I think we should not get befuddled by what things look like or what we might compare them to visually, but more straightly checking the properties, wanted and unwanted, that a knot may provide.
There is also an important feature to consider here, friction, more precisely the friction between different parts of the rope that is knotted. Friction may be one of the most important features to consider in a rope, apart from breaking strength, elongation, water uptake and other. Now Dyneema, HMPE, is a very special case when it comes to friction.
So regardless of whether the ends come out perpendicular to the standing parts or not, friction is what holds things together in a knot, and HMPE is a very poor companion when we want friction. In any knot, there is always a force component along the rope axis at any point within, and this force tries to slip the free end into the knot to undo it. If friction is sufficient to withstand said force, and if friction builds up more than the length-wise force when applying load, the knot will hold, it will not slip back and undo itself. The most important property for this is friction, and HMPE does not provide much of that property.
So discussing whether the ends come out at right angles or not is not fruitful, as the important question is: Will friction within the knot prevent that it pulls the ends back into it and undo the knot? This question is crucial, whether the ends are aligned with the standing part or at other angles to them. Only testing will confirm or falsify any claims. Many here might have seen how a doubled bowline in HMPE will undo itself at a much lower force than the breaking force of the rope.
It is interesting to find knots that work in HMPE, and I think that factors that must be addressed also are ease of tying, and ease of undoing. Would these not be important, splicing is the answer. We know that splices can provide enough friction - also without any right angles.
We may also sometimes accept a "second best" solution. One property we look for when using HMPE, mostly is little elongation, while ultimate breaking strength might not be overly important, as the load will mostly be much less than the SWL of the rope. So if the knot will not slip at 1/10 of the SWL, it might still serve its purpose, if it will never be loaded beyond that point.
HMPE in boating is used, among other purposes, for halyards, where it is often spliced. It can also be used for sheets, although less frequently. The main important feature in halyard use is low elongation, and low weight comes second.
Neither halyards nor sheets are frequently knotted, but spliced to a hook of some kind, which is attached to the clew. Only in an emergency would sailors use knots, and very seldom would a bend be required. True that I use the bowline for attaching my sheets, but they are not HMPE. Also, if using HMPE, mostly it would have a braided polyester sleeve, and as the SWL is not even closely approached in normal use, but it is used mostly for the short elongation, it can be knotted with normal knots in normal use.
Nevertheless, it is interesting to find knots that work in straight HMPE, and the zeppelin formation might be a viable starting point. When I tried different sizes, I found that in lines with more friction than HMPE, doubling the first round turn does not achieve anything more than complicating things. It is the final turn that should be doubled for lines with more ordinary friction. Whether this holds true for HMPE should be tried before anyone puts anything important at stake by knotting unsheathed HMPE.
I think that simplicity is also an important feature with knots. If the knot is easy to tie, simple to inspect and thus "minimalistic" in appearance, those are features that might also be of great importance, as one wrong tuck will make a totally different knot, or perhaps no knot at all.
So for HMPE, i never tried bending it other than stuff with a PE sheathing, which I used for short slings, bent with double fisherman's (grapevine) knots. I have tried such a sling with my hydraulic jack, and I failed to break it or pull the core back.
I haven't tried to attach such a sheet or halyard to a sail, but if so, I would favour the #1843 anchor bend form, preferably with a snapshackle. This knot may also be less prone to slip back with an extra turn if needed. I think it should be tried, but it might behave much like the doubled bowline.
http://web.comhem.se/~u77479609/anchor_bend.htmlTwo of those hooked into each other would be reminiscent of a doubled zeppelin. Whether this would be a viable knot or not, is something to try out for anyone interested. I would prefer something else than HMPE, but in an emergency, a bend that really works could be a boon.