New Bugs, Rev 5879+

[Argh]

I don't really know. After reading the source, it didn't look like mip levels were being applied to AirLos anyhow. But yeah, the previous code capped the AirMipLevel at 30- in the interest of efficiency, I raised it to 30 (why pay for more detailed searches, when it's going to show you stuff that's not in LOS anyhow). Switching it to 1 cured the problem.

Lemme try 29... maybe it's being effected by the problems with LODBias or something equally unlikely.

I should note, again, that it all "works", but that doesn't mean it necessarily works well. I have yet to do any real stress-testing, to find out whether it's broken by other events.

<tests>

Yup, it's being effected by LODBias

[Argh]

Yay, new problem. Setting LosMipLevel > 1 seems to multiply the distance at which Units get LOS on CEG events. IOW, I can't raise it to 6, the previous maximum, to gain back all of the efficiency benefits of using a cruder representation of the heightmap, without giving players a virtual cheat-code, showing where their enemies are, and what they're doing- because in P.U.R.E., practically everything is emitting CEGs constantly

I've disabled that stuff, for now, but meh, that's aggravating.

[Noruas]

It seems that mines cannot be ran over anymore.

[KDR_11k]

Does the yardmap on them still say "y"? That's implemented now so I might have removed the kamikaze=1 check.

[Noruas]

they all said yardmap=o; and did not work like before, however
yardmap=y; does work, but i dont know if it works in the current version.

[KDR_11k]

76b1 just ignores the yardmap and checks the kamikaze tag, setting it to y won't break anything (I think most mines are still set to y).

[Argh]

As of rev. 5960, Spring runs well, and P.U.R.E. runs correctly, on the ATi Radeon 9600. Well done, whoever fixed it.

[hoijui]

i can confirm that too.
the SNV log sais trepan committed 5960, and only change i see that could have anything to do with that, is:

Code: Select all

- changed to NEAREST filtering for gui_game_setup.lua's font rendering

makes sence, as it was always slow then the build menu showed up. slower when it contained lots of items with text.

thanks trepan!

[Guessmyname]

I'm getting some weird texture-breaking here:



Also:

[FLOZi]

I get this, too. Have done for a while.

[Argh]

I get the black-bar thing, too, on both the Radeon and the GeForce 7800. I hadn't mentioned it because it was non-fatal. Maybe an issue with GL_FOG, and the really huge water planes?

[Argh]

Um, I'm having a very weird issue with the World Builder demo map, and the Radeon 9600- it won't run at an acceptable framerate. And yes, it ran Ok, with minimal settings, before.

My guess- and this is just a guess, mind ye- is that it has something to do with the different DDS compression used. I used 8888, instead of DXT1, which is what the old, inferior MapConv used (along with other problems).

People with newer video cards aren't reporting problems, and on my 7800GT, the map runs just fine. I will try recompiling the map using DXT1 and see what happens during the next beta of P.U.R.E., but I wanted to put this out there, as it may have some relationship with the weird issues other people have been reporting with user's tool. The old map code, which I read when user designed the new MapConv, is format-neutral, so I'm wondering what has been changed...