After finding some great examples of Human Shadow Residue at the base of the Sagrada Familia (see previous post), I began looking at other tourist attractions with a lot of foot traffic. I also wanted to examine places I had visited in the past, so I started off in London with St. Paul's Cathedral (a personal favourite). There was some HSR, sure, but there were also non-erased people sitting on the main steps and various benches. Some of them had a smidge of dimensionality, but most of them were ghostly 2D images.
Curious, I thought of another attraction nearby that would surely have a line up - the Tower of London.
Sure enough, taking a look at the Tower using Google Maps (above) revealed a long lineup, a substantial crowd around a small plaque or statue (a light Googling indicates that this was the spot where execution scaffolding was erected), and plenty of people scattered about, walking or standing, and also sitting on benches.
Switching to Google Earth, above and below, we see that an interesting thing has happened. There is plenty of HSR about, but just like at St. Paul's there are ghostly images of people; we can see the colours of their clothes and perhaps a little hair colour.
The screen grab below reveals something new, however - gelatinous blobs of people-clumps, skinned with the same kind of ghostly, coloured imagery. It's as if groups of people were melted together into misshapen piles which then had some of the colour drained out.
The variation in the height of these crowd globules is very interesting to me, and also the question of why they erupt in some some areas of congregation but not in others. The above is a great example of this. One blob near the front of the line reaches human height, but other lumps look like they only reach a couple of feet high or less. Plus there are sections of the line with no height at all, just those ghostly people-images painted on the ground.
As I have speculated in the case of the amorphous crowd blobs painted with Human Shadow Residue at the Sagrada Familia, this is likely due to the technology Google Earth uses to determine elevation (see How accurate is Google Earth elevation? MES, July 2018). For ease of discussion, let's say that satellite radar is used to figure out elevation. When there is a crowd of people in an area, a physical mass is created that has some stability over time. The radar samples heights in that area multiple times as it is gathering data; examination of multiple shadows of stationary objects suggests this happens four times (assuming that radar and image data are collected at the same time). The fact that multiple shadows are discernible at all suggests that there is a not insignificant time delay between each radar sampling/measurement. That's not a problem when it comes to an immobile building, but crowds are different. They move and flow and change shape. That means Google Earth has data snapshots of four different crowd configurations, which are then combined to produce the summed data for that area. When a particular point in an area is occupied by a person in each of the four samplings, the radar believes there is an object there of human height. That object has to be skinned with something, and the only visual data available are multiple images of crowd configurations that have been captured along with the radar data. Because each image only constitutes 25% of the full informational package they have to work with, any image of any individual will only have 25% opacity. Unless, of course, that person stood perfectly still for all four samplings, which is not very likely (though maybe I should try to find some really committed Buddhists who like to meditate outside.) And so a weird overlapping melange of people images is created, and used as skin to coat the 3D model determined by the radar passes. They must have some rule in the algorithms that figure all this out, along the lines of "If the point is only occupied two times, calculate the height as one half." That would explain why some of the lumps are only waist-high or knee-high.
It makes sense that some people sitting on benches would look a little more solid, as the odds of them being in the same place at least twice during data collection is higher.
It looks like the algorithm removes individual people when they are standing alone, leaving just their shadow or part of it - classic HSR. In some cases, like at the Sagrada Familia, that process is also extended to groups of people, such that all that's left is an overlapping blanket of shadow fragments. For some reason, in England, they decided to leave the images of people in, letting each person's ghostly 25% contribution to the final picture remain and constitute a vague humanity-paint that is applied to imperfectly constructed crowd globules.
Or something.
This crowd on Tower Green conforms roughly to the bell curve in two ways - from above, in 2D, and in profile, in 3D. The profile curve supports the idea that heights are calculated statistically, i.e., averaged over four data points. Note, too, that from above, the individuals in the crowd shift from coloured imagery to HSR as they leave the center mass and hang around the periphery. That's a weird decision to me. Pick one or the other, either all Human Shadow Residue or all Human-Painted Crowd Blobs.
Also, some great floating trees on Tower Green. Worth a look for those, if nothing else.