Log in

No account? Create an account

Agent Teddybear

Seen from my office:

  • Someone I used to know has decided to mix the adorable teddybear look with a secret agent look. It is a very strange combination.
  • Someone out taking their robot for a walk. I think they were testing some kind of voice recognition software, as they were irritated when their verbal commands failed to get it to change whatever path it decided to take. Ahh, CMU.
Changes in drawing of faces, ponderings, and some sketches:Change towards drawing sketch-style rather than comic style: draw heads as they mask the environment first. The "front surface" of faces with that terrifying smooth gradient towards other parts of the head can and should be lightly implied - additional task of "holding non-front-surface-bits back" so as to correctly estimate position/size of face relative to head. This takes a bit of attentional and stack space, but is doable, and the results are much better.

Proportions continue to be a challenge.

Faces: not like scenes or most other objects - first, we judge them to a higher standard and parse them with a different part of the brain used for fine visual judgement, second, they have lots of smooth transitions rather than hard borders. Drawing a table or a wall is not hard - it's simple arithmitic/geometry with optional perspective, and one can spend a lot of time adding incredibly fine levels of detail if one wants. Learning to draw roundness: much tricker, more delicate, requires near-full shift to working with shades rather than surfaces.

Drawing tablet: Theoretically helpful - could use layers to sketch surfaces, pulling them off when not needed, avoid mental effort. Could even avoid hard borders around objects (there seems to be a visual effect we live with every day that "draws" infinitely-thin-but-prominent hard borders around all objects - I suspect this is done relatively early in visual parsing - past V1/V2, but before the mid/late phases - it is not actually there) - would making the brain do this itself make scenes look more realistic, or is it possible to provide similar enough context for the brain to latch on this way? We certainly lose depth of field.. Other attributes: ability to zoom and undo are awesome.

Drawing tabletのproblems: interaction between pen and tablet is very unsatisfying - too slidy. Also, software stack does not feel fully mature. Oddities with brush sizes (having brush size be relative to view size rather than image size would be very nice), tool quirkiness, colour sampler seems to think in terms of points, not areas, cat hair is greatly attracted to drawing tablet. Cats are also greatly attracted to drawing tablet (and pen) when it is in use.

Drawing as videogame, difficulty level:tograph (easy)

  • Live model (moderate)
  • Imagination of a nonreal person (hard)
  • Real person from memory (deathmatch)
Judgement scale (incidentally, I like how the word judgement has multiple spellings. The most common american spelling, 「judgment」, hurts my æsthetic sense. 「Judgement」, a British spelling, looks pretty good to me. I think Noah Webster and others of his kind really screwed up in places (although both the British and the American spellings of hæmorrhage almost make it seem unpleasant):
  • Looks nothing like the person, or indeed any person
  • Looks like a person, you might see how it looks like the person if prodded
  • Reminds you of the person
  • Immediately recognisable as the person
  • Exemplifies/defines the person - you would recognise the real person if you had this as a source
  • Looks like a photograph
I am probably somewhere between the second and third levels now when I put the time into it, on easy mode.

Behold, my latest sketches. Some CMUfolk might recognise other CMUfolk in these, thanks to facebook for providing photographs as sources.

It's pretty clear to me that I did not inherit the artistic skills of my mother, and I will likely not get a lot further than this. It's a good amusement though.