[J3Vr6]

Hi Bobboau!!!



(I didn't want you to feel lonely when you woke up and saw you were all by your lonesome).

[Bobboau]

happy birth day too me,
happy... birth...   day?

[kasperl]

Happy birthday, again, as in the thread in the main forum

I'm waay to tired to do anything but browse forums and read a book. I'll try this out later.

[Bobboau]

I was just kidding, but it is demoraliseing to be up to 3am working on something post it and 6 hours later nobody responds.

[J3Vr6]

Hey, I responded!!! What am I, chopped liver?

[Taristin]

I haven't been online for over 12 hours, if it's any consolation.. Otherwise I'd have DLed it and tried it, I'm sure.

Plus, you did post it at 3 in the AM... I sleep at around 12...

[Ace]

Originally posted by Bobboau
the screens shots are from BWO, mission called blindsided in wich the Golgotha blows some **** up.

I thought it was Blindsided Make sure you have the latest version I made. The events play out better, meaning that the Golgotha firing shots should be even more impressive.

[Lightspeed]

just got the build. Will be testing it later on

[Deepblue]

Gonna nab that build as soon as I get home.

[Lightspeed]

Deepblue: Using the quote tags in signatures tends to make them incredibly huge.

Bob: Okay, I've done some excessive testing with this build, and as always I've found out quite a few things (bad news first):

1) After playing a couple of missions, suddenly there seemed to be some missing textures. For example, the dark bits of the Demon destroyer weren't there, and the Lucifer was completely transparent, the Lilith cruisers also had no textures anymore. I quit the game, and restarted it - and alas, they were there again. I could NOT replicate this bug till now, but it might be worth being looked into.

2) Someone seems to have been fiddeling with the particles again screwing them up like when RT tried implementing the HTL particle system - they have the same issues as they had back then, they're way too small and seem to be spawned sometimes, and sometimes not at all.

3) The build still causes some odd slow-down (I think its down to env mapping) which it technically shouldnt.

(now for the good news)

4) No crashes whatsoever, it ran pretty stable except for the bugs listed above.

[Bobboau]

I've been screwing with the particles, exactly as you think, the old system will be preserved for people who don't have suport for point sprites.

I think bug #1 is the one I'm most concerned about

I'll make a comand line option (or two) for environment mapping, alowing you to disable it or turn off it's dynamic nature (switching render targets can lead to major slowdowns and environment mapping does it 7 times)

[Lightspeed]

Well, the particles *do* work, but not as they should

As for the env thingy... "turn off it's dynamic nature (switching render targets can lead to major slowdowns and environment mapping does it 7 times)" - what exactly does that mean, and how would it affect the in-game looks?

[Lightspeed]

I'm talking about this when I say they're screwy: Click

This is an image from an old bug report of mine on Mantis showing exactly the same situation (when RT was working on them).

[Bobboau]

do the particles seem to be diferent sizes (from one another) or are they all the same (wrong) size?
some cards may not allow point sprites to be diferent sizes, also check them in a default fov.

the turn off dynamic cubemaps will have virtualy no visable effect generaly, the only time you'd see a diference is if the mission incorporated animated background effects or if it was in subspace (in wich case you would see a very noteable diference). switching render targets is what you do when you render to a texture or something, so we switch the render target once for each of the faces of the cube (6) + 1 for returning it back to the back buffer.

well right now I think I've got the decal generation code to never generate illegitimate (containing fewer than 1 poly) decals again. there are some places on some ships that would never get a decal applyed, like the entirety of the Seraphis, nothing on that model will take to the decal code, and I'm not completly sure why, but I now have code wich I am confedent will never fail to generate decals. (about to test in 'surender belasarius', bet it doesn't work )

[Bobboau]

meh... it's even worse than before

[Lightspeed]

I have an idea

Wouldn't it be perfect then, to check if the mission is or isnt in subspace and THEN deciding which method to use for the env mapping?

That would speed up things and look exactly the same (as the subspace env is still animated). I think this would be the ideal setting for most people.

As for the particles: TBH no idea if they're all the same size; does a Radeon 9800 support differently sized point sprites?

[Lightspeed]

Okay, more info:

I did some more testing, and managed to replicate the missing textures bug. It occurs almost every time I play several campaign missions in a row. It seems to appear after/while a mission has been loaded as in the first mission that is played they're still there, but in the briefing of the second mission they're already missing.

About 50% of the game's textures seem to vanish (went through the techroom after it happened) - there seems to be no order behind which vanish and which not (it's all mixed up).

Erm to replicate load up your favourite campaign, play a few missions in a row and check if the textures are still there.

[Bobboau]

hell I didn't even need to play two missions, all of the sudden Nephelems are gone, wonder what the hell is going on there, I havn't touched the texture system with a ten foot poll

[ShadowDrakken]

you may be capping out the video card's texture memory with the decals eh? I had a similar problem back with Diablo 2 when I was using an 8mb card, went to 128mb card and the problem fixed itself... so somewhere, memory is not being released like it should...

just an idea here, not certain since I don't do any serious coding and don't have a clue with FSO's code

[Bobboau]

I've got some interesting numbers in the final mission, a total of only 522 decals ever get used, if everything is working right, this is the number of the most decals that are in mission at a time, only 3487 polys ever get used at a time also. (an average from this will show 6.68 poly per decal)
in the test I ran (final mission) 1234 decals are made (total more than half of these are recycled though) each one has an average of 9.357 polys, the average polys per decal has been an extreemly eratic number, it was 4.727 the first time I tested, and 6.412 the second time.

it seems that I over estimated how many polys and decals would be needed by nearly an order of magnatude, if these numbers hold up for the other mission I could probly get away with 3500 polys and 1000 decals (current prealocated block is 11000 polys and 5000 decals)