Good catch,
@alastairstevenson
And no,
@Fastb. I applaud your desire to give this guy the benefit of the doubt. But Lightroom will NOT change the EXIF shooting data (at least on its own) to compensate for "digital zoom". This is for a number of reasons, but one big one is that we always want to know the original shooting information when viewing our or other people's photos. Another is that since this "digital zoom" is actually just cropping and resizing, making such an adjustment would be impossible to do accurately because so many steps could be used in the process. I have Lightroom CC due to it coming along with Photoshop these days as part of the subscription model Adobe now uses. While Lightroom may well be fantastic, I don't use it much because, quite simply, I'm more used to and productive with good old Photoshop.
Still, it's nice of all of us to give folks the benefit of the doubt whenever possible.
But, to me, the smoking gun here is this:
The photo posted as "Stock" has the following characteristics when I examine its EXIF in Photoshop:
The actual time the photo was taken is shown as: 2019-07-22 16:31:11.001
It says it was taken using an 18-300mm f/3.5/6.3 lens set to 21mm focal length
And even more proof is that the exposure info for this shot is:
1/1000th sec; f/22; ISO5000; Manual; Pattern metering
On the other hand, the photo posted as "3x zoom" has the following characteristics:
It was shot with the 18-300mm f/3.5/6.3 lens, but set to 300.00mm this time
And now, the original file date/time is: 4:31:22 pm (so this was taken 11 minutes later than the first (wider) shot).
And the exposure information for this shot is:
1/1000th sec; f/22; ISO 12800; Manual; Pattern metering
That means that he shot the second image 11 minutes later, using the same lens, but this time zoomed out to its maximum of 300mm (instead of at 21mm like the first shot) AND, while he may have tried to shoot with the same exposure settings, the camera was set to auto-ISO mode, and in order to expose for the shadow in which the car in question was located, it bumped up the ISO from 5000 to 12800. So very clearly, the two "digital zoom" images were NOT made from the same image.
He cheated and zoomed the lens from a relatively wide angle setting (21mm) to a relatively long focal length of 300mm. Further, he shot with a different exposure setting (even if he didn't realize it) so the dark shadow area where this car's license plate was at was more properly exposed.
The "5x zoom" image does appear to simply be a cropped and upsized version of the second image. You can really see the digital noise in the image shot at ISO 12,800.
I'm not sure what the guy was trying to prove or figure out by doing this. In effect, what he did was prove our point that you need a camera zoomed in (longer focal length lens) and set for the proper exposure if you want to read license plates effectively. He made OUR point for himself, but then still didn't want to accept it. That's sort of odd if his initial intent really was to figure out how to do LPR and/or facial recognition.
You cannot pull the license plate info out of the first (as he called it "Stock") image he made at 21mm with his DSLR. Even it doesn't get the job done. Not even close.
Pinch Point(s)
Enough "pixels per license plate" (or as we used to say in the macro-shooting forums "Pixels Per Ladybug").
Proper exposure.
The problem at my place is that there is a buttload of traffic past my house and I don't have any good vantage points to do proper LPR. Still, I'd like to play with it, but the angles for the most common traffic will be at about 45 degrees from the plates at best. It may be doable, but may not be worth my effort here.
On the other hand, the OP may well have a way to get a good zoomed-in camera aimed at cars entering and leaving his parking lot. He just has to accept that this will be a separate camera from his overview camera.