wraitii

It would, but the hard part here is actually number 1. Drawing the waves on screen properly is harder than I expected, at least to me. I may be missing something, though.

I'd really like some sort of Asterix cheat. Something like "Cacofonix" (the bard, Assurancetourix in French), that would replace every melee weapon prop with a fish (even if just for the graphics). Any idea how this could be implemented? (I'm willing to do this completely on my own, as it's really about the most unnecessary thing ever).

Very hard to have a nice result given that it must rotate 360° and it has to scale (and textures too) properly.

Tried that but it was unsatisfactory in some cases. Right now I'm using the "distance to shore" parameter to have the shader add a texture, but it's really not that good-looking. Your idea could work, I'll have to see how hard/easy to do it would be. I assume you'd require some sort of knowledge of perspective, and a way to keep track of waves/update.

I certainly can, and it certainly can be deactivated until someones finds a better looking way. The only problem here is committing to SVN efficiently.

Allright. I'm really having troubles with the waves, and i can't figure how to do it in any kind of a beautiful way. I have right now a system that has waves following the curve of the ground nicely, but it's ugly. I'm really unsure how to improve on that...

Myconid, would it be easier for me to work from a clean SVN (ie Kieran's master branch) or from your merged branch?

Okay I'm back from holidays and will try again with the waves stuff.

You need to reflect the water height too. So for each height that has a water plane, you'd need to re-render the whole screen. Even with proper optimizations, given that the terrain is rarely perfectly flat, this would probably be a fairly big performance hit.

Okay, so I've forked from the latest update in 0ad-git, and I've applied the necessary changes for my water patch, and only those. It's two patches because I forgot some stuffs on the first commit. I've tried (and succeeded) to compile and make it run from this clean patched branch, so I assume there is no problem (unless in the last two days someone changed everything but I don't think that happened. I've linked them as attached files... Would that be enough for committing in SVN? (I've changed the Atlas stuff so it doesn't use the super fancy water, and since I haven't yet ported to ARB It'll always use GLSL). 0002-Clean-commit-of-my-water-Shader-in-theory.patch 0003-Fix-messup-ready-to-SVN.patch

I can confirm this. And if the map is intended to be the Loire, I'd recommend making the river a tad larger. Thing's big.

To be perfectly logical, shouldn't that actually be reversed? Make a lot of money selling to the foreign market, make little selling the ally's goods in yours? I'm thinking a 80/20 mechanic would work better than a full 100/0 here.

BTW, I've given a shot at this but it wasn't so good, so I've restarted slightly the other day but I've been busy with the water stuff so if somebody else wants to give a go at texturing it, I'm OK with it

We'll change things when we get that in. But yes, it does look very nice thanks to the depth buffer (particularly around edges, where the normal fancy way gives really wrong results in some cases). I think I'll pick the second option, it sounds like it's the most obvious and the easiest one to implement. Atlas is slow because it recreates the texture every frame, in case the water height has changed. It ought to only when the water height changes, or it could be deactivated in Atlas... Is there a way to know in-code if it's Atlas or not? That should be a very easy fix.

Allright, I've finished the "clean" version. It fixes most remaining problems: water texture/foam texture looping issues, small depth buffer things that could prove distracting, and MEM_ACCESS crashes (I don't think there remains any, but it's always possible that there's a well hidden rare case). The only thing lacking is waves, but in the concern of putting this in the game, and since I believe there will be a lot of trial and error with waves, and I'm not there this week-end+, I think it's better. So it's all committed here. I'm not sure what the procedure on committing to SVN is, I don't think I can do it, or at least am unsure. Since there are probably other modifications than strictly the water stuff in this branch (which I believe I forked from myconid's merged branch some time ago, and anyway it's not up-to-date with SVN), would it be better if I took a clean SVN, merged with the water patch myself (by hand?), and then committed those files/gave those files to someone who can commit? Is the GIT mirror for SVN up-to-date? In this case I could fork that and patch, and then it'd probably be easier to put into SVN.

I do realize that . Okay, so I'm not happy with how the waves look. I could pull out a sort of workable system that would be bugged in some situations but would work in others, but I don't think it's really that good-looking (and given how the rest of the water now looks, I feel it's a step back). Therefore, I'll now try to clean up my patch for the water (without the waves) so it can be put, whatever happens, in alpha 11, and then I'll work on another solution for the waves: cutting the water mesh in two: a "coastal" part and a "high sea" part, that will allow me to have both shaders more efficient at what they do (the main bulk of the code would be the same, but some stuffs about foam or reflections could slightly change). With that system, I'll get rid of annoying mesh and z-fighting issues, and I think I could pull of some niceties a-la AOE3. Perhaps. Not as good. Perhaps. If that's okay with you.

Could use a stronger orange under the forest, but I really like the grass there.

Wouldn't it be easier to store them in subfolders?

I'll give a look at their shaders, could always learn something. I've got a system that should be all-right, but won't look that good…

Thing is, I don't think proper waves texture will do given the way this works. The necessity to do it 360° means that even when upscaling (what I've got on my working copy) there will be some (pretty serious) artifacts. The other solution would be to put waves only in some places, but that wouldn't work so well I think. The best is probably to use a 1 dimensional gradient and then apply a texture (like my foam texture) to re-create the illusion. I'm not surprised that even Red Alert 3 doesn't have waves on the shores. I'm still pondering simply having this done texturally to have the shape of the coast correctly.

You can probably merge some things together. In the ARB shader, to stay under the 8 texcoords used, I had to put the v_tex and the v_los vec2 together in a vec_4, for example.

Okay, so I've finished my first shot at waves. It rocks along with the foam, slightly moves up and down but not too much. The vertices are a bit rough, it doesn't look extremely smooth, there are a few textures that get squished here and there. I haven't had time yet to do that. The system is pretty easy to understand, harder to have in practice: I basically look for the shore, create a point there and another a bit farther at sea, and then connect all those contiguous points. This has the benefit of completely lacking the "holes" or overlapping when using quads. It may however produce artifacts if the water is a too concave pond (I think the RM map Oasis should be able to showcase that), or when two patches of land are too close together. The effect is sort of scaled with waviness for now, but it's not perfect. Here's the mesh around the afore-screenshoted island (you can see why it doesn't look smooth: it's pinned to the vertices' position to some extent): Here's two shots showing the waves when far from the shore and when on the beach: I'll try to smooth the effect now, and fix a few problems, and treat a few special cases. If somebody cares, it would be great to have a nice animated wave texture (from the sea on to the crashing on the beach, then the starting to retreat to the water). Or at least something a tad nicer than what I've added. Warning: I have not really optimized the wave mesh generation, so Atlas may be very slow for now. Also, it only works on superfancy, I think. It may crash in some circumstances, I haven't really tested yet.

Got some quad stuff going on with the waves, but I'm not really happy with it. It either looks rough (given that to do that properly I must go 360), or there's clipping problems... or rendering problems. And I can have other issues with concave stuffs, and with angles getting weird. I may go with some clever texturing job, or some clever axis-aligned quad, it looks like it would basically be easier and just as good. I'll see.

I think you should aim for half way between the first and last draft. If that doesn't work, choose the first.

I still think your HDR looks oversaturated, Myconid. While I like the effect for a screenshot, it'd probably get tedious in-game. Of course, you've probably set that as a setting by now .