference between cloudy and cloudless images. But according AA opinion, this difference is much less significant. From it they conclude that the cloud on the right Russian MoD photo had been photoshoped either.
After that AA locate a spot near one of military vehicles and notify that that was a pool of oil leaking from the vehicle. According the pool growing they manage to judge about the photo timeline.
Without giving specifics I’ll just say that there is everything, either oil timeline, or the timeline of the further analysis, there is every indication that Russian MoD images are fabrication and Photoshop-made. The second image was exposed to similar operations with the same conclusions, that this is deception and fabrication.
Hereafter there is analytics of the higher-quality version of one of the Pics, but there is no metadata for it. Without any doubt that looks incredibly suspicious according AA of the report.
The Conclusion is predictable,
I can hardly control my scoff, so let’s start the dissection. First of all let’s deal with accusation of Photoshop software using. This argument is fabulous, for Photoshop trace in metadata = the photo was modified.
[here is scoffing laughter of the piece author]
There is Photoshop trace in 100% photos of this blog, for I use it for changing a pics size. It’s obvious that Russian MoD used it for the same purpose. I advanced further and took a casual photo from the Bellingcat site. I uploaded it with the same software tool the report AA used and… whatever your guess could be? IT WAS ALSO MODIFIED USING ADOBE PHOTOSHOP SOFTWARE. What felinity!
Now come over to the tool revision. The AA assert that using ELA it is possible to determine if an image was modified or not. As I said earlier they used http://fotoforensics.com site for this purpose. I decided to try this method and used this site for checking of funny pic I found in Internet recently. It is on the left, and the analysis of loss caused by compression.
Hum. Indeed a text stands out. Also a posy looks a bit modified. The rest is obscure. Could it be the real? The image turned to be too specky. Is it possible that the analysis doesn’t work? The AA claim that a non-modified image has to be evenly dark. I found a must-be real photo and uploaded it to this web service.
It is cool if you can see any differences between the first and second right photos, for I don’t distinguish any. It seems that the method reliability is much dubious. But the pattern must be the same. Therefore I found another site (http://www.imageforumsic.org), which generates the map of compression loss on the photo basis. Waiting for getting evidence, but…
Even another web service shows that two men on the first pic are real. In other words, It is wrong to make conclusions about Russian MoD images on the basis of this method and judging from the pics provided in the report.
As it turned out that I wasn’t the only person doubting adequacy of the Bellingcat report. Despite very few people (including me and the AA of report) have an idea of the term, ELA sounds very serious and inspires with respect. Accumulation of strange and intricate terms tends to seem cogently.
There is an article by Charles Wood at the 7mei.nl. He is a New Zealand digital forensic researcher working in specialty for more than 10 years. Reviewing the report he makes the conclusions similar with mine.
Exemplarily Charles provides several original images showing signs of tampering according “the Bellingcat method”
The German programmer Alexander Gehret conducted analysis of his own and found errors of the Bellingcat method indicating lack of qualifications of its researchers. This is a fragment of his analysis of report.
Consequently uniform errorlevel is optional for an original image. It is well illustrated through testing of casual photos by pattern matcher. It turns out that the Bellingcat researchers don’t know beans about those tools they use. However we still need an expert opinion on this problem. And we have one.
At last an ELA service creator’s answer to a straight question about his immixture to the Bellingcat investigation,
Besides he proves the Bellingcat to publish fake images. The issue is about that very “Buk missile system starting” uploaded at the beginning of this year.
Thus ELA made by the Bellingcat group has nothing to do with verification and doesn’t provide any evidence if photos by Russian MoD were forged or not. We could stop here. But there is still diversity in dates.
The AA of report give finesse demonstration of deception determining that oil was leaking from one of the vehicles. Oil flows slow and sure. That is why a pool grows with time and allows drawing a deduction about images’ timeline. I’ll save my breath about mentiferous talents of those researchers, who elaborated liquid content via the Internet. I’ll alert of much more obvious detail.
Strange that the AA of report didn’t notice it or probably just draw attention to it on purpose. They write about a unit of equipment situated horizontally on the photo. But they don’t make clear what exactly vehicle it is. They also don’t pay attention that in spite of leaking oil, a vehicle started, moved and changed its attitude at the pic 4.
So, my friends, there is a most likely curtain-sided truck at the photo. The British experts must be ignorant in technics if they couldn’t identify it at a glance. However its shape is quite clear. Its body is broader and higher than other parts. It gives bigger shadow compared to the cabin. You can even see a whitish hood. Seemingly this is ZIL-130 cargo truck or something of that kind.
At this point the bold guess about leaking oil collapses like a house of cards, because all Soviet period trucks are front-engined. An oil leak would be near its fore wheels, not near back ones by all means.
Much more real version of a mere rainpool instead of oil. But such pool doesn’t fit the rest smooth story of the investigation, for pools appear after a light rain and disappear at once. Pools show nothing concerning the time of a shot.
From the very beginning I wasn’t sure about exact date of photography in Google. It must be kept in mind that Google service doesn’t make those photos by itself, but orders them at various companies. Photographing satellites move along a strict orbit and shoot accidentally during a clear weather (that is they make photos constantly, but just cloudless pics are used). It is often the case that the terrain representation is pieced together using photos of several satellite fly-byes over a certain point. In this case the notorious date of shooting is out of question. Also it is questionable if companies would date each tile for Google.
The same says Charles Wood,
I went extra mile and decided to conduct mini-investigation to find out if Google clarified its stand about exact map dates. Officially Google says nothing about it. BUT,
It did it earlier. In the old Google Earth help articles run very clearly,
The approximate date. Now those help articles were redesigned, so that this topic is there no more. But it was saved in cash.
Thus there is no possibility of drawing any conclusions from comparing satellite images with Google dates. Each and every “averment” is either disproved or they cannot be used for ascertainment the images fake.
Most of the report statements fail to stand a test with “for (what)”? This is very effective method. Just say a testable argument and end it with a questions “for what?” For instance,
1. Russian MoD retouched clouds over the Ukrainian forces deployment for (what)? It is clear that there is no sense in it. Consider the pics’ dating situation,
2. Russian MoD used old satellite images for photomosaic for (what)? Any unmilitary man understands that there is an army of spy-in-sky satellites of various countries frying over the military conflict territory. So Russians could easily use latest images for the truthlikeness. There are lots of such discrepancies destroying the investigation.
The report itself consists of a large amount of messed facts, complicated technics, wishful thinking and such innocent insinuations seeming seriously impressive for a layman. In other words, I would be very suspicious about such document. But this is not the end of my post.
Just over a half a year ago in Kiev I managed to attend an American journalist’s lection. This man was actually dealing with the subject of [journalist] investigations. He advised me to analyze a fact through the system of other surrounding events, not only to research it in itself. Like zoom out and put the problem into perspective. Let’s look at other investigations by the Bellingcat group.
Now I just list headings of the last posts at their site in a row.
Abundantly clear that British activists are too thrilled because of the Ukrainian theme. May be I would find something interesting about Odessa?
No investigations. How mean. Don’t you think this is one-legged attitude? Let’s scroll a bit more. A piece about a new North Korean missile
Indian aircraft carrier
Now Iran is up to something
An article about chemical attacks in Syria
What are Chinese constructing in the South Sea?
Comment is needless, however I’ll speak out. This is one of the first comments under that very report I started this post with.
We live during an uneasy time. An entire information war is going on. There are no good or bad in it. There are only side A and side B. There are no either any rules, or honor, or moral in it. Thus don’t you listen open-mouth and believe everything you are told. Even if you much want it would be truth. Keep calm, don’t become a mindless tool in other’s hands and don’t allow manipulating you.