Spring RTS Engine

Holy Jesus on crutches, with three flaming sticks, two potato masher grenades and a tommy gun gowning out of his nose...

Thats how cool that specular lighting is.

I still can`t get it to load,, keep getting this same error


failed to build fragment shader

muffdvr, make sure you extracted the directories intact into yout ta spring folder.

That is so say, umm there are folders inside the zip. Make sure that the folders get unzipped to the taspring directory.

Umm am I making any sense? And Muf, do you have a microphone and Teamspeak? Pm me.

Yup,, did all that but its a firm no go,, I dropped all the files contained in the zip into my spring directory but I just keep getting that error ""failed to build fragment shader"" and get chucked back to the desktop, I have an ATI card please tell me this is not another nvidia only feature

I had to check the "never use shaders for sm3 maps" box to get the map to load and of course I got no bump mapping of the textures and no specular stuff like I said I have an ATI card.

Updated screenie. Wow, the new lighting options are absolutely superb. Totally kick in the nuts stuff.

http://img85.imageshack.us/img85/2553/screen234im7.jpg


@Muf, it shouldn't cause ATI any issues that I am aware of. I myself am on nvidia, but all the new stuff is within ATI specs afaik.

Well I need to see a post from an ATI user to confirm this, I have completely un-installed spring and done a clean re-install and then dumped the contents of the "spring-sm3-v4.zip" into my spring directory and it still won`t load the map unless I turn off the specular stuff.
Thats where Im at right now,, later on I will post screens of my directories for you to look at thanks for your patience
unless I can get it to work for me theres not much point in myself persisting with this format any further which is bloody disappointing

These shots may help

I get a comparable error on windows (with nvidia): "failed to build vertex shader"
last part of infolog (seems inconsistent / cut off?) This happens both with 0.72b1 data plus the -v4 zip, and also if I replace the shaders with the shaders from SVN.

It doesn't seem to matter which settings I use, only "use no shaders for sm3 maps" does work (obviously).

On Linux, it doesn't work either but there it's because DevIL somehow can't load the images (didn't yet get it far enough to test shaders). in LoadHeightmapFromImage() the file is succesfully opened but the ilLoad returns an error. After adding some code to show iluErrorString(ilGetError()); the error appears to be "png library error", even tho everything else on my install can perfectly fine read PNGs (and AFAIK it all uses the same libpng).

Obviously I converted the heightmap to .bmp and to .jpg and tried again, .bmp failed to load too without a meaningfull error message, .jpg did load, but then it couldn't load the typemaps . The conversion trick didn't seem to work on those

Also note that in all cases the binary did open the file succesfully, but it was ilLoad that was bailing out.

This linux thing could very well be a DevIL or CFileHandler bug tho, I'll investigate some more another time.

EDIT: After installing 91.31 nvidia drivers it changes to a "failed to build fragment shader" error. Last piece of infolog.txt: EDIT 2: got it, need to set SM3MaxTextureStages to 5, I suggest you try lowering that config value too mufdvr222

Its nice to see someone else with a similar problem, I tried lowering the "max texture stages" but it still won`t load.

EDIT 2: got it, need to set SM3MaxTextureStages to 5, I suggest you try lowering that config value too mufdvr222

You really shouldn't have to do that with an 7800. It is even worse than "don't use shaders on sm3 maps" and switches to the absolute ugliest mode of rendering. It's also been a while since I tested it, so I'm not exactly sure what happens there :)

Maybe the nvidia linux drivers are older than the windows ones?

mufdvr222: It should work on ATI, but indeed it might be an ATI problem right now. I'll test it on another pc with a Radeon.

Thanks for testing everyone!

Read my posts a bit better (and also check my PM), that was on Windows with 91.31 drivers. On Linux I dont even get that far because DevIL doesn't want to load the files.

EDIT: hm now I got it working here with SM3MaxTextureStages at 20. I wonder what's I changed in between.
(it could be that I didn't test the shaders of the zip with the 91.31 drivers)

EDIT 2: MinGW build of SVN HEAD fails the same way as Linux build.

The nmapsys2 branch is set up a bit stupid , I wasn't planning on using it long. The shaders in the zip are the most recent, but they aren't in the nmapsys2 branch (because it is only a branch of the rts dir).

But since your errors are about DevIL, I really don't know how that should be fixed... :/

Possibly the DevIL versions used on MSVC/MinGW are different versions / configured differently, or gcc is stricter in some way that breaks ilLoad()... anyway, I'll find out sometime.

About the issue on Windows, indeed my old drivers (84.XX IIRC) borked with "vertex program error", no matter whether I used trunk or -v4 shaders, the 91.31 drivers bork with "fragment program error" with trunk shaders and work fine with -v4 shaders. So I should just use compatible stuff together

Looks great btw!

the specular things still look more like plastic than steel, just like in all games.

bamb wrote:the specular things still look more like plastic than steel, just like in all games.

True, but look at Crysis, it has a 16 million Euro budget and still looks plasticky.

I still like this though. And I guess much of the shininess is due to the textures or blendmaps or whatever in this case. I'm sure you can have parts of the map shiny (as in wet rock) and some parts dull and without any specularity at all (as in sand).

I'd love it if maps could be made with a more gritty, dusty look too them, but specularity is useful and nice, so hey, great job :D

Actually if people want to attempt their own stuff, mappers are free to include their own "shaders/terrainFragmentShader.glsl" in the map archive, so in theory you could completely modify the lighting model for a particular map. (But preferably after I debugged the thing for radeons)

AFAIK basically all games use the blinn or phong lighting model, which looks a bit like plastic. Find me a good lighting model that doesn't require more than a normalmap and I'll implement it :)

mufdvr222: My radeon gives the same error, so at least we know what causes it now.

jcnossen wrote:
mufdvr222: My radeon gives the same error, so at least we know what causes it now.

Thats good to know
FYI I have a Radeon x300/x550 series

Same problem on my X850XT can't wait to get back to testing.

works on my Radeon 7000 good fps too

Err, ATI takes a bit longer because of my exam in 2 days and yet again missing OpenGL functionality in the ATI driver. I read on a few gamedev sites that for ATIs you actually have to ignore what the driver says it supports, and assume it will work :/