tasuki’s blog

It’s been happening for longer time. I’m getting stupid. As in “not clever”. I’m learning new ways to solve specific problems, but my general problem-solving skills are very bad nowadays (in the timeframe of several years, not specifically at the moment (actually work helps a bit, I guess)).

I used to be good at solving easy problems that many people get stuck at.

Let me give you an example… There are two ways of making a thumbnail from a picture:

In the first one, you get width and height of the original picture and the boundaries (width, height) into which the thumbnail must fit. You resize the picture so as to preserve the width/height ratio. What is the width and height of the thumbnail?

The second one is trickier, but still falls into “must be easy” category:

Again, you start with width and height of the original picture, and width and height into which the thumbnail must fit. The difference is that now you must fill the whole thumbnail and cut off the exceeding parts. Suppose you want to have the thumbnail in the center. What are the left x-coordinate/y-coordinate and width/height of the area to cut the thumbnail from the original picture?

In case you say that I’m so dumb I can’t even describe the task, here’s a picture that (hopefully) explains everything.

Ok, now go and solve it. Seriously, take a pencil and paper, or fire up your favourite text editor and solve it. In case you “know it’s trivial but can’t be bothered to solve it”, don’t even bother to read on.

This kind of problem that doesn’t require (almost) any prior knowledge and has simple input and simple output is the kind of problem which used to be much easier for me to solve than for most people. Well, not anymore. And I wonder — can I do anything about it?

I thought that various logical puzzles might be a good way to force my brain to learn to solve problems again. So I searched a bit and found some excellent puzzles: blue eyes, three hats, and prisoners with hats in a long row (of course the comments under the post are wrong). The puzzles do not require any kind of “trick” to solve.

PS: Please don’t post solutions to the thumbnail problem. If you solved it, you know you solved it. If you are not sure, you haven’t solved it.

For newbies, by a newbie.

• It doesn’t really matter what kind of a camera you have (ok, no cell phone camera, you need some kind of a lens).
• Most modern cameras can decide the exposure settings just fine in 90% of cases. Nevertheless, you should learn to operate in manual mode too (for example, the closest distance my camera is able to autofocus at is about 15cm, using manual focus makes it about 2cm (it used to be my dad’s camera and he thought for two years that 15cm was the closest it could do)).
• Flash is evil. (funny, exactly the same rule applies in webdesign)
• The only exception to the above rule is using flash to light nearby dark objects in an otherwise well lit scene.
• If you do not have enough light, feel free to increase the ISO speed. Or bring a tripod. Or emulate it somehow (don’t forget to use a short timer to avoid touching the camera when it fires).
• By far the most important thing is the composition. (there are also other good resources about composition)
• Actually, there are no rules for composition. But it is still the most important thing.
• Postprocessing is your friend. (the examples here use GIMP, but similar techniques will work in other image editing software too)
  • Three alternative ways to do a Very Important Thing ™:
    • the easiest and fastest: Colors > Levels > Auto (this works when you have both real “white point” and “black point” in your picture)
    • slightly more control: Colors > Levels, and play with it (you usually want to leave the output levels alone and shrink the input levels according to the histogram, you can set this individually for each color channel)
    • and finally: Colors > Curves (this is like Levels but with more control — you do not only set the starting and ending point of the curve, you can also change its shape (slight S curve often improves the image a lot)
  • Orton effect (aka dreamy look): duplicate layer > Filters > Blur > Gaussian Blur (set something reasonable depending on your picture size, experiment) > set layer mode (multiply, overlay, hard light, or whatever… experiment) > set layer transparency (uh, guess what… experiment?)
  • Tone mapping: duplicate layer > duplicate layer > Desaturate > Invert Colors > Gaussian blur > Layer opacity to cca 70 > merge layer down > set layer mode > set opacity
  • Vignette effect (aka lomography): new layer > E (elliptical regions) > select accross the whole picture > invert selection > Select > Feather (set to something reasonable, experiment) > fill selected area with black > set opacity (usually 20 or so)
  • And while all these effects are great, try not to use all of them at once.
• Finally, publish your photos on the internets (flickr, picasa, etc).
• Flickr is still the best photosharing site around (even though they have the bloody videos now — which even started polluting Explore), it has the most amazing community and most talented photographers
• By the way, all that I know about photography (ok, not much) I learned on flickr, just watching Explore, and sometimes reading the comments.

What a mess of a post.
On the other hand it does contain plenty of good info, it’s just a bit too much concentrated.