That Hex game looks good... is it availible for the TI89?

Hope it is or will soon be ported....

1

It looks like the graphics are too big, and would have to be made smaller.

Darn... people need to make stuff for the TI-89 and THEN port it to the TI-92+/v200. It's much easier that way. Oh well, still looks nice.

It will definitly ported to the TI89. I just made it full-screen on the v200 because that's the calculator i own myself. The graphics will look nearly the same, since the real playing field will fit into the smaller display (i hope so...). Planned features are also AI and link mode.

Really? Great!

NO. Enough stuff is 89-only and you have to bother with a smaller screen and screwed up keys. Make it for both.

I agree. I hate having to make my programs 89-compatible, since you'd need to write both separately or use only part of the 92+/v200 screen, which gets really annoying...

I challenge everyone to make a Earth Illumination Simulator faster than Sun. I'm still using floating point mathematics, so there is a lot of calculating time to reduce by using integers.

Good Luck!

Interesting ^^

Interesting program Björn and so is the challenge too...

This 'sun' thing...is it like that 'life' thingamabob I keep hearing about? :D

Awesome programs.

It shows how the sun illuminates the earth at specific times. Does any one know if it factors in seasons?

If I understand you correct, you wonder if the program supports summer, winter, etc.

Well... Earths' declination (angular tilting to the sun) is the year dependent variable, that is set when defining the specific month.

Satisfying answer?

Thank you.

i'm just glad that i never have take that risk of exposing myself to sunlight again, thanks!

In trying to write a program like this, I ended up with 5 multiplies and 2 trig functions per pixel plus 2 trig functions per pixel row. A rough estimate yields about 2 seconds per frame at 128x64 pixels (I'm doing this for my TI-86, which is _slower_ than the TI-89/92+).

How many multiplies and trig functions does your program do? It seems awfully slow for drawing only 8x8 pixel blocks. Also, why does it need the sundata variable? Can't the program compute the position of the sun on the fly? It's not that difficult.

I'm taking your challenge, by the way.

Well...

This program only calculates the value of four trig functions (neglectable). The time consuming part is the 20*13*8 floating point multiplications. 20 is the width, 13 is the height and 8 for the zenith angle calculation.

I think I have explained the routine in the included PDF-file.

The sundata contains the compressed background picture and precalculated sin() and cos(). As explained in the PDF, everything else is computed on the fly...

I actually just skimmed the PDF file, as it contains a lot of math -- a lot more math, in fact, than I thought was necessary for such a simple problem.

Anyway, I first wrote this program in TI-86 BASIC (before my comment above), and it takes only about 50 minutes to draw the whole screen pixel-by-pixel! It uses a simple error-diffusion dithering, but my ASM version will use Bayer dithering for better results.

So...

You think it contains a lot of math? How did you derive your equations?

And how did you manage to make it so fast? Infact, you let the CPU compute approx 1e5 times more operators than my version.

It'll be interesting to see your results.

The math in mine is quite simple. First it finds the x,y,z vector (length is 1) of the sun. Then for each screen point it finds the corresponding x,y,z vector on the surface (just a multiply per axis).

If the dot product of the earth and sun vectors is positive (they form an angle < 90 degrees), then the sun is visible at that point on earth and therefore is lit.

Currently I have a working prototype, and in VTI it takes about 2.5 seconds (close to my guess of 2) to draw the whole screen. I just have to clean up the code a bit, add some user interface stuff (to set the time and date), and write a Bayer dithering routine before I upload it.

Good news, everyone.

I just reduced 5 mulitplies down to 4 per pixel plus 1 per row. I also wrote a Bayer dithering routine from scratch which worked on the first try (then I optimized it :).

It renders the whole 128x64 pixel screen in about 3 seconds on a real TI-86 (which seems to be a little slower than VTI).

I still need to write a user interface to allow the user to input the month, day, and time.