My 'computational' smartphone produces noticeable artifacts in its photos that aren't in the original scene. The degree/extent of those artifacts is dependent on the nature of the objects in the scene.
One wonders how the factual integrity of such images would stand up in court when challenged.
If an apple looks like an orange in such photos how does one prove it's an apple?
I've quite a number but most of which are not used as phones per se.
Essentially, they all suffer the same problem to a varying extent but the worst one is a Motorola Moto G6 Plus.
The problem became immediately obvious with a photo taken with this phone that I took of an intersection of two streets. It was a relatively wide shot but not that wide, it showed the intersection and one of the roads running off into the distance. However, the shot was sufficiently close to show only a single house on the corner of the intersection and even then only about half of it was visible. (This meant that all objects in the photo should have had sufficient resolution to determine easily what they are.)
The house was surrounded on both sides (i.e. at 90° in line with the intersection) with a white picket fence and behind it for about 25% of its length was a large green bush similar to that seen in many hedges and in places it towered over the fence by between about two and three feet.
Everything in the photo was clear and in sharp focus except for the bush which appeared as a large greenish out-of-focus amorphous mess. There was no wind at the time and I'd again emphasize that everything in front of and behind the bush was in focus.
This wasn't just a one-off. It was so obvious what was happening that I took several dozen photos under slightly varying conditions to determine the extent and scope of the processing error.
I could get detail in the bush (see its leaves in the photo) if I panned just above the fence (thus essentially cutting it from view) - and all else being equal. BTW, the bush could be seen through the usual gaps of the picket fence.
Frankly, I was shocked at how bad the problem actually was, especially so given my background. (I've worked in video surveillance and for years I've railed against the use of heavy compression on grounds that it mitigates against authentication (as artifacts from the compression distort the image and thus lower the probability that an object in the image can be authenticated)).
Nevertheless, I'd never seen anything quite this bad or quite as object-selective as was the distortion of that bush in images from this Moto camera.
I now use that location/scene to specifically test digital camera errors.
Shame I can't post images here or I'd show you the problem firsthand.
Given the 'variable' and unpredictable artifacts this phone/camera produces, lawyers would have a field-day by demonstrating how useless it is as a means of obtaining reliable evidence.
3 comments
[ 0.24 ms ] story [ 22.9 ms ] threadOne wonders how the factual integrity of such images would stand up in court when challenged.
If an apple looks like an orange in such photos how does one prove it's an apple?
Essentially, they all suffer the same problem to a varying extent but the worst one is a Motorola Moto G6 Plus.
The problem became immediately obvious with a photo taken with this phone that I took of an intersection of two streets. It was a relatively wide shot but not that wide, it showed the intersection and one of the roads running off into the distance. However, the shot was sufficiently close to show only a single house on the corner of the intersection and even then only about half of it was visible. (This meant that all objects in the photo should have had sufficient resolution to determine easily what they are.)
The house was surrounded on both sides (i.e. at 90° in line with the intersection) with a white picket fence and behind it for about 25% of its length was a large green bush similar to that seen in many hedges and in places it towered over the fence by between about two and three feet.
Everything in the photo was clear and in sharp focus except for the bush which appeared as a large greenish out-of-focus amorphous mess. There was no wind at the time and I'd again emphasize that everything in front of and behind the bush was in focus.
This wasn't just a one-off. It was so obvious what was happening that I took several dozen photos under slightly varying conditions to determine the extent and scope of the processing error.
I could get detail in the bush (see its leaves in the photo) if I panned just above the fence (thus essentially cutting it from view) - and all else being equal. BTW, the bush could be seen through the usual gaps of the picket fence.
Frankly, I was shocked at how bad the problem actually was, especially so given my background. (I've worked in video surveillance and for years I've railed against the use of heavy compression on grounds that it mitigates against authentication (as artifacts from the compression distort the image and thus lower the probability that an object in the image can be authenticated)).
Nevertheless, I'd never seen anything quite this bad or quite as object-selective as was the distortion of that bush in images from this Moto camera.
I now use that location/scene to specifically test digital camera errors.
Shame I can't post images here or I'd show you the problem firsthand.
Given the 'variable' and unpredictable artifacts this phone/camera produces, lawyers would have a field-day by demonstrating how useless it is as a means of obtaining reliable evidence.