Back to the drawing (painting) board

OK, so recently I made some attempts in painting on computer (meaning open a blank canvas in Photoshop and paint on a tablet). After spending just a few nights on it, I can’t help wondering ‘why haven’t I done this earlier?!’. Anyways, to help saving fellow mathematicians from making the same mistake, I’m writing this post. (okay you got me, I just want to post because I think it’s cool >..< Oh well, I guess travelling salesman has to wait q few days.)

This one was painted last night (you should blame it if you can’t wait to read about the travelling salesman problem). After reviewing pervious attempts, I found somehow I tend to use less saturated colors when shifting from canvas to digital. Hence I picked a circus scene for color training (the scene was from Disney’s Hunchback of Notre Dame album), seem to work~ Also, before this I have been sketching at a smaller size hence details a little smoother on this one, click to enlarge.

OK, here comes my complete digital painting history to date:

The very first sketch I did the night my tablet arrived (by the way, I’m using a Wacom Intuos, small). This is a scene I saw back in Evanston, when I used to take mid-night long walks along the shore of lake Michigan, there were days when moon rise from the lake, creating a shiny silver region that silently sparkles in the mideast of the otherwise dark water. My picture certainly doesn’t do the justice, but this was one of the views I have been longing to capture but can’t do with a camera. (low light + moving water) People, if you every visit Northwestern, I highly recommend checking the moonrise schedule!

After the first try (which is mainly black and white), I decided that I should get familiar with color mixing techniques, since the eye-dropper tool is non-existant in the real world. So I did a ‘master study’ i.e. pulled out a hard copy of Pixar concept drawing (originally in colored chalk I believe) and tried to get all the colors working ^_^ So here comes my version of radiator springs from Cars~

The next day I went out of my office, sat on the stairs for a couple hour with a computer and tablet, did this quick sketch of “the only interesting piece of architecture around the math department” – Lewis library by Frank Gary. (never liked the green and orange blocks on the building hence I intentionally deleted them :-P) Can’t get into any details because it was past midnight and that day was fairly cold.

In conclusion, digital vs. traditional:

1. Much better at anything to do with color (mixing, getting color from another part of the picture, etc);
2. Faster (used to take me weeks to finish a painting)
3. Unlimited resources (creating custom brushes, color won’t run out)

1. More personal satisfactory (well, I guess nothing compares to holding canvas in you hands)
2. Can actually see the brush and stroke at the same time (took me some time to get used to not seeing where my hands are, but in fact not as bad as one might think)

Anyways, I’m about a week old in this field, so don’t believe anything I said :-P. By the way, a side effect of doing this is, I have now started to think about the RGB value each time I see a colored object in real life >.<.

Hope you had fun! (I certainly did)

Simulating paintings on computer

Recently I’ve been playing around in our CS department and discovered there are actually people working in this field called computer graphics. Given that I’ve been playing with photoshop even before knowing what complex numbers are, this is quite exciting!

I guess I got a little too carried away in this homework item about coding your own non-photorealistic filter.

Anyways, I was never quite satisfied with any of those ‘artistic filters’ that came with commercial software. For example, here’s the photoshop ‘colored pencil’:



and dry brush:

Well, just believe I have tried to adjust the parameters to produce reasonable results…

I head down to MoMa and Metropolitan museum on the weekend and looked at some classical paintings with the intension of figuring out how might one go about programing it. (Since color mixes quite differently on canvas and on screen, there are actually some quite interesting problems to solve in order to make the computer “paint”.)

Anyways, here’s what I ended up with: (I would really like to make a Van Gogh filter…although haven’t gotten anywhere convincing yet…)

Pointillism: (I found this being the obvious one to start with):

real painting (Georges Seurat):

my pointillism filter:

(applied to a fairly ordinary photo:)


classical oil painting portrait:


Just for fun, I also did one for cubism:
(obviously inspired by Picasso’s violin)


my cubism filter:

Anyways, that was quite fun to program (especially for someone who haven’t touched code since high school).

In fact, I even found it rewarding in improving one’s painting skills. For example one thing I figured out was the distribution of luminance in painting is normally quite off from that of the real world, each artistic has their own ‘luminance curve’ and shifts the hue/saturation in a particular way. Umm…maybe coding filters should be added to all fine art curriculum :-P