@Joe: I just need more practice, that's all. Drawing only from life after the methods presented in Betty Edwards' book however is not sufficient practice because that only trains, as you know, ones' own perception and ability to see whatever is there but only as flat 2D shapes, as the contours which are formed as a projection of the real world volumes onto a flat surface. The result appears 3D if the perception and translation to pencil strokes was accurate but it does not mean the artist understands the volumes or that he/she could render them accurately from a different perspective without the help of a different reference photo.

I am aware of the fact that it isn't necessary to have any constructional knowledge to arrive at good life drawing skills (given many hours of practice of course) if you just draw what you see and keep making sure you really see it right. But that is not the main focus here.

The main focus/goal is on arriving at an understanding of that which is there as 3D volumes of width, height and depth and to be able to quickly render those from any angle from imagination, even angles which are completely different from the referenced real world object and angles from which an object hasn't even been seen before by extrapolating from what is already known and by imagining how it might appear from that unknown angle.

Now, if, and I stress IF I already were an expert at that, at spotting and correctly placing/feeling the volumes as they extend into the space in all directions, there would not be much difference in either drawing what is seen as a flat 2D image or drawing what is seen as 3D volumes first and then rendering the contours, lights, shadows and halftones based on those volumes.

The latter would have the advantage (over a mere 2D copy) of enabling the artist to render the referenced real life object(e.g. a photo) with a different light source or from an entirely different angle.

@wishie: Posemaniacs is great (worked from that in 2009 a bit (as seen on some older page of this thread)) and I should also get back to working from that again. I'll also check out the other site you linked to which seems interesting as well (and it features real naked people too!).

Working at bigger size is also a good idea (although in theory it doesn't make a difference since an accurately constructed or copied image should feel right at any size).


Now here are the results of todays practice.

First is a drawn comment on "your images are rife with symbolism":


Second, some 30sec posemaniacs studies:


Third some quick doodles trying to imagine how that same old pose would look from different angles (see... if I already had the skill I'm training for, these would look right instead of awkward ):


And fourth, oldschool flat drawing practice from reference photo without any construction practice involved (failed to capture her likeness again, I blame a lack of practice ^^):

process video: http://www.youtube.com/watch?v=9u0U2hq13tw

result:


And finally another deskshot (with a new solution for fixing the camera in a place where it doesn't get in the way):

Hrrmmmmpffff.

I tried to make a construction based on the photo but then quickly caught myself alternating between constructing and copying. My construction however wasn't perfectly true to the angles of the head in the photo, so the result ended up falling apart with random features placed based on my construction and others being placed and drawn as referenced from the photo.

Also, the end result does not look like her at all and while drawing I often found that even very minor differences of less than an estimated two degrees in an angle already made a huge difference in the overall appearance. Same seems to be true for line thickness: e.g. That one thick line almost seems to make the lower lip appear 1.5 times as high as it really is.

I could probably digitally keep adjusting it until perfection (or directly draw on top of the photo) but that seems pointless, so yay, another abandoned study which once again proves that a flawed construction is the root of all evil. So I think I should really practice construction more than drawing from life. Or perhaps the simplified head (ball and plane) construction is flawed in itself because it's not an accurate simplification of the skull?

I intended to draw a full portrait of her but I felt intimidated by her beauty and thought there was no way I could be able to capture that with my rusty and crude pencil-fu, so I just drew an eye. Even in that little portion I made a few mistakes and got some angles and stroke lengths wrong. The most obvious error is the fold which covers the upper eyelid, it seems a little swollen in my drawing thus already destroying the perfection found in the original. Also, it does not seem to wrap as softly around the bone as it does in reality and just ends abruptly on the left with a sharp edge.

Furthermore, I misinterpreted the dark stuff on the lower left as lashes from the lower lid where now after comparing my drawing side by side with the reference I think those dark lines are actually shadows cast by the lashes from her upper eyelid. Also, the dark lines inside the colored part of the eye are probably reflections of the lashes (got that wrong too while drawing and assumed seemingly random pigmentation).

Ashbad, the hard way would be to be a creeper and draw people from life outside (Public transportation works well for that but be prepared to get some angry looks when you stare at people analyzing their features. If you wear headphones and listen to loud music you can at least prevent them from actually starting to talk to you. After some time you can move on to sketching them in your head, still being a creeper staring at them but without the additional awkwardness which arises from them seeing you scribbling down lines into your sketchbook while you keep looking up and down.)

Other than that, others have suggested studying Loomis which is what I'm currently doing, reading "Drawing The Head And Hands". I'm still at the beginning of the book but already learned a couple of things which helped improve my head drawings. Of course, posting your drawing on the web and receiving critique is also a good way to learn. Avoid family and friends for critique for they tend to be mild on you and dishonest in that they won't point out your flaws because they're too afraid to hurt your feelings.

----
Today I tried to make an arbitrary construction without looking at the 3D model. This time I started with the ellipses which form from the sliced off sides of the main ball of the head and then constructed the head around them. Afterwards I looked at the 3D model and picked the orientation which seemed to match it most accurately (which turned out to be D3). Based on that I found too many flaws and after a bit of trying gave up on correcting all of them which leaves me with the abandoned version on the far right:


(this proves once more that if the construction is flawed, no amount of tweaking and screwing around on the individual parts afterwards will be sufficient to make it look right as a whole)

My mind seems to be controlled by some strong force of habit which always wants to go back to rendering the face from a straight symbolic front view and it also refuses to accept that some things are hidden underneath others and thus should not appear in the drawing at all, it's like the mind does not like leaving out anything and wants to spread out and draw even the things that lie covered around and behind an edge. This becomes most obvious whenever I get towards drawing the mouth and chin region.