. /../Known Noctis IV Bugs, and.../ 1..1819202122..71
written by Doozer on Feb 07, 2005 20:56
Something is wrong with the compass. If I face north, then turn 90 degrees clockwise, my compass shouldn't read west.
doing pushups
written by Megagun on Feb 07, 2005 20:59
It's made like that, because Noctis itself is bugged in that way. Before, if you would go "east", you would actually go "west" (on the lat/lon display thing)...
!?
written by Ginrai on Feb 07, 2005 21:02
*digs around the forum*

Here's what SL said on the issue a little earlier in the thread:

Sl said:
As for the compass, it's not broken, the e/w directions had to be reversed because Alex did something bizarre with the code. Perhaps because he's calculating x,z from -sin,cos instead of x,z from sin,cos or x,z from cos,sin. Or maybe it's because of something else. I don't know.

Specifically, right now, if you move NORTH, your Z decreases. If you move SOUTH, your Z increases. If you move WEST, your X decreases. If you move EAST, your X increases. To reverse the e/w directions so that they're "correct," EAST would have to DECREASE X and WEST would have to INCREASE X - but that's not the case. Go to the landing-sector map in the stardrifter and press the right arrow (to move the landing point EAST one sector), note the X coordinate increases by 1 when you do that.

Something's mirrored or something in the original code; I merely made the compass match the directions. It looks wrong because to do so, I had to change it from n/e/s/w to n/w/s/e.
written by Doozer on Feb 07, 2005 22:10
How delightfully confusing. Evidently I don't understand why we can't just call an increasing X "east". Or are you saying that when you go East (by the common definition) you actually end up going toward the sector to the West (by the common definition)?
written by Mr.fop on Feb 08, 2005 13:17
Yes, that how it goes :/ I'd prefer making the compass turn in the wrong direction, but so that when you turn face north and turn right, you face east. Moreover, I'd like to replace the directions with arrows.
whatever.
written by Bgreman on Feb 09, 2005 15:25
As I noted, Noctis uses a left-handed coordinate system for some reason. That's why he uses -sin, cos instead of cos, sin, as is standard for a right handed coord system.
absent more than alex
written by Shadowlord on Feb 11, 2005 07:07
P.S. In an entirely unrelated note, in my own programs I've always used (sin, cos) instead of (cos, sin). Why? Because it makes an angle of 0 point down instead of to the right, and pointing to the right for 0 feels unnatural to me. Maybe Alex used (-sin, cos) for the same reason, though I don't know why he used -sin instead of just sin, unless it was to make angles proceed clockwise instead of counterclockwise. I'd guess he didn't know about the compass being b0rked, but I didn't even really notice it until I had implemented sector transitions and was trying to move in one direction by using the compass, but was really moving in an entirely different direction.
written by Mr.fop on Feb 11, 2005 10:07
This is my trigonometry books take on the unit circle (or that's how we call it):

As you see, an angle of 0 would point to the right (coordinates 1,0).
absent more than alex
written by Shadowlord on Feb 11, 2005 14:42
If you were saying that for my benefit:

Shadowlord said:
I've always used (sin, cos) instead of (cos, sin). Why? Because it makes an angle of 0 point down instead of to the right, and pointing to the right for 0 feels unnatural to me.
I know (cos, sin) is "normal," and that it results in 0 pointing to the right. I don't like that, and I'm guessing neither did Alex, and there's no real reason you can't use (sin, cos) instead. (cos, sin) for (x,y) is mostly arbitrary anyways. You really only have to change how you use a few other functions if you're doing that. For instance, with atan2 you have to reverse the order of the arguments.
absent more than alex
written by Shadowlord on Feb 11, 2005 20:35
Changes in R8:
(See the 'Fixactis Releases' thread for the download links)

  • I corrected several places where I had 112 instead of 119 (119 is the maximum latitude, 1 is the minimum).
  • Made animals less common in temperatures below 0 degrees celsius. However, that's based on the base temperature (since animals are created when the sector is, and only the base temperature is known then), and it can get somewhat colder than that if you are high above sea-level (e.g. if you've found a deep chasm, and are on top of it, or a tall mountain or spike).
  • Now when you're in the lander, pressure won't decrease as you move higher (or increase as you move lower). That just means that pressure should remain standard while you're in the lander.
  • Successfully made target-by-parsis target the star at the given coordinates, even when the coordinates are really for a planet that's a distance from the star. If there's no star there it just acts like normal parsis coordinates. BTW, if this doesn't work for you, let me know, and include parsis coordinates so I can check it out.
  • If the F1 page was open when you started the game, it won't be anymore (it might or might not have been before - the 'open' variable wasn't being initialized, so it had a not-quite-random value).
  • Fixed Stella's crash bug (running out of power on the surface and then crashing when returning to the drifter) - though I don't know why it started happening, since even in the original source, the drifter was still using power while you were on the surface. Of course, you couldn't see status messages on the surface before... Anyways, simply not calling additional_consumes while on the surface works, and the power should still be drained after you return to the drifter, but you won't crash (At least, I don't seem to anymore with stella's current and surface.bin).
  • The terrain distribution was decidedly non-random, due to the random algorithm (which is in BC.31's libraries and can't be changed), because local seeds for adjacent sectors were only slightly different, etc. The local seeds are more different now, still without any duplication (In addition to what it was before, it's gained 6 more bits (which are the least significant bits; the others have been shifted left by 6), which are mixed from the lowest 6 bits of lon and lat using logical OR, AND, and XOR, 1 per bit). The result is a more random, but not 100% random, distribution. I tested using the mersenne twister instead of the default random number generator, but seeding it was too slow.
  • The terrain is laid out more sensibly now, and there are also two new terrain types, forest, and shrubland. Forests have many many trees (and will probably run slow unless you turn on the debris-filter on older computers - I think this is because trees' fractal structure is generated/calculated every time they're drawn, instead of just once (due to memory limitations, I suppose)). Shrublands are between plains and deserts in terms of rainfall and habitability, and trees tend not to grow there.
  • Deserts now have cacti-like plants, and icy/snowy regions can have trees (depending on how far they are from the poles). The names of icy/snowy areas in the F2 info box are more detailed, too (Taiga, tundra, icy, etc, and taiga can be dense, sparse, very dense, or just you're average normal taiga with no adjectives).
  • On surfaces, when there're large numbers of trees, ones farther away aren't drawn. The more trees there are in the sector, the closer they have to be in order to be drawn. This helps to keep Noctis from slowing way down due to there being too many trees visible. However, it may not be enough to keep it from slowing down on really old computers. (BTW, the distance check is fast, it's done once for each 'sector' being drawn)
  • If you attempt to set remote target, start vimana, or set local target while you're preparing to land and the cupola is closing, your landing will be cancelled.
  • Calling for help works again.
  • In orbit, there's now a terrain-filter which shows you what type different sectors are from space. Press the 'x' key to toggle it on and off (it'll turn off after you land too). It tends to look better with the antialiasing off. Here's why: The colors aren't all one consecutive spectrum (in terms of their color values, not their palette indices), so the antialiasing can result in showing the color of a totally different terrain type between two terrain types, but I've arranged them in the palette so that they (usually) should look sensible: For instance, if a desert is next to a plains, the antialiasing will the shrubland color between them. The most noticable one, though, would be icy terrain next to pretty much anything else.


Did I mention you can press 'z' to see your speed in various units? Including, if you're curious, TOS star trek warp speeds (since the TNG/DS9/VOY ones are bizarre). Those might communicate more to your mind than just seeing that you're travelling at 300 thousand times the speed of light.
written by Mcwgogs on Feb 11, 2005 23:44
It's very very nice!
the trees could be largerr in forests...
hello! :) felysian
written by Hello! :) on Feb 11, 2005 23:56
Thinking about it... how desperate willing are you to learn the PNG format...
absent more than alex
written by Shadowlord on Feb 11, 2005 23:57
I'm quite content to let it continue writing screenshots in BMP format. If you have the folder set to be compressed, anyways, then voila, windows transparently keeps them compressed for you.
hello! :) felysian
written by Hello! :) on Feb 12, 2005 00:11
I said nothing about compressing them. I just asked (implied) if you would be willing to learn to write the screenshots in the PNG format.
written by Mr.fop on Feb 12, 2005 00:57
Shadowlord said:
.
But wouldn't using (sin, cos) make degree 0 point upwards to (0,1)?
reading this thread
no members are reading this thread
. /../Known Noctis IV Bugs, and.../ 1..1819202122..71
42624, 13 queries, 0.113 s.this frame is part of the AnyNowhere network