[Lightspeed]

Originally posted by Sticks
What I would be willing to do, if it would be useful, would be to add a command line that would divide the displayed speed by 20 so that modelers could scale everything appropriately. Though, I'm not going to muck about with this if no one's going to take advantage of it.

Dogfighting at snail-speed™. Whee!

[Fineus]

Oh god no....

As for FOV - it's ok, but rather hard to get "right" at the moment.

[Flipside]

I think it's more a question of relative angles, if you take two shots at the same FOV for a 1m shape then the same shape with the camera positioned at exactly the same relative point on a model 20x bigger, you will be able to see surfaces on the 1m model that are occluded in the 20m one,

This is an enhanced still from an old version of my Atlantis video on Sectorgame...



You see how the front jet section is side on to the camera, whereas the section that the pipes feed into is facing slightly towards the camera? This is because in this version, the model was tiny, actually about 30x smaller than it should be, which exaggerates the effect a little, but you tell the model is small because of these angles, surely if the pipes are heading in my general direction, that front jet mount should be at the same angle?

I got as close as I could to the same image on a larger model, I use a landscape rendering program for this, so it's not an exact art, but this was the outcome....



As you can see, the ship looks a lot 'straighter' here

I'm not a 3D Maths guru, and I'm not saying this is the be all and end all of the problem, but I think it is part of it

[StratComm]

That's just a FOV problem flipside.  In a video game, you generally assume that the point you are viewing from always has one and exactly one possible viewing position.  However, in a render engine, you can have approximately the same image displayed with a variety of pan, zoom, FOV and camera position settings.  That's why you have a more or less "straight" render, but you could produce the exact same image if everything was scaled up 10, 50, even 1000 times.  By the same token, you can also almost always achieve the same exact image with a much smaller model, although limits on the precision of units can screw around with this in a lot of modeling programs.  This really is, pure and simply, an issue of FOV that should not be under discussion again.

[IceFire]

Got to be very careful with sizing things in FS2.  Venom did that for OTT ships and the whole thing went to hell in my opinion.  Fighters were falling like flies, the AI didn't know where it was, and there was of collisions and that sort of problem.  Venom didn't regard it as much of a problem but when I went to balancing it was definately VERY tricky.

[Flipside]

Maybe so Stratcomm, but I hadn't touched the camera at all, merely resized the model and moved the camera. If it shouldn't be under discussion again, why is there a thread about it?

Maybe someone can do something similar with some higher level software than mine, I'm willing to accept it as camera curve or the like, but why, with precisely the same camera, can I see more of the front of the main jet on the lower image, when the camera is actually slightly closer and further towards the back than the original? The only difference between the settings in the two images is the size of the model.

[Flaser]

AHHHH!!!!!!

NO! NO! NO!

For heavens sake forget FOV it has nothing to do with the actual problem, it's like putting on a scope and pretending that everything is bigger.

Forget the VELOCITY / SIZE issue too.



Sandwich has already solved the quetion of as of what's the problem with FS ships:

The FS meter is 1/20 of what it's supposed to be.[/b]

One suggestion I heard earlier - and I immediately wrote off as STUPID was to increase all the speeds by 20 as well as the scales.

You said that wouldn't change anything?
WRONG!!!!

What everyone failed to realise so far - including me - is the fact that we're speaking about LINEAR SCALING

That's why the actual speeds and DISTANCES match.

However if I set a linear scale to 1/20 of the original all the ships will be (1/20)^3 = 1/8000 of their original size[/size].

This means the Collosus will look like a 10 meter big piece of brick!

[Sticks]

It's not a stupid idea, just completely and utterly impractical, from a coding standpoint. Feel free to scale the models by 20 and adjust all the table velocities, though, and see what happens.

Actually, to be fair, adjusting just the physical models themselves is about 2 lines of code. However, then the problem lies in the AI, ship speeds, weapon speeds, particle sizes, collision detection, z-buffering, physics, beams...

I could go on, but won't, as I think everyone is getting the idea.

[Flaser]

Sandwich always believed it be a coding issue - we just didn't know the extent of it.

Thanks for giving us the details Sticks.

BTW where could the problem actually come from?
Are the "models too small"  or the way FS scales everything is to blame for the fault?

[StratComm]

Flaser, if everything is linearly scaled up, then what's the point?  You still get the exact same image if the camera is treated as part of the scaling.  Don't believe me?  Look at these two shots.




Ignore the ship, it was a failed experiment.  Anyway, if you look closely at those two images, you will see that they are identical except for some mapping of the vasudan bits.  Now, in the first picture each dimension is normal, in the second everything has been scaled up by 10 times.  So, you are looking now at something 10 times as long, 10 times as tall, etc, from a distance precisely 10 times as far.  That is your linear scaling.  Now, lets look at a third picture.  



This picture is notacibly different, but viewed from the exact same position as the first.  In this shot, the FOV has been reduced from 45 degrees to 22.5 degrees.  Obviously, if the range to target is the same, and the ship looks so much bigger, then your problem is solved.  So, to make ships look more massive you are looking at changing the FOV, and that's why all the coders have been saying that it's already done.

[mikhael]

I dunno. For a failed experiment, it looks damned snazzy.

[Sticks]

My pleasure.

It's actually the way the file convertors scale things. So yeah, I guess you could say the models are too small, although everything else in the engine has been built on that principle.

That's why I suggested that I could add something that would change the way velocities are displayed. That way, modelers could scale up the size, and then edit the tables to make the ships twenty times as fast. In the code, the displayed velocity would be divided by twenty, thus making it appear that the velocities of things never changed.

That's really the only feasable solution, in my opinion.

[StratComm]

Sticks, I really don't think it would change anything, other than introducing more reason to get close to existing limits again.  If we call the units 1/20 meter and scale everything up to be "proper" size again, then it would still look the same from, lets say, 200 "meters" (displayed units) from the viewed target as it would have looked without scaling at 200 game units from the target.  People forget when talking about scaling stuff up that if they will also be increasing the distance from which the objects are viewed.  In truth, the problem lies in that Volition created the viewpoints with the intention that the player could see "around him" through his/her computer screen, and so a much larger visual field is presented in a small space than would exist if the computer screen was simply a "window" into the game world.  Playing with FOV alone may not cut it for some people, but in truth that's how it works.  A meter, viewed from one meter away, looks just as long (to a monoscopic eye) as a mile viewed from one mile away.  It's all just geometry, and it works out the same no matter what units you use.

[Sticks]

Yes, well, this is what I explained in the first thread, but no one believed me. So I figured that they could have their twenty times multiplier, and everyone could be happy.

But then you had to go and bust my master plan...

[StratComm]

What can I say, I try

[CP5670]

It's good someone sees what is going on with that unit changing thing; I was about to go on a rant about that myself.

Anyway, the FOV setting actually does solve the whole scale problem, but it introduces the background problem as a side effect. I also tried it with a nebula mission a few hours ago and the poof bitmaps were also appearing too close to the player. As I said before, if there was some way to make the display distances of the background sphere and the nebula poofs dependent on the FOV setting (or let the user specify them individually), the problem would pretty much be solved.

Being able to specify some kind of multiplier on the speeds of the debris animations would also be a good idea. I think you all would agree that 100m/s looks and feels more like 5m/s or so; this could possibly fix that, which would indirectly lead to the distances seeming more believable.

[Bobboau]

I havn't read much of this, but I posted this elsware, and figured it would be relevent







that herc is 13 meters wide 11 tall and 23 long, prety much the exact same as it is listed in model view

[KARMA]

btw what are exactly the problems with fov?
just make a list of all the problems that happen when you reduce the FOV, from background to nebulaes, make sure the coders know about that and give em feedback.
FS2 has been built spcifically around a defined FOV, just let the coders work in adapting it the best possible to a FOV that make things look more realistic.

[Flipside]

Ok, thanks for explaining that for me people, at least the bits I understood

As I said earlier, I'm not an expert at this side of things, but then you don't get to be an expert without asking questions

Flipside

Edit : And fair enough Sticks, I know you guys are dedicated, but there are limits

[Lightspeed]

*applauds to StratComm*

thanks for demonstrating what I was ranting about all the time