> And if you look at the HSL values of the juice in product photos like the following, you'll get hue values of around 52°:

So, Just clicking around your first OJ example and checking lots and lots of pixels, in all the areas of the juice, the juice pixels seem to be mostly around 38-42, with a low of about 32 and a high of about 50.

So, as I said, closer to orange (30) than yellow (60), though also definitely in between.

> For comparison, here's the first result for "banana" in Google Images:

> It's 48°, even closer to orange than orange juice.

Most of the brightly illuminated top seems to be in the 48-52 range, the indirectly-lit side is mostly in the mid 40s, though there are some pixels right along the bottom edge that are also brightly lit (from the reflection from the white surface of the light source above) that hit around 60.

Its not more orange than the juice, for sure.

That's very interesting. We tend to use a lot of logic as a layer on top of our senses to tell us what we are looking at. One of the most interesting cases (which I can't find a reference for, so sorry) of this that I've come across was an accident in Amsterdam where the person whose car was involved in an accident swore that the other car in the accident had gone against traffic. But from the situation after the accident and eye witnesses it made no sense to believe this. But much later some security cam footage was submitted as evidence that showed the car that had gone against traffic somehow flipping into the position that showed them coming from the right and coming to a rest just like the final scene after the accident was. So, in spite of all defiance of logic the occupants of the first car had it right and the 'eye witnesses' and the accident investigators all had it wrong (and the occupants of the vehicle that caused the accident were lying and could probably not believe they were getting away with it).

I'm sure that in the larger fraction of all accidents eye witnesses will at least get the basics right and that parties are not going to lie about what actually happened. But it's interesting how two minutes after an accident with five witnesses there are probably 10 conflicting stories about 'how it happened' and it makes me wary of any eye witness testimony, especially in less than ideal conditions (night, rain, distance etc). When I first heard about the above case the explanation was that eye witnesses probably heard the crash, looked up and immediately created a mental reconstruction of what must have happened based on what they saw and then reported this as the fact rather than to simply say that they didn't see anything until the moment of impact.

I really should dig a bit more to see if I can turn up something about this case, it is very interesting given your comment because you have provided another example of what I think is the same thing.