posted 03-02- 07:21 AM               

K guys this started in the post about the F4F

"Would you like to see this in the game by Saturday??"

If you need to get up to speed:
http://www.fightersquadron.com/ubb/Forum7/HTML/000424.html

K I did some quick research and this kicks MOTHER ass. Really.

Well maybe i'm just excited.

We said we could have low poly Bombers for the AI to fly right.

Well i tried my idea and its not even the half of what we can do.

I took almost all the LODS and put the highest LOD switchin at 200 so they never switch in I don;t know what the units are I wanted to have them still switch in but ya get whats happening anyway.

check this difference.
1st pic the lanc with the 2nd level LODS showing all the time. (with the first at 200)

Check that FPS. Thats my p2 266, banshee with fog at 10 miles, 800x600,texture and object slider full. and clouds on. lets just say thats a good FPS for me.

check these 3 out with the stock Lanc. at different distances.
everything is the same same graphic options.

K I personally find this really interesting.

We can make all the bombers have a really close (pretty much never) switch in distance for the highest detail LODS. I say we keep them cause they are used for collision detetion remember.

To do this would be a bit of a hassle cause we would need 2 versions of every bomber we have if we just wanted to do it to the bombers but i think its worth it if we can get more bombers in the air.

This also brings up another really neat thought that i think should be applied to the next verion of SDOE or maybe even SDOE when we see the open source thing come along.

2nd level LOD bombers and actually all aircraft for the AI.

How hard would it be to have a switch in the game code that says (k this ain't programing by no strech)

;start
ask human/ai
if human use hi level lanc
if ai use lo lever lanc
;end

HAHAH cracked myself but you get the picture.

AND i had another thought. Why do AI pilots need dashes? or refective canopys or moving control surfaces? we still want them but they don't need to switch in till a human player gets with in say a couple 1000 feet.

If it was coded right into the game we wouldn't need 2 versions of the lanc to pull this off. the game would do it automaticly on the fly.

So what do you guys think? I think this could blow this game wide open.

posted 03-02- 07:27 AM               

K i was looking at it again and boy is this cool.

Look at the first pic and look at the FPS.

Notice that i am closer in this pic then any other.

Now notice how much lower my fps is in the closest look at the hi level LODS.

Its getting pretty close to half well a strech but you get what i mean.

I relize there is a law of deminishing returns but if we get it down to half (i really only did about 1/2 of what we COULD do) then we can have almost 2 times as many bombers in the air.

at least 1/3 more. HOW cool is that?

posted 03-02- 07:34 AM               

Ok so how does that work? Whats the formula?

posted 03-02- 08:19 AM               

Yeah definatly right with the detail slider but i don;t think we will be able to convince everyone to slide it down.

This could allow more planes with still hi detail for the plane your in. the other method won't

posted 03-02- 09:22 AM               

I bet you will find that in big missions you wil still get like 5 to 10 FPS. All the bomber AI work and physics calculations add up fast. EAW can do more planes because it is not stuck doing all the physics work.

I think the more planes you have, the more physics is the killer over polys. Statisticaly, the more planes you have, the more lower LODS each person will see anyway. Remeber, only what you see can hurt you

So as the amount of planes increase, the total polys rendered curve levels out... more and more planes don't change the poly count at all because more of those planes won't be seen anyway - or at least they will be further away and switched to low detail.

Now there is an exact 1 to 1 ratio of planes added and physics calcs. So each new plane takes a bite out of FPS here. This is the weak area...

This is why SDOE will kick ass more and more as time goes on. People get better systems and we don't need SDOEII to kick the competitions ass! We just add polys and break the planes up into smaller pieces! Now we have better graphics, better DM, and better FM!!!

Imagine SDOE on a 5000 procesor! We could have very detailed planes - breathtaking! We're talking about modeling each slat of a radiator! Moving gas-filler caps LOL!

Also imagine the DM, shoot of small pieces of wing skin! Take out main spars! What about the FM? More FPS = better physic calcs. More detailed inertia boxes. More detailed FM changes with damage.

If OpenPlane keeps up the way it is, we may see OPenPlane enhancements to give us detailed engine modeling, better control surface manipulation, moving ground objects... wow!

Just dreamin' here

------------------
-Sv =FC=

WWI in SDOE!

posted 03-02- 09:40 AM               

YEp your right SV i tried a mission with 9 lancs

and i went from 7 FPS to 8 heheh

All i'm saying is everylittle bit counts.

Look at the difference that made with one plane. Thats pretty cool.

and if you notice the plane does't look that bad.

Now do you really think we need the highest level lod to switch in very far out? i don't i think it shouldn't switch in until that lanc is almost filling the windscreen of your FW. But i bet it switches in WAY sooner then that.

I personally don't think we should have to upgrade our computers every year. What we need is PC games that are more optomized for the pcs people actually use. instead of the pcs they will be using when the game is 4 months old.

Look at consoles. the games get better every year by leaps and bounds and the hardware is the same.

the good thing is MH is doing exactly that. he is putting some the code optomization he did for the rc sim into SDOE.

posted 03-02- 10:06 AM               

This brings up an interesting question. Couldn't the physics, motion, and other processor related issues of the AI planes get simplified when they are out of visual range. Couldn't the movements of say 4 planes get grouped into the motion of 1? ... kind of like the "player bubble" in Falcon 4. That way, we can be sure that the framerate isn't getting massacred by stuff that is going on off screen.

posted 03-02- 10:17 AM               

As far as the AI idea JT, don't forget that this could be a group of planes dogfighting out there - they need to do their stuff in order to make missions fun.

We have 3D cards to off-load the 3D processor work, right? What we need is the first PC based FM card! Think of it! It could be optimized to do physics the way a 3D card does hardware based 3D math. This would probably make SDOE 100 times faster. Imagine if your systen had to do all of the 3D work? Wouldn't we get like 40 seconds per frame instead of 40 frames per second?

So a game just calls APIs to create airfoil objects, intertia objects, etc. Then the card does the work. That would be cool...

posted 03-03- 12:22 AM               

SV. We used to have those back in the day remember. Math co-processor.

posted 03-03- 06:22 PM               

Reaper. Yep exactly. But I think that since physics as far as I know is just a bunch of # crunching that we just need better FPUs in the CPUs