Spring RTS Engine

I don't mind the polygon numbers themselves but I do mind unnecessary waste of polygons.

Caydr, could you comment on the Part number shown in your tests? Is it the number of parts included in the 500 visible models? Because it looks like the GEM fighters use less parts, which might be counteracting the loss of performance from more polygons. Not trying to disagree with you- I'm hoping you are right, after all! Just curious to cover all the factors.

EDIT: Of course, a part disparity of 2x probably isn't covering up a poly disparity of 20x, but it is worth mentioning, I guess.

Um... having seen a GEM fighter... yeah, they don't have many parts, compared to a Kbot.

More importantly, what I've observed in my tests is that polycount starts mattering when you're doing transforms (i.e., COB animations) because of the data-overhead involved with each part. IOW, it costs a lot less to have a single object that is 50K polys than to have 50 1K objects, all undergoing transforms at once... and even a 50K poly object that is undergoing transforms due to the main code (y'know, going over terrain, etc.) and then being transformed by a script gets kind've twitchy.

All of that said... for AA, these polycounts will be fine, I think, so long as Caydr's very, very careful to optimize his animations and to not use the FALL explosion event unless he has to (this is quite expensive).

Hadn't considered that... Well... the fighter used in the test has several pieces:

the base model
fire point 1
fire point 2
engine point 1
engine point 2

An average kbot has more than this, an average vehicle or structure has fewer.

Does the (when specified only) caching of non-moving pieces that used to work in TA still work in Spring?

Caydr, u wouldnt mind if Xta borrowed the models if they were really good, would you?, id also like to see a few models myself so i can play with them and try to texture them myself
id like to see if i could make those models high poly 3dos like these:

I haven't decided on how much or how little I'd like to see other mods using them. Inevitably, even if I say "NOBODY USES THESE THEYRE MINE HAWHAWHAWHAW!!!!!!!", somebody's going to use them and nobody's going to stop them.... I think I said earlier that I'd just want credit to be given.

If polys REALLY dont make much of a diffrence like you say, then knock yourself out with the poly count, I don't know much on the subject myself. I do know that high poly models look sweet though.

Noruas, I think Caydr is doing s30.. God forbid if he isn't...

sigh. personally i think caydr's model, though inefficient polygon wise, are good (better than his GEM ones), but I guess, just because it is Caydr, everybody hates it.

TheRegisteredOne wrote:... but I guess, just because it is Caydr, everybody hates it.

Sorry?
Nobody hates them, at best there are people giving feedback on polycount.
Why so negative?

Geometrically, the models are fine but I agree that there's a boatload of unnecessary polies lying around, and the models really need to be optimised.

Tim Blokdijk wrote:

TheRegisteredOne wrote:... but I guess, just because it is Caydr, everybody hates it.

Sorry?
Nobody hates them, at best there are people giving feedback on polycount.
Why so negative?

i wasn't being negative. I stated I liked the designs. and did it occur to you those who dislike the models did not post their opinions on these boards? the spring boards is not the only place people talk about spring.

I know what TRO is talking about. Don't fret, keep going with your work Caydr. I'd offer help to make the load lighter, but I can't model... yet.

I would help, but I can't..

I refuse everyone cuz I am working on my mod XD

i havent read all posts but i think ur doing great cydr keep up the good work..
i know ota is a holy cow but i think aa does need a bit of change..
i mean arm and core should be more different stylewise..
each side needs to be in a certain style ill hope ull achieve better results
in that part of the design proccess.

He's remodelling not redesigning Yan.

So now my dual nVidia 7900 GT's will be taxed more then 5% right?

Mainly it's the unmerged faces, like on the unit's feet that bug me.

They overall look quite good, especialy the storm, but merge the polies that can be merged (look at the front of the toes) and i'd recomend removing some complexity from the eges of the "window" not necesairly because it adds too many polies, but rather because it looks cluttered (unless it's just the wireframes making things look odd) The "polygon density" on the storms launchers and main body is preety optimal for higher poly models.
After that just make sure they have good lods and they should owrk fine...

(By the way, amke bumpmaps for whenever spring gets to use those )

Caydr, here's a point: The game industry uses pixel shaders. Spring will probably be using modern pixel shaders soon. For this kind of sheer quantity of work, doesn't it make sense to plan for the future?

Take these models, do low-rez versions of them at around 300 polys, 600ish tris. Nvidia has a tool for creating normal maps for low-rez models from high-rez geometry. Use it. Then you get most of the detail from your high res models, with less polygons and a much better final appearance. Right now you still have a thing or two to learn about optimization from what I can see - this will allow you to cheat that and get the best results possible.