Starting with your cylinders around arbitrary minor axes, by and large you've done a pretty good job here. Your ellipses are smooth and confident, your side edges are straight and consistent, and you're doing a good job of identifying even small discrepancies with your minor axis lines, which helps a great deal with avoiding the risk of plateauing as your work gets into that "good to the naked eye" territory.

One thing I do want to point out however are the cases where the side edges of your cylinders get a little too parallel, where they fall into appearing as though your intent was not to have them converge, but rather to force the vanishing point governing them to "infinity" as discussed in Lesson 1. For example, we can see this in cases like this and this. The reason these are concerns is that in this challenge we're effectively rotating our cylinders at entirely random orientations, but a vanishing point only goes to infinity when the edges it governs in 3D space run perpendicular to he angle at which the viewer is looking out into the world. We don't actually control the position of the vanishing point - just the intended orientation of the form, which dictates the vanishing point.

Given that random rotation we're going for, we're much safer simply assuming that the alignment will never be so perfect, and thus the edges will always have some convergence to them. There are a lot of other cases, like this one where your cylinders appear to be fairly parallel on the sides, but there's enough convergence to keep you from falling into this issue. I'd recommend using that as your general "minimum" amount of convergence.

Continuing onto your cylinders in boxes, your work here is looking quite good. This exercise is really all about helping develop students' understanding of how to construct boxes which feature two opposite faces which are proportionally square, regardless of how the form is oriented in space. We do this not by memorizing every possible configuration, but rather by continuing to develop your subconscious understanding of space through repetition, and through analysis (by way of the line extensions).

Where the box challenge's line extensions helped to develop a stronger sense of how to achieve more consistent convergences in our lines, here we add three more lines for each ellipse: the minor axis, and the two contact point lines. In checking how far off these are from converging towards the box's own vanishing points, we can see how far off we were from having the ellipse represent a circle in 3D space, and in turn how far off we were from having the plane that encloses it from representing a square.

In checking your line extensions correctly, you're continually getting feedback on your estimation of these proportions, and in paying attention to what those line extensions tell you about the success of your approach, you've been able to gradually refine your estimation of those proportions quite well over the set. There's certainly more room for improvement, but as it stands, you appear to be able to achieve proportions close enough to satisfy the naked eye, regardless of the orientation of the given box in space, which should ultimately come in quite handy throughout the next lesson.

One suggestion I do have however is that as your boxes are by and large focusing on fairly shallow foreshortening, I would recommend throwing more dramatic foreshortening into the mix. This will ensure that as your skills continue to develop, that they will apply more evenly across all possible circumstances.

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