You had me a little worried on that first page, where your cylinders all came out pretty much the same - and more importantly, appeared to include no convergences of the side edges at all. While you corrected this by the time you got to the second page, I did want to quickly explain why it would be incorrect for there to be no convergences whatsoever.

The only situation where a set of lines would have a vanishing point at infinity (as discussed back in Lesson 1) is when those lines run perfectly perpendicular to the viewer's angle of sight - basically running right across their field of view without slanting towards or away from them through the depth of the scene. Should that set of lines slant even a little in this regard, they would have to converge, even if only very gradually, towards a concrete vanishing point, rather than running perfectly parallel on the page.

In this challenge, since we're rotating our cylinders freely and randomly in space, we can pretty much assume that they would not align in so perfect a manner.

Looking at the rest of your cylinders around arbitrary minor axes, you've largely done a good job here in drawing your ellipses with confidence, and checking their minor axis alignments - although there is one other point I wanted to raise. Right now, you seem to be drawing the ellipses on either end of your cylinders with roughly the same shift in degree between them - that is to say, you either draw them with no shift at all, or a fairly minimal shift.

Foreshortening manifests in our cylinders (and really in all forms) in two ways - one being the shift in scale caused by the convergence of the side edges, basically where the far end is smaller in its overall scale than the end closer to the viewer. The other shift is in the shift in degree, where the far end will always be wider than the end closer to the viewer, for the reasons explained in the Lesson 1 ellipses video. Both of these 'shifts' tell the viewer how much foreshortening is being applied to the form, and therefore a combination of how large the form itself is, and how much of its length exists in the unseen dimension of depth (which cannot be captured directly on the two dimensional page).

Now, the thing is this - if your cylinder's scale shift tells the viewer that there's a lot of foreshortening being applied (due to the far end being dramatically smaller than the closer end), and the degree shift tells them that there's very little foreshortening applied (due to the degree shift being minimal), then we have a contradiction. Both of these must work in tandem, shifting in roughly the same amount (you don't have to be hyper accurate, just ballpark).

There weren't too many examples in your set of really dramatic foreshortening unfortunately, but the few that I saw - such as this one (be sure to number things in exercises like this so I can just reference them that way), had contradictions such as the one I just described. Keep this in mind in the future.

Anyway, moving onto your cylinders in boxes, you've done a good job here and have shown a good deal of growth. This exercise is all about having students train their instincts in regards to constructing boxes that feature two opposite faces which are proportionally square. We do this by testing how far off the ellipses themselves are from representing circles in 3D space (by checking whether their line extensions converge towards the same vanishing points as the box), and in doing so, we check whether the box's planes that enclose those ellipses represent squares in 3D space.

Towards the beginning you were definitely struggling with maintaining consistent convergences for the box itself, which threw things off, but as you progressed I can see considerable improvement both in the convergences of your boxes' edges, as well as the proportions we're looking to train here. So, all in all I'm quite pleased with your work here.

So! I'll go ahead and mark this challenge as complete.