So far, we've looked at some basic issues with the claim and the image, and now, I think it's time to .. stand back. We will also get out the
magnifying glass (with some provisos that will be outlined in Part 3), but for now, the next step is just to look at the image as a whole.
1.f. What does the overall image tell us?
Forgetting the claim for a moment, just look at the complete picture - what can we tell from it?
How would you interepret this scene if you had not heard the story? What does it look like at first glance? What time of day was it? What were the
weather conditions, and would they affect the image (eg fog, raindrops on windows, sun glare)? Are there signs of motion blur? If there are
directional light sources (eg the Sun, or streetlamps) do the shadows and lighting look correct? Are there any road signs or other things (houses,
buildings, etc) that might help identify the context?
So let's do that for the original image that was posted. Note that this image is NOT I repeat NOT the full-resolution image that was supplied later.
Here it is:
OK, so we have some rooftops, and a blob under a very grey sky - clearly the sun wasn't out, but there are no other clues as to the time of day...
As a photographer, the thing I then noticed was the dullness and underexposure. Not that unusual for a picture taken under a gloomy grey sky, but
unfortunate as it means there is little usable detail in the 'blob'... The next thing I noticed was the fuzziness of sections of the image.
Initially I thought this looked like (vertical) motion blur, but when I took a slightly closer look, the effect is strangely unbalanced - the blurring
effect is very noticable at the top of things, like the blob and the roofline, but not so noticable below - note the difference between the top and
bottom of the blob... Very odd. Could this effect perhaps come from a very badly aligned lens element?
Now at this point, there is a motivation to zoom in, but here we need to pause, for the traps in enlarging an image are HUGE.
So on to the
next stage in the analysis - how to (and how NOT to) examine the image in more detail.
First up, the image is clearly not a full-size original - it is reduced. Whenever you reduce an imge and re-save it, there are two nasty things that
- you lose detail. The smaller the image, the less detail.
- (if it is saved as a jpeg) new artefacts, especially around the edges of objects, will be introduced on resaving. FALSE detail. Detail that was
NOT part of the scene. The more compression used, the more (and uglier) false detail is produced. Not only does jpeg compression cause false detail,
it also reduces the colour quality and may introduce 'posterisation' - visible banding in color gradients. More about that later..
Even a casual glance at the OP image shows horrible jpeg artefacts, eg the haloes around the blob and the faint rectangle (which is a jpeg
'blocking' artefact) that surrounds it.
Clearly, enlarging this image on screen is going to make all of that image corruption more obvious. Take a look, this is the blob from the OP image,
placed next to a crop (magnified 200%) from the higher resolution original for comparison:
For the record, I have used non-interpolated
(more about that in next instalment) enlargement to magnify the right hand (full-res) image by
200%, and then similarly enlarged the left hand image to get to roughly the same size. I also adjusted the 'gamma' of both images (to ~ 0.7) to
better show the artefacts.
It's pretty clear why you should ALWAYS try to get hold of the full-size original...!! All of that junk around the left hand image (including the
rectangular jpeg blocking that some have misinterpreted as copy-paste evidence) came from the reduction and re-saving that was done to that image.
It's ALL false.
The full-res one does have some artefacts (and shows 'noise' that has been wiped out of the other during the reduction), but
it is obvious that the left hand image is almost completely worthless and should never have been used for magnified analysis.
But there's yet another trap!! When you enlarge an image beyond 'actual size' or 100%, you are at the mercy of the software you are using. Most
likely, that software will use one of a number of enlarging techniques called 'interpolation'. That means they try to smooth out the pixel edges by
*guessing* what would be there. MORE false detail!!!
Just what we need..
In the next instalment, we'll look at magnifying an image,
how not to do it, and some (free) software that allows you to magnify WITHOUT interpolation (like I have done above)..