the full color spectrum crayola box (that everyone has access to see to verify colors themselves cheaply) backed with a gray card?
There are several problems with that test; they've been pointed out before in other threads. They don't remove all of the usefulness of the test; I think your example shows that you can get a presentable image with significant overexposure, at least under certain conditions and within certain quality criteria. I'd not dare go much further than this somewhat vague qualification, because indeed, there really are things that I feel stand in the way of further interpretation:
* The subject brightness range of the setup is very low. Hence, the test isn't very representative of most real-world photography.
* Both color and contrast meander in the test; basically, the colors 'wander about' a little around average center values. This points to an influence of the digital post processing that twists the results; we're not just looking at negatives, we're looking at
scans from negatives. Of course, we're virtually always looking at either scans or prints from negatives, but this does draw into question how much we can say about the negatives to begin with.
* The subject matter used relies on pigments in the crayons and printing ink on the carton for the hues. These are all relatively low-chroma hues compared to real-world, natural scenes, and therefore this example won't really show what happens with significant over/underexposure of a real-world scene.
* The subject matter involves inanimate objects. This may sound kind of silly, until you realize we pick up hue variations a lot more if we're looking at e.g. human skin tones. If you translate the present example to something like that (human skin), you'll see that what presently feel like small variations (partly as a result of a low-contrast, low-chroma setup to begin with!) end up very significant. They'll make a perfectly normal face go from deep red with underexposure to sickly cyan with even modest overexposure. Mind you, in
your particular example, with your choices of post-processing/scanning. I'm actually a lot more optimizing on how skin colors
can hold up with proper filtering in overexposed negatives, but this
will come at the cost of some other hues in a real-world scene shifting about significantly.
Again, despite all this, moderate overexposure can still yield acceptable results; it's just that this particular example doesn't really demonstrate it. It
seems to support the conclusion, but this argumentation breaks down if you dissect the methodology. Which is something I kind of feel lame at doing, because I understand the time and effort you took in preparing it and it's one of the very few examples along these lines we ever get to see in the first place - but because it keeps popping up for years on end, I feel we should also collectively understand its limitations as well as its merits.
