It's very cheap on any shader-capable card. The per-pixel sin is slightly expensive (although the shader compiler will typically optimize it to a single sincos instead of two sins) but that balances against a lower vertex submission cost due to no subdivision (and not to mention the ability to keep everything in a static VBO if you want to go that far).

Doing shader-based versions of both water and sky means that every single surf can go into a single static VBO, in fact, as no per-vertex data needs to be updated on the CPU at all (you'd use a cubemap for skyboxes in this setup - Doom 3 has some code for converting Q1/Q2 skybox layouts to alignments suitable for cubemaps and I've independently written the same conversion myself a few years ago so I'm reasonably certain that it can't be rocket science). Get some draw call batching in and you're suddenly running up to 5 times or more as fast in big, heavy scenes. That does require a higher GL_VERSION than Fitz aims for, as well as some massive surgery to the current codebase. Probably not reasonable to expect in any kind of short/medium timeframe, if ever.

The fullscreen warp has some complexities in how you handle the edges. You can use GL_CLAMP_TO_EDGE but it won't handle the bottom edge which juts against the sbar. There are solutions (I used a "control texture" that fades off the warp effect at the edges, another option is to just destroy and recreate the texture at the appropriate size if it needs to change, which generally doesn't happen at any performance-critical time).

Generally glCopyTexSubImage is going to be slower than using an FBO for this because it needs to shift more data around. Overall you can expect to lose maybe one third of your performance owing to increased fillrate even in the best case.

Another way to do it that doesn't need shaders (and will even work on GL1.1 hardware) is to use a grid of quads (much the same as what Fitz does for r_oldwater 0). That can give acceptable enough quality.

The third way is just to rotate the projection matrix a little based on time, which gives a reasonable enough effect at no performance cost. It's like Fitz's current stretch'n'squeeze effect but with some rotation added in too. You need to extract your frustum from the combined MVP rather than calculating it separately if you do this though. Easy enough (and a little more robust than calculating it separate).

Despite all this seeming complexity and multiple options it's probably cleaner and easier to integrate with any engine as it doesn't need to touch any code outside of it's own subsystem. So - detect if we're underwater (via r_viewleaf->contents) at the start of the scene, and either (a) just set a flag and/or (b) switch the render target. At the end of the scene, either capture the scene to a texture or put the old render target back, then draw using the appropriate method depending on what hardware capabilities are available.

For an engine like Fitz I'd recommend asm shaders rather than full GLSL. The main reason for that is that they're available on a much wider range of hardware, so they seem more in keeping with the Fitz philosophy. 

good news!
thanks, Baker. 

Would it be possible to do alpha animated textures?

Something like this:
{+fire1
{+fire2
{+fire3 

I'll let Baker answer on his own behalf; this opnion is mine and mine alone.

Yes, it should be possible with the necessary code changes. However, by this point in time you're stacking up multiple "first character" hacks (alpha animated water with alternate anims anyone?) and things will begin to look a little delicate.

So some form of alternate system would be needed. Since Q3A shaders are out there, since we have a GPL working reference implementation (the Q3A engine), and since they don't require any format changes (just add a .shader file for anything you want different behaviour for), that would be my own preference. 

It might already be possible to do them:

Set r_texprefix_fence "+alpha" and load your level.

It should process all the textures that begin "+alpha" and likewise set a flag to mark that surface for alpha mask rendering.

I hadn't expected someone to want to use alpha textures like that, but I bet it works.

(Note: fullbright pixels in alpha textures are supported (except color 255 of course). Replacement textures replacing an alpha texture that has an alpha mask are also supported.) 

Baker this makes fitzquake look more like darkplaces, joequake, qrack e.t.c?

record demo is awesome... 

 

If by chance you are looking for something like JoeQuake, you might try Engine X. Engine X supports FitzQuake 666 protocol and looks like JoeQuake with the kind of very recent engine features one would expect (video mode change, etc). 

Obrigado Baker

I will see it ASAP!

thanks 

Thanks for the reply!

Also, regarding the alpha textures, I think there is a bug with the way they are rendered, or it may be intended. The transparency textures bleed through the floor and you can see the void of the map.

Here is a picture showing what I mean:
http://imgur.com/ZFOz1

Even if I paint the bottom of the glass brush with a solid texture, it still shows the void through the floor. 

Turn your alpha brushes into an entity like illusionary, door, func wall etc. Use a really good q3 fence texture, enable texture scroll and be amazed. 

Yup.

Because of the way QBSP seals the world and chops off outside faces, putting this texture type on a brush that's not part of an entity will produce this result. It's expected behaviour and a result of normal QBSP operation, not really an engine bug (there's actually nothing that the engine can do to work around it as the data required isn't even there in the BSP to begin with). 

See above. I almost decided to enable alpha textures only for entities (func_wall, etc.).

But an experienced mapper knows all the rules and how the tools work and allowing it for the world is no different than, for instance, using the skip-remove tool.

Alpha textures should generally be limited to entities (func_wall, platforms, doors, etc). 

I have been testing my map with the mark V engine and for some reason all my models (map objects, items, weapons and monsters) randomly go black and monsters do it while patrolling.

I have checked the area the monsters patrol and my weapon does not go black. It is not a subtle change either, they are normal (full bright) one minute and then solid black the next.

Is there a special reason for this? (I know this is a counter productive thing to say but it does not happen in Fitz85) Is there a command line parameter I need to stop this? 

do they stay black with r_fullbright 1?

If yes, may be a texture problem.... if no, may be a lighting problem. (not a problem with your level, i mean a problem with the engine) 

I loaded the map, typed in r_fullbright 1 and I could see everything, no blackness. It looks like r_fullbright removes the light map from everything. It is very strange the blackness on certain objects, I thought it was specific locations where I have spawned stuff but it happens to wandering monsters.

I checked the original Q1 maps (played episode 1) and could not see any problems. I am using the latest compiler tools, hitting a couple of max limits but the engine still copes with it. 

It could be because I made the light check test against world submodels (plats, lifts).

I have a new revision to this with new features essentially done. I'll put a cvar in disable shadows/lighting from world submodels to see if that is the origin of the issue.

I also changed the lighting depth check (because MH did) from 8192(Fitz 0.85) to 2048 (original Quake) because MH had a comment about performance (8192 is a long, long distance).

The new version should be be available in 6-7 hours. 

The code that obtains shadow surfaces in Quake is actually very imperfect and can fail to "hit" a downtrace for lighting. Prior to the trace, I may have it reset the result color to black (full darkness).

Either way this shouldn't be hard to solve. 

Awesome, thanks :D I will download it and test my map once you got a new link ready.

Another bug:
When I start the engine (windowed mode) the +mlook does not work (even if I have the window selected) The old Fitz engine when it started windowed would never give up the mlook but at least I could test my maps without have to go full screen all the time. (I usually have to exit Fitz to use another application because the mouse cursor is not available to any other windows apps).

Is this a windows issue? a problem with active window focus? Or is the application (mark V) not able to control the mouse look when windowed? 

I also changed the lighting depth check (because MH did) from 8192(Fitz 0.85) to 2048 (original Quake) because MH had a comment about performance (8192 is a long, long distance).

Note: the reason i extended this distance is because CZG had a really deep pit in one of his maps, and the scrags floating over that pit were black instead of taking the lighting from the floor. 

the light depth check could be tweaked via a command line parameter? Something to set before the engine fully loads? 

How much the increased depth check costs to make. Is it really worth reducing back to stock quake for a little extra performance? 

The lighting depth check should be cvar-ized. Having to restart the engine to change something like this is stupid.

2048 is probably a little too low, especially with the "run the test for static entities once only at startup" trick (which was where the main performance impact came from in tests - and mind you this was quite an extreme case too). It also fails to light models properly in Zerstorer when this low.

It's probably worth finding the max top-to-bottom extent of the world and using that value as well - that'll be guaranteed to catch anything. 

end[2] = start[2] - (cl.worldmodel->maxs[2] - cl.worldmodel->mins[2]);

Why make the player do it when you can do you yourself? 

@Necros: The lighting check is one of those super slow recursive world checks. I'll do as MH suggests and default it to the Fitz 0.85 --- I wasn't looking to make an changes in how Fitz 0.85 operates (I lacked the imagination of a situation where 8192 could actually be needed, like the CZG map metslime said that induced him to raise the distance check).

[I'll add in MH's idea of a world bounding box check .. and if set to 0 will check lighting from unlimited depth. These things are going to push the update version to tomorrow it seems. But getting things absolutely right is something worth taking the time to do.]

@Sock: Any chance you could zip your map plus save the game exactly at the point the view model (weapon) lighting is wrong? It would help me avoid investigating things that aren't actually the source of the problem.