MV3D Development Blog

We are pleased to announce the release of MV3D 0.80! The highlights of this version are a revamped toolchain which should enable a better workflow for building new content. Some of the improvements include tabbed tools instead of docked windows, a property grid for editing, and support for 3rd party asset types. Also in this version is the first example gameplay system– the stats system. To support it is a new component system that loads objects from an xml template and attaches components such as the new stats system. Finally, many improvements to the low level server code were made including initial support for asynchronous database connections, and a new JSON based protocol which can be used instead of PB to support clients written in languages other than Python.

Contributors to this release include:

• Mike Handverger
• Judd Cohen
• johne5
• Mythros

MV3D is an open source multiplayer game and virtual world framework written in Python. It encompasses a scalable server with dynamic load balancing, a robust set of content creation tools, and an extensible 3D client application. MV3D provides the foundation to build anything from 3D chat rooms to full MMORPGs with the goal of letting you concentrate on creating a unique virtual world instead of the infrastructure to support it.

The next release will be 0.85 and will focus on tools, usability, and example gameplay.

For more information on MV3D and this or future releases, please visit the website. The full release notes for version 0.80 are available online. For further inquiries, feel free to stop by our IRC channel on irc.freenode.net #MV3D.

Retrospective

What went well?

• py2exe client and tool releases include support for both Panda3D and Ogre3D.
• Datastore load test showed excellent performance. 
• Multiple projects are now using MV3D in various forms.
• Relaxed release process. Otherwise it would have been another 3 months! Time will tell if this is a “well” or “less well” though.
• Improved support for traditional static world MMOs.

What went less well? 

• 18 Months is way too long in between releases.
• MV3D is still not very user friendly.
• Not many flashy features were added.

We are very happy to announce the release of version 0.75 of MV3D! This was mainly a bug-fixing release with more than 65 bugs squashed. Also, in this release, MV3D gained support for Linux with the Ogre3D renderer along with Mac OS X with the Panda3D renderer. This means that MV3D’s client, server, and tools are available on Windows, Linux, and OS X! With bug-fixes across the whole platform, this is also the most stable release of MV3D to date.

MV3D is an open source multiplayer game and virtual world framework written in Python. It encompasses a scalable server with dynamic load balancing, a robust set of content creation tools, and an extensible 3D client application. MV3D provides the foundation to build anything from 3D chat rooms to full MMORPGs with the goal of letting you concentrate on creating a unique virtual world instead of the infrastructure to support it.

For more information on MV3D and this or future releases, please visit the website. The full release notes for version 0.75 are available online. For further inquiries, feel free to stop by our IRC channel on irc.freenode.net #MV3D.

Screenshots:

Retrospective:

What went well:

What went less well:

The next release is 0.80 which is set to include features aimed improving MV3D’s usability as a platform for traditional MMORPGs such as instances along with increasing tool usability and functionality. As mentioned in the less well section, we need more help. Even just help testing out new features would be very much appreciated. MV3D is also a great way to learn about client / server game architecture!

In svn trunk, MV3D now supports all three platforms! That includes server, client, and tools. This is a pretty big deal as until fairly recently, MV3D didn’t support running anything but the server on Linux. Both Ogre3D and Panda3D are available on Windows and Linux. OS X only supports Panda3D at the moment. There’s a build slave configured for each supported OS combo, and there’s now a wiki page showing supported platforms.

Panda 3D on OS X:

Ogre3D on Linux:

The 0.75 release is wrapping up now, but we’ll need as much testing on all supported OSes as we can get. Unfortunately, there isn’t an easy install for Ogre3D on Linux just yet, but that’s something we’ll work on in coming releases. The first step here is to get it all working. Next up is making it easier to use. On the other hand, Panda3D is a breeze to install for any supported platform.

I’m currently working out what the next release will bring. There are a lot of options and a lot that needs to get done before we hit 1.0. Stop by IRC or the forums if you are interested in helping out or finding out more about MV3D.

We are pleased to announce the release version 0.70 of MV3D! This release includes a massive amount of new features and over 70 bug-fixes. The main focus was on usability and involved a large amount of work on tools. In 0.70, Panda3D support was brought up to par with animation, terrain texture splatting, and sky-domes. One of the things we are most excited about is the Windows development environment installer. This is an easy installer that will quickly set up any Windows system to develop MV3D. If you’ve been watching the project and want to get involved, there’s no better time!

MV3D is an open source multiplayer game and virtual world framework written in Python. It encompasses a scalable server with dynamic load balancing, a robust set of content creation tools, and an extensible 3D client application. MV3D provides the foundation to build anything from 3D chat rooms to full MMORPGs with the goal of letting you concentrate on creating a unique virtual world instead of the infrastructure to support it.

For more information on MV3D and this or future releases, please visit the website. The full release notes for version 0.70 are available online. For further inquiries, feel free to stop by our IRC channel on irc.freenode.net #MV3D.

Screenshots:

Retrospective:

What went well:

What went less well:

Release 0.75 will be focused on bugfixing and adding full support for Linux + Ogre and OS X + Panda. We are currently looking for help porting and testing on these platforms. Let us know if you are interested!

Probably the biggest barrier people have to helping out with MV3D is getting a development environment set up. It’s pretty easy (on Windows) to run from the prebuilt binaries, but those don’t allow you to edit the source code. As of today, a Windows install package for the development environment is now available! This includes:

• Python
• Panda3D
• Ogre3D
• All of the prerequisite Python libs.
• Combinator - a tool used to manage development with MV3D’s dev process.
• Directx and vc redistributables.

There are a few caveats though. You must pick either Panda or Ogre3D. Although it will install both, one of them won’t be fully configured since there’s only one active Python install. Also, there will be many sub-installers that come up. Make sure to choose the default options on everything with those.

Finally, and most important. Do not use this if you have an existing Python installation other than Python 2.6 in C:\Python26! It will mess up your current install otherwise.

Thanks,

Mike

We are pleased to announce the release of MV3D version 0.60! This release focused on scalability of worlds and includes support for splitting a single area across multiple server processes with automatic load balancing and redundancy. Areas can now be connected together using gateways to build worlds limited in size only by the amount of available hardware. The Overseer Cluster Management tool was upgraded in order to better handle many processes across multiple physical servers. Camera controls across all content tools have been unified. Better support for Panda3D clients was added as well.

MV3D is an open source virtual world and multiplayer game framework written in Python. It was designed with scalability in mind and is able to distribute a world across as many servers as needed while dynamically balancing the load. The simulation framework is not specifically slanted towards any one genre of online game, and can just as easily be used for a space game as a fantasy setting. Objects on an MV3D server are simulated using the ODE physics engine. A single server can host thousands of simulated objects. MV3D can use either Python-Ogre or Panda3D on the client for visuals.

For more information on MV3D and this or future releases, please visit the website. The full release notes for version 0.6 are available online as well. For further inquiries, feel free to stop by our IRC channel on irc.freenode.net #MV3D.

Video and screenshots of our new Demo World, the Town of Wellston:

Retrospective:

What Went Well:

• Last minute shader and dynamic light support!
• The client is now authoritative for its movement which means things are much smoother.
• Fixed some crazy bugs including a few with the persist code which were pretty serious.
• The persist upgrade system works as designed! The old demo world’s databases were automatically updated with no intervention or errors!
• New Ogre .scene import functionality in Builder is pretty awesome and saved hours of work on the demo world.
• New triangle mesh colliders including automatic converter from Ogre/Panda meshes makes indoor world building much easier.
• The new representation file format is a definite win!
• Official support for Panda3D client and tools on Linux.

What Went Less Well:

• Client movement is still jerky (but at least we know it’s a client only problem).
• The release took 2 months longer than expected.
• Panda3D support is still not 100% there. Missing: sky, animation, and terrain texture splatting.
• The tool workflow needs to be simplified.
• Linux support for Ogre3D is still lacking due to inability to successfully build Python-Ogre.

Release 0.70 Planning:

The next release will be 0.70 and is focused on tightening up the toolchain as well as adding new client side features. The current list of tickets planned is available here. We will be meeting on #MV3D to go over the plans for 0.70. I’ll post a comment here when we decide on the time and date.

The next release is nearly ready. We’re looking to get some people to help with testing before officially giving it the thumbs up. You can grab the latest binaries and check it out. The demo server will be updated over the next week, and expect to see a new and improved demo world. Here’s a little preview:

One exciting thing you’ll notice with the binaries is that there are now Panda client and tool binaries for Windows. You’ll need the Panda3D runtime installed to use those.

Look for the new release in the next week or so if all goes well!

MV3D 0.50 has just been released! This release took quite a bit longer than intended and ended up being 2 months late. On the flip side, there’s a lot of new functionality such as support for the Panda3D rendering engine and a path-finding system for NPCs. In addition, many bugs were squashed and the performance problems of the last release are pretty much gone.

What Went Right:

• Panda3D support is great!
• The NavMesh + A* path-finding system is pretty much industry standard and includes automatic NavMesh generation.
• A client Patching System is available complete with patch building tool.
• Documentation. There is now a User Manual.
• 50 bug tickets closed! This is the most stable release to date.

What Went Wrong:

• Underestimated the time it would take to write a fully functional NavMesh generator, which extended the release by 2 months.
• Started documentation too late in the project (this isn’t specific to this release).
• It is still pretty hard to configure a whole cluster. This will be addressed in the next release.
• MV3D’s tools could use a polish pass and more documentation.

I’m very excited about the next version as well. One of the planned features is splitting up of areas across multiple servers. Another big push is for making MV3D more accessible to new users. Part of that is expanding the user manual as well as simplifying configuration.

For more information on MV3D, see the website at www.mv3d.com, or the release notes. There is also a #MV3D chat room on irc.freenode.net which has been getting more use lately. If you don’t have an IRC client such as mIRC, Pidgin, or Trillian, you can use the web interface.

MV3D 0.44 is ready, and 2 days ahead of schedule! The release was focused around building an asset tool-chain to facilitate adding content to MV3D worlds. In that respect, it was an outstanding success. The sprint only lasted about 3 months, which is a lot shorter than most MV3D releases. I’d like to keep up that pace moving forward. The most amazing thing to me for this release is that for the first time, you can spin up an server and populate it completely using tools!

The Good:

• Solid asset tool-chain!
• Much more stable than 0.42.
• Guide (the GUI data-binding library) is awesome!
• Overseer cluster manager really speeds up iteration time and should really help managing servers.
• MV3D’s Client now works on Linux!
• Users can download server, client, and tool binaries for Windows which makes it incredibly easy to try out MV3D.

The Bad:

• Performance is pretty horrible on the client and server.
• More tools still needed. Plus, the new tools are just at a bare minimum state.
• Even though there were many bugs fixed, quite a few big ones remain.
• Not enough work was done on guidece and guideweb, the CEGUI (in game) and web versions of guide respectively.
• Even with the GUI, it’s still really hard to set up a cluster of MV3D servers. Partly due to bugs but mostly due to all the settings that have to line up exactly.
• Still some bugs that can cause datastore corruption on upgrades.

All in all, it was a solid release and a lot was done in a short time. For next release, which will be 0.50, the current plan is bugfixing, performance, and properly functioning areas with gateways.

Including the plans for next release, there are a few major features missing which are required for a 1.0 release. The ones I am currently aware of are load balancing an area across multiple servers and pathfinding AI. I’m starting to put together a roadmap for these. It will be updated more soon.

Head on over to MV3D and get the new version! Binaries will be up there tonight. If you are in a hurry, you can grab them hot off the press from the Hudson builder.

MV3D 0.42 released today! Check out the release notes or download the MV3D Windows Client! In this release, the core systems were made ready for tool building and content creation. This included a redesigned persistence mechanism and a cross system GUI layout.

The Good:

• New persistence system is pretty awesome.
• For the first time ever, you can export or import sections of the game world!
• Pretty cool start to a new way to build UIs which should save tons of time and allow for quicker iteration as well as reusability.
• Made an unreleased play in browser app for MV3D.

The Bad:

• New persistence system may have some performance issues.
• Tons of bugs still need to be squashed.
• Didn’t get to any tools which was the original focus of the release.
• No visual features added– nothing cool to show off really.

What’s next? Well, the next release will focus on the content pipeline. This means upgrading Ogre (since it’s long overdue) and tools, tools, and more tools. The one major unknown is how the content pipeline will actually work. Ogre has a lot of half functional or very basic exporters for various 3D modeling packages, which is an unfortunate start to the pipeline. That’s not so much something that MV3D can or should solve, but it’s fairly disappointing.

After exporting the raw files, they have to be added as assets in MV3D. Currently, RED can do this in a very basic way. It does include the ability to SFTP the asset files to the server that will be distributing them which is handy. This will need to be extended and made smarter and easier to use. RED should definitely get in app previewing of assets. It’d also be nice to add a preview image to assets in general. This could be displayed in RED, in the IGE, or on the web editor. Speaking of the web editor, it’s completely busted more or less. That’ll need fixing, which may come in the way of guide (the GUI framework which also supports web GUIs). What’s awesome about fixing it with guide is that the same GUI can be used in IGE or RED without any modifications.

With the assets imported, the real fun begins. RED will need a plugin that lets you add and arrange objects in a vacuum (i.e. not anywhere live in the game). Then you can use Freeze (the importer/exporter built in 0.42) to save off pieces of the game world. An additional RED plugin should be able to connect to a server and edit existing areas. Here you could import the stuff you created before and place it in game. This should also have the same functionality of the other plugin so that you can edit anything you want.

One other missing piece is the collision volumes. These need to be manually placed (or generated from static geometry in some cases). This means the other required tool will be something which is able to position colliders for models.

But wait, there’s more! There needs to be a way to edit bodies of water, and in fact, the current ocean object should probably be made into a water volume since right now it’s just an infinite plane. Similar to water, the sky should be editable, and if you are editing the sky, physics properties of the realm need some love too.

Finally, it’d be nice to be able to have more than one area in the game at some point. Part of the RED area editor should include connecting areas together. At this point, it’s been so long since this was tested, that it may not even work any more.

Sounds like I have quite a few tools to write. Any volunteers to help?