Starting with your cylinders around arbtirary minor axes, the biggest thing that stands out to me is the fact that your cylinders don't really appear to have been drawn with very much variation in terms of the rate of foreshortening being applied.

Right now I'm seeing that your cylinders either are drawn with a bit of foreshortening, where the far end is a both a little smaller and a little wider than the end closer to the viewer, or drawn with the far end basically the same size as the closer end. Most of them fall into this second category, which is incorrect for two reasons:

• Firstly, the change in degree (with the far end being wider than the closer end) works hand-in-hand with the change in scale (far end getting smaller than the closer end) to represent foreshortening to the viewer. You're not going to have one change really dramatically, and the other change only slightly. Similarly, you're not going to end up with one changing a little, and the other not changing at all. These scenarios would result in a visual contradiction to the viewer, where the inconsistent representation of foreshortening tells the viewer that there is either more or less distance between the ends of the cylinder (depending on which sign they're looking at). That's more or less foreshortening's job - to help the viewer determine just how long a cylinder is.

• Secondly, the only situation where you'd end up with no visible foreshortening (which is basically what you're asserting with the lack of scale change from one end to the other, and the side edges remaining parallel on the 2D page), is when the cylinder itself is perpendicular to the viewer's point of view (basically what we discussed in Lesson 1 about what makes a vanishing point "go to infinity"), not turned towards or away from them at all. This is a VERY unlikely occurrence unless you're specifically aligning your cylinder that way on purpose. In this challenge, we're rotating our cylinders completely randomly, and so we can pretty much assure ourselves of the opposite - that we'll never end up with this specific orientation. Now, if the cylinder is not in this specific orientation, then having no foreshortening tells the viewer that there is 0 distance between its opposite ends. Of course, we can see right on the page that these cylinders have length to them, resulting in another visual contradiction.

The cylinders that fell into the first category - that is, those that have a little foreshortening to them, such as numbers 93, 96, etc. - are fine, but you have quite a few more that appear to try and keep their side edges parallel on the page.

Moving onto the cylinders in boxes, the same issue seems to occur, where you are much more obviously now trying to keep the side edges of your boxes parallel on the page. Compare this to your box challenge work where you actively and correctly focused on how your lines converged (whether rapidly or gradually) towards a shared vanishing point, it's very clear that you went off in the wrong direction here.

As far as actually checking your ellipses' own lines against those of the box (that is, identifying whether the ellipse's minor axis and contact points align towards the box's own vanishing points in order to identify whether the ellipses represent circles in 3D space, and in turn whether the planes enclosing them would therefore represent squares in 3D space), you did a good job - but this key issue of completely eliminating this overarching principle of perspective and drawing in 3D space really overshadows that.

To put it simply: you just can't eliminate the convergences on your lines to make things easier to draw. Convergences are eliminated specifically by the orientation of the form you're drawing, not by your own choices.

I'm going to assign some revisions below, just to make sure you understand this, and to give you the chance to apply the exercise in more correct circumstances. Also, don't forget that the assignment itself did mention that you should include lots of variation for foreshortening, in bold.