I don't understand what the problem is. Type A uncertainty always exists even if we don't take multiple measurements to assess it. It all makes sense from metrological standpoint if we can consider the author an uncertainty contributor
The uncertainty associated with using clouds doesn't necessarily exist, though?
Like, my point is we have one scaling that is, say, cloud height -> giant character.
Then we have a second scaling that is, for example, car -> giant character.
Cloud height is very uncertain. It can vary by huge amounts. Meanwhile, we know pretty exactly how large a car is. So that latter scaling step has no cloud uncertainty in it. At best it has some uncertainty in it from "maybe the author drew it not up to scale", but that is more speculative and can ultimately not be excluded for the first case either, even if it's the more focused one. Ultimately, I would be more willing to believe that the clouds that day were just a little thinner or thicker than usual, as clouds just sometimes are like that. (Not like the author would have researched cloud thickness anyway)
Similarly, let's say we have one scaling using a focussed shot and angsizing to measure things and another scaling using direct pixel scaling, but the shot is less "focused". I would still take the latter above the former. Angsizing uses viewing angle assumptions that are even harder to expect a mangaka to consistently meet.
Or, if we have a 3 step scaling of focussed shots vs. a 1 step scaling of a less focussed one, I still take the one with less scaling steps. If we can't even expect the author to always draw things consistent in size to each other, we can even less expect them to get the size right between objects that only indirectly compare.
Or, one more example, say in one version you scale something in the foreground to something in the background. That always somewhat decreases the size of the measured object. And then you have another scaling method, were the measuring stick and object are closer together. Even if the former is "more focussed", I would take the latter, because it just doesn't have the foreground-background deflation.
And, generally, I also think more consistent portrayals of size take some priority over less consistent ones. Like, that's just statistics. The more often a result appears the less likely it becomes that it did so by chance.
So, like I was saying, I believe the rule should only be used as tie-breaker between otherwise equally good scaling methods and that should be made clear.
Another thing is that I think the formulation isn't very good. I feel like what is more focused is somewhat subjective in many cases. Furthermore, I think it misses the key argument that was made. The idea was that with objects of wildly different sizes, the author might have difficulty portraying both at once while keeping size consistent. If that is the main argument I think the rule can reflect that better. I think a good formulation could go more in the direction of "Scaling methods in which the size of the measuring stick is close to that of the scaled object are preferable to ones where measuring stick and scaled object are of vastly different sizes".
In total, I would propose something more something in the direction of:
One should consider that an author has to depict events in a fashion visible to the viewer. As such it can at times happen that the size of an object is displayed inconsistently, for the sake of showing a large and comparatively small object at the same time or other artistic considerations. This should be taken into account when evaluating which of several possible size scalings should be used. Generally speaking, a scaling where the measuring stick used is of similar size to the object scaled with it, is preferable to a scaling where the measuring stick and object are of very different sizes. However, this guideline should not take priority over criteria which cause a similar or greater uncertainty in the scaling, such as one method taking more scaling steps than another, one using angsizing while the other uses pixel scaling, one measuring stick having a much less certain size than the other, perspective in one method preventing precise results or general consistency of the sizes obtained by various methods. Exceptions might apply in truly extreme examples, where common sense should be used.
vsbattles.fandom.com
