That's great! Thanks! I'll certainly take a look at that.
As for whether the software actually works:
What it tells you the "dynamic range" is will depend on what thresholds you set for dynamic range. (Not sure if the video covered that so I'll just redscribe it here. It took me a while to understand this part) Essentially the software takes -3EV, +0EV, +3EV shots and does a sort of HDR type thing on them, assigning output tone values from 0-255 and, for each tone, the +/- EV of the patch with that tone. (The -2EV patch from the -3EV shot is probably considered the -5EV patch, and I'd bet that it's just extrapolating beyond -5EV.) So you get 118 or so around EV0, maybe 50 at -2EV, etc.
The software is configured by default to consider whatever it determines for level 20 as the low end EV for "dynamic range", and whatever it finds for 245 as the high end EV. So what you actually get for "dynamic range" and "clipping range" is going to depend on these settings. Defaults are:
The Middle T one, dynamic range (-), clipping point (-), clipping point (+) and dynamic range (+) settings will differ depending on how you think about image reproduction. At Sekonic, we use the following values as our default standard values. The numbers in parentheses indicate the standard range.
Middle T one 118 (113~123)
Dynamic Range (-) 20 (15~25)
Clipping Point (-) 35 (30~40)
Clipping Point (+) 230 (225~235)
Dynamic Range (+) 245 (240~250)
So basically what this means is that they just assume
that at 20, for example, the tone response plot for your sensor takes a bend toward a very gradual slope that would keep you from getting any real detail (anything but highly posterized "detail" such as 10-20 covering the range of 1EV or more).
It does not do anything intelligent like "gee, at level 20 the plot goes from something like a line with a steep slope to something close to a line with a very gradual slope so therefore this must be a good point to consider the -EV side of the dynamic range". It just assumes "20" which may not in fact be the point in the curve where it significantly changes slope.
I don't know if they didn't try or if they just couldn't come up with any objective thresholds for the amount the curve has to change to consider it "no detail" below that point.
Essentially what you have to do is just look at the curve yourself, and tell it your opinion on where it becomes so gradual that anything below/above is not useful. (You can slide marker lines around the graph to set these or type the values in.) Some perfect camera out there might have a perfectly linear response from 0 to 255 for all I know, so in that case their defaults would obviously be bad assumptions.)
So it basically just gives you a (hopefully accurate) curve and no objective definition of dynamic range or clipping range based on some statistical analysis of the curve.
If they really want to improve the software here they should put some continuous gradient or pattern or something on the target somewhere, then show you a zoomed in area of detail on that to give you an better idea of how bad the posterization gets at what level. A sample of the actual effect on the image detail in that range would be more useful than just a line on a graph. Right now if you want to do this you'd have to load the images into Photoshop or something and try to figure out the best threshold based on what seems acceptable to you as you mess with the levels controls to expand just the extreme light/dark areas to the full output range. (Like use levels to do 0-30 -> 0-255 and look at posterization.)
Anyway, the short answer is that the 5D might have such close-to-linear response that the +/- thresholds should be expanded to 10-250 or something instead of 20-245.
Obviously if you're a Canon marketing person then you're going to have the convenient opinion that the detail is just fine in the 2-254 range or whatever you can get away with and make your dynamic range claims based on that. That is, unless there's some response slope standard for these ranges that I don't know about.
So I guess whether it works depends on whether "works" means doing something other than default level thresholds for these ranges.
In any case, the only thing that these range numbers affect is where the fancy light meter will put some little marker dots on the LCD display at a given ISO setting. It may suggest something based on these default numbers, but you can just tell it to display whatever you want.
The part that it should be able to do properly by itself, which is hopefully accurate, is determine the mid tone and, based on what you told it your light meter said for mid tone, tell you how inaccurate your light meter is and what EV compensation to set in it to "fix" the error. So if my light meter says 1/60 f/11 and the software finds that the middle gray patch is -1EV from what it should be, then I know I need to tell my light meter "+1EV" to calibrate it. Of course you need their fancy light meter because it lets you set a different value for every single ISO setting, but I found that on my camera with my light meter it didn't really need any EV compensation at any ISO setting.
Of course it isn't really THAT hard to do this by just bracketing a bunch of shots of a 50% gray card, looking at all of them in an image editor, picking the one with the gray values closest to 118 (sRGB), taking the exposire settings from that, comparing them to what your light meter says, and calculating the EV difference yourself. The sekonic stuff just automates this process while only requiring 3 shots to do it at a given ISO.