Getting the Most Out of Modded BC

[Morgan]

Added to the list along with speculars and enhanced glows. Thanks Aces.

Also added suggestions for performance improvement through Bridge Plugin, mostly based on my own experience and just things that seem to make sense to me. Any further ideas (or corrections) are appreciated.

[King Class Scout]

please tell the non computer savvy (and those of us who's tech savvy is a couple generations behind) HOW to shut off this T&L.

[Dalek]

Go to Configure/Graphics. Look at the first few sets of buttons. If you see some text which has the name of your graphics card, click on it. A little menu should pop open and then select the option with your graphics card but doesnt have T&L written next to it.

[JimmyB76]

ok im confused (as always lol)

so if i understand correctly, only people with decent computers should use the T&L option for better quality? 

[Kirk]

One with a decent graphics card should. If your graphics card doesn't have enough memory, turning on T&L will not help performance.

[Morgan]

One with a decent graphics card should. If your graphics card doesn't have enough memory, turning on T&L will not help performance.

Can anyone give us a range of "decent?" Say 256mb and below should leave them off, etc.

[moed]

Actually in my experience, 256mb and above are ok to leave T&L on.

I have found that the biggest hit on performance hands down is the LOD of the models in use, the sizes of their textures, and the amount of ships/objects on screen.

If any of you are like me and have a ton of different ship mods installed from the different what I call "eras" of BC modding, then depending on the ships/objects you play with will be a huge determining factor on your overall quality and playability.

For instance, whenever I play with a bunch of older p81 ships (I keep then around for nostalgia... and also for the fact that they're a great improvement over the original BC ships) I can play everything at maxed out settings and even those old ships from a distance don't look too bad.

On the other hand, when I play with a bunch of DJ, LC, CG, or any other high res models on screen at higher resolutions then I definitely start to take a hit on quality/performance... and I have a pretty powerful system.

It's all about balance.

[King Class Scout]

also include a refrence for those people that still run XP windows.  not everyone has upgraded to Vista or Seven.

I find that, if an individal ship download totals more than 10 meg, it gives my 128 meg card conniption fits.  so basically, pare down ships to match about 10 percent of your graphics memory size for good performance.

[Nebula]

using the DL size of a mod is a baaaad idea... to fix your issues just reduce the size of the textures... (that makes up the bulk of the file size anyway) so if it is a 1024x1024 texture reduce its size to 512x512...

[King Class Scout]

i did...with Baz's JJ's (which were the ones that gave me the most trouble).  it didn't help.

another suggestion.  make sure the graphics card is up to date.  the chipset in my comp needed 17 meg worth of driver updates!

[Morgan]

Okay I added information provided by Moed about 256mb and up being ok for T&L. I also added a section on high-poly, massive texture-sized ships. I originally thought a section like that wouldn't be necessary (I considered it obvious to be honest) but there's a few posts here going into detail about it so I won't argue with the consensus.

[ACES_HIGH]

Added to the list along with speculars and enhanced glows. Thanks Aces.

Also added suggestions for performance improvement through Bridge Plugin, mostly based on my own experience and just things that seem to make sense to me. Any further ideas (or corrections) are appreciated.

actually you have mipmapping backwards, as I understand it, Mipmapping simplifies the model, so it will use less memory, I could be wrong about that part, but I'm pretty sure you want it on if you want to improve performance.

[Morgan]

actually you have mipmapping backwards, as I understand it, Mipmapping simplifies the model, so it will use less memory, I could be wrong about that part, but I'm pretty sure you want it on if you want to improve performance.

From the BC Instruction manual:

Click this button to turn mipmaps on or off. Turning mipmaps off will save some texture memory at the expense of rendering quality.

Of course this gives no mention about the model as you suggested (about simplifying it), so for all we know this feature may go either way performance wise...can anyone help clarify this one?

[Kirk]

Mipmapping does nothing to the model, only the textures.

[ACES_HIGH]

whoops,  I guess I had that backwards all these years.  I could have sworn it saved memory.

[King Class Scout]

a little experimenting on my part yeilded some interesting results.  high-poly ships reduced to the original BC's maximum sized 512 textures saved me a lot of greif for 30k poly ships.

[Morgan]

Yeah textures are probably one of the most common slowdowns in BC. I personally won't go past 1024 (unless it's an install without NFX).

[Kirk]

I believe Mark, while creating the most recent version of the Intrepid bridge, did do a BC engine stress test with many of the people here to determine what was an acceptable poly count and whether polys or textures put more strain on the engine. In the end I believe the verdict was textures put far more strain on the engine.

[JimmyB76]

actually, it was the Ent-A bridge with several dozen high-poly chairs put all over the place lol
i was there at the time (with others) providing feedback on a mid-range system...   it was funny; chairs sticking out of the walls, ceiling, everywhere lol

[R-TEAM]

Hi,

it is all time the whole system (CPU+GPU+GPU-RAM) and the settings that give the
performance.

If you have an old CPU that dont get the data for the graka fast enough out, then the
grake sitting in wait and doing nothing.
T&L is a good performance boost alltime ! (but remember the point with GPU-MEM ! )
If you have an cheap MoBo with an non-optimal design that basicaly runs but not all
good speed then the data com to slow to GraKa or the CPU<>RAM is to slow.
Now to the GPU-RAM:
It is simple->
To operate at full speed - ALL! data that the GPU process MUSST fit in GraKa RAM.
If it musst swap RAM with MAIN RAM (even with PCI-16) its SLOW down like hell!
(if you have a good Graka - if you have an old who the onboard ram isnt faster as the
AGP8 transfer rate then you will notice only little difference as example ...)
So the hint with T&L is normaly a bad set, this move the expansive T&L calculation
to the CPU and initiate an extra data overhead on the graka<>CPU bus.
Only if you use an old driver who is bad with T&L THEN it isa good solution,but then it is better to update the driver ...

Now to the REAL GU-MEM Eater ->
AA and Mipmap
If you have an Texture that is RAW 20MB.
With Mipmap it musst hold in different detail size in RAM (at now RAM stand for GPU-RAM) so you get with 4 level at maybe 30MB RAW - with compression
you get maybe final 3-7MB on the bigest.(depends on the texture - if he can good LOSLESS compressed)
Then you use AA - EVERY AA level musst hold the actual mipmap Texture in RAM.
So at 16x AA you get at close range with the 7MB texture an memory amount
from around 110MB.So if you have 2 ships realy close with both high-res textures
then you get with 2x 110MB + 30MB basic textures + 3D Data + shader Data very
close to 256MB ... this is optimized in different ways ... but the AA give the biggest
Memory impact.
If the RAM can hold all data T&L on is always the best (with driver that operate flawles).

Regards
R-TEAM