Horde3D

Hi,

I finally decided to get my hands a little dirty with Horde. I can't remember what I've posted in the past, but my intention is to use Horde from a .NET/Mono environment, so when downloading the engine today I went straight for the .NET bindings.

I installed the latest release of Horde and the latest release of the .NET bindings from the Download links on the website (not from the Subversion repository, if that matters). I followed the installation instructions, extracting Horde to a new directory and then extracting the .NET bindings in the same directory (which matched up so that I ended up with a C# subfolder under the Bindings folder, just as it seems I should).

I opened up the project file, and there were a couple of minor issues (I had to convert to Visual Studio 2008, the project files pointed to \Source\Horde3DEngine\Horde3D Engine.vcproj rather than \Source\Horde3D Engine\Horde3D Engine.vcproj and same for the Utils, and I think one or two other small things) but there was no real trouble at all. I spent a little while looking through the .NET wrapper (it's really impressively light and elegant compared to similar wrappers for other engines/libraries, no doubt largely due to Horde's small C interface).

Then I decided to test everything out by building the samples and playing with them. Here's where I ran into trouble. The projects built just fine, but when I tried running them things were definitely not good. I could see the models and animations, and I could move the camera, but everything was orange and no textures were showing up. This was true in both the Chicago and the Knight sample. Additionally, there was a somewhat paler orange rectangle overlayed in the lower-right-hand corner of the screen.

I had intended to take a screenshot of this behavior to show you guys, but unfortunately I'm now having another problem -- the Chicago.NET application will no longer launch, instead just briefly showing a gray window before crashing. I'm assuming that restarting my computer will fix this issue (temporarily at least) but I have this post mostly written so I wanted to send it off before I try.

Yikes. Well, I figured out that the reason I couldn't open the samples anymore was due to a conflict with FRAPS, which I had opened in an attempt to take a screenshot (weird, right? I've never heard of FRAPS conflicting with anything). So I disabled FRAPS, and tried launching the Chicago sample again . . . this time, it "worked" for about thirty seconds and then hard-crashed my PC. (I got a blue screen with a note telling me that my computer had shut down to protect itself from damage, that I should restart my computer if this was the first time it had happened, and to contact the vendor if not. It wasn't even the normal BSOD, which I haven't seen in months . . . this was white block letters at 640x480, and looked more like a BIOS setup screen.)

Am I possibly having driver conflicts, rather than this being a .NET issue at all? I have an ATI X1950XT, and I know there have been ATI driver issues . . . and I haven't updated mine in a month or two.

I'm back to thinking it's something with the .NET samples. I just tried the normal Horde3D samples, and they seem to be working perfectly well for me.

The trouble is that I'm not sure why it's the .NET samples. I've been looking around, trying to find whether there's some texture path set wrong, or anything like that, but it kind of looks like the material paths are determined by an XML file, which is the same for both the .NET and the normal samples.

I know this is a stupid issue and no doubt has an easy fix, but I'm not really sure where to start looking. I've been teaching myself to program for the last couple of months, and I've made huge improvements, but even though I'm to a point where I can more or less understand the code I'm reading, I don't have much experience with large scale software, or familiarizing myself with the structure of another developer's code, or the "nitty gritty" details of Visual Studio's projects/solutions, properties, resources, or build system -- all the things I think I would need to be able to figure out what's going wrong.

So if anyone has any thoughts . . . I'd really appreciate it! Oh, and now that I've figured out FRAPS was my problem, I can take screenshots using ATI Tray Tools. The first two are the .NET versions, and the last one is the standard Chicago sample just to show that it seems to be working fine for me.







(Broken though it is, the Knight sample looks ridiculously awesome like this -- I actually think I prefer the accidental stylization of my broken .NET version to the original!)

Normally the orange "textures" are due to the missing of textures. The default texture is just an orange pixel. Did you check the EngineLog already?

I only ajdusted the Pyropix .Net wrapper and made a quick test if the samples work. There might be still some issues since I didn't have the time to test it intensely.

I did check the EngineLog, yes. I wasn't totally sure what to look for, so I actually went ahead and used DiffMerge to look for differences between the EngineLog for the .NET Chicago sample's log and the standard Chicago sample's log. There actually were no differences at all, except in the numbers in the left column of the table display (I assume this is the time passed in seconds?). Both logs had a couple of issues: notes that "Extension ARB_texture_non_power_of_two not supported" and "Extension EXT_framebuffer_multisample and/or EXT_framebuffer_blit not supported", and a few warnings about "Validation failed - samplers of different types are bound to the same texture image unit".

However, these warnings were exactly the same in both logs, and yet the normal sample works fine while the .NET sample doesn't show textures.

Of course my first thought when seeing this was that the textures weren't being loaded, but I'm not sure where to look to try to fix it, and I don't understand why the .NET samples have the problem and the normal ones don't.

EDIT: It's so weird. I've been going through the code for both the .NET Knight sample and the standard Knight sample. I didn't think I was comfortable enough with C++ to do that, but it's similar enough to C# (at least superficially) that I'm not having any problems. The thing is, though, that the relevant parts of the two samples (it seems to me like the problem would have to be in the App code, where the resources are loaded) are virtually identical, line by line! I'm very confused.

The .NET bindings on the pyropix and corresponding SF page are still for version 0.13. Do you use the latest ones from here?

http://mm-werkstatt.informatik.uni-augsburg.de/project_details.php?id=48

Yes, those are the ones I'm using. There are quite a few references in the code (which are generally commented) to Horde 1.0.0. I'm also using the version of Horde 1.0.0B from the download page -- which I know isn't the most up-to-date (since I've seen in the forums that a few patches have been committed to version control), but which I suspect was the version the .NET wrapper was written for.

I actually still haven't been able to resolve this issue, although I haven't had more than a couple of hours to try fixing it. Does anyone know if the .NET bindings and samples actually work at this point? Has anyone tried downloading a fresh copy of Horde, a fresh copy of the .NET bindings, and just building/running the demos? I'm not so much asking anyone to do that -- it's a lot to ask of volunteers -- but if someone else has already tried, I just want to know whether they do work for anyone else, which would at least indicate that there's something going on with my system in particular.

It was indeed written specifically for a newer svn build.

I was having the same issue of textures not appearing a while ago. Rebuilding the engine from the svn resolves it.


Just thought I would mention it.. better late than never.

Thanks for reporting that. Sorry I forgot to have a look at that problem again.