Currently FriendlyFruit is just a proof-of-concept. There are various problems around 3D models and RPC that I have been trying to solve, and FriendlyFruit is the testbed. However, it does run, and it could be the foundation for a real Raspberry Pi multi-user game.
I've been using Panda3D on the basis that Panda is probably the easiest game engine to learn. Currently there is a demo scene that you can walk around. When you start the game, you will see a (very well modelled) building from 3dWarehouse in front of you.
If you walk to the right hand side of the building, you will see a crude humanoid figure walking up and down with a walk cycle that doesn't quite work! He was made in Blender by yours truly. In spite of the fact that he's not the world's greatest model, though, I'm very pleased. He has a raspberry badge UV-mapped onto his chest, demonstrating that the textures work. He also has a walk cycle of sorts, demonstrating that armatures and animations work. That's really all we need.
If you log on to the game twice, the characters will be able to see each other. They both start in the same location, so if one of them walks backwards, it will see the other. If one walks forward, it will be seen by the other.
If you want to get FriendlyFruit working, you will need to install Panda3D (the developer kit, not the runtime) from http://www.panda3d.org/. If you want to run your own server (and you do, because there are no public servers at this time) you will need the following additional dependencies:
MongoDB (http://www.mongodb.org/) and PyMongo (http://www.mongodb.org/display/DOCS/Python+Language+Center). I am using MongoDB 2.0.4 from http://www.mongodb.org/display/DOCS/Ubuntu+and+Debian+packages but you should be able to use the version from your distro without problems.
One word of warning is that Mongo tends to waste disk space if you aren't storing a large amount of data. When you create a new database, it assumes you will be storing a lot of data, and it creates some very large files for it! For this reason, edit /etc/mongodb.conf and add the option
smallfiles = true
to the end. This will cause the largest per-database file to be 32M. For the same reason, if you run multiple server instances, try to use the same database for all of them (unless they become very large). That way you only take the big file hit once, rather than once for each instance.
It's also worth adding the option
journal = true
so you don't lose your data in the event that your machine crashes. (This is the default with some configurations of Mongo, but specifying it explicitly doesn't hurt.)
- Google Protobuf (https://developers.google.com/protocol-buffers/). You should be able to install this from your distro. You need the Python support and the compiler.
- Scons (http://www.scons.org/). This is used for building the Protobuf RPC stubs. Again you should be able to install it from your distro.
Pyflakes (https://launchpad.net/pyflakes) is a lint tool for Python. You don't need this unless you're making a lot of changes to the source files, and you want to check for silly bugs.
There is a script called run-pyflakes which invokes this program on all the manually-written Python files. It's probably sensible to run this before checking changes into git.
Run scons in the top level directory, to build the RPC stubs. These are pure Python and presumably cross-platform; nothing is compiled to machine code.
If you pull new revisions from the repository, don't forget to run scons again.
You should have a file called server.cfg.sample. You probably don't need to change this, in which case the simplest thing is to symlink it to server.cfg. If you do need to change it, make a copy instead, and then edit the copy.
The network section allows you to change the addresses that the server listens on. You could for example make it listen only on the loopback address, for extra security during testing.
The database section gives the details of your database server. The host and port should normally be left unchanged; you only need them if you want to run the server on a different machine to Mongo.
The prefix allows you to run more than one instance of FriendlyFruit with a single database. The default ('main.fruit') tells the server to use the database 'main' and to prefix all collection names with 'fruit'. To run two instances against the same database, you could give them prefixes like 'main.raspberry' and 'main.strawberry'. If they got big enough that you wanted to use different databases, you could use 'raspberry.fruit' and 'strawberry.fruit'.
Start by running the server. It takes no arguments because all its configuration information comes from server.cfg.
Once the server is running, you need to create a user account. Run the client like this:
./client --register -u myname -p mypass localhost
You should see a response from the server indicating that your account was created.
To play, you do much the same thing:
./client -u myname -p mypass localhost
(I know the handling of the password isn't very secure or convenient, but it's not the most important thing at the moment.)
Once the client is running, you control the camera with the WASD keys, plus the left and right arrow keys.
I've tested this with Ubuntu 11.10. If that's what you're running, and you have a 3D-capable video card, you stand a good chance of success. Theoretically Panda supports quite a lot of platforms, though, so you should stand a reasonable chance on other systems too.
Finally, be aware that this code is at a very early stage. Don't expect anything polished, instead expect to mess around quite a bit in order to get it working.
Please let me know how you get on.
Not all files in 3Dwarehouse are downloadable. This is annoying because non-downloadable models clutter up the site and make it difficult to find what you want. One tip is that the Advanced Search screen has an option to ignore models which cannot be downloaded.
Once you have a downloadable model, you have to decide whether you will be editing it further with Sketchup. If you won't, download it in COLLADA format and go to COLLADA to Panda, below. If you will be editing it, download it in Sketchup format. When you have finished editing, go to Sketchup to COLLADA.
If you have a model in Sketchup, and you want to use it in a game, you should export it as COLLADA. The export options should be set as follows:
[X] Export Two-Sided Faces
[ ] Export Edges
[X] Triangulate All Faces
[ ] Export Only Selection Set
[ ] Export Hidden Geometry
[ ] Preserve Component Hierarchies
[X] Export Texture Maps
[ ] Use "Color By Layer" Materials
[X] Preserve CreditsThis will give you a usable COLLADA file. Continue to COLLADA to Panda to get it into Panda.
Run dae2egg on your file. This will give you a text file which can be loaded into Panda. You can optionally run egg2bam on the egg file, to compress it and speed up loading.
You may find that the paths to your texture files don't end up pointing to the place where you want to keep them. If this happens, it's often easiest to fix it by editing the egg file (with a text editor or a script). It's a simple text format so it's easy to make this kind of change.
If you have a Sketchup model, I would suggest that you edit it in Sketchup if you can. Converting it to another format is unlikely to preserve all the information in the file. However, you may be forced to convert, perhaps because you don't run Windows or because you want to combine it with something that is already in Blender.
Blender ships with a COLLADA importer, but there is a bug which results in the wrong textures being assigned to some of the faces of the model. You could reapply the textures, but this would be rather tedious for a complex model.
A better option is the Open Asset Import Library, http://assimp.sourceforge.net . At the time of writing, you need to download a pre-release version, like this:
svn co https://assimp.svn.sourceforge.net/svnroot/assimp/trunk assimp
I ended up with revision 1231. If you try this and it doesn't work, try checking out this exact revision.
You then build Assimp yourself, following the instructions on the website. Once you have built it, convert the COLLADA file to Wavefront:
assimp export model.dae model.obj
This can be imported into Blender using the Wavefront importer.
(There are other programs which convert COLLADA to Wavefront, but as usual, many of them give unsatisfactory results. In particular, many of them fail to export the textures properly.)
Follow the first option (YABEE) on this page:
http://www.panda3d.org/manual/index.php/Converting_from_Blender
Before exporting you must apply all modifiers except the armature. Be very careful that you undo this after exporting, or (assuming the modifiers are required) you will lose your work.
To create for example a walk cycle, first create it as a regular Blender animation. You can have several in the same file, one after the other. Then, during export, enter the start and end frames of each animation into YABEE's dialogue box.
Underneath this dialogue box are various options. Set them as follows:
[ ] Animation only
[ ] Separate animation files *
TBS generation: No
Tex. process: Simple
[ ] UV as texture
[ ] Copy texture files* This one is up to you, but I think it's convenient to store everything in one file.