I think more advanced animation capabilities like something Unity has. Unity has an easy way to build a state machine on top of the animation controller that lets you set up states and how to transition between them. It’s pretty powerful and would be a wonderful addition to JME. I know I have something like that setup for my project but not nearly as awesome.
Particle systems like those found in Unreal 4 and Unity would be pretty awesome too. Thinks like animation curves for particle attributes would allow a lot of creative control. I had to setup a system like that for my game but it isn’t integrated into the sdk which is cumbersome.
Well david is a bit too old to be a student (no offense @david_bernard_31 )
If we pick something core, like particles, animations or deferred rendering, we have to make sure the student is very capable. Also if we go to GSoC this year, and I’m a mentor, I’ll be a bit more picky on the student.
Also remember that the project must appeal to Google so that we can have slots…
I’m not sure what exactly the rules are for the GSoC but would it not make sense to make an open source game using JME rather than trying to make add-ons or improvements to it.
That way students who want to use the engine learn a lot about it from someone with advanced knowledge of openGL and what not.
You’d think students wouldn’t be in the position to contribute directly to the engine due to inexperience and lack of confidence to take on a huge project, but rather would learn to use the ins and outs of the engine itself with the help of a more experienced developer.
Yep, it would. We had it as an idea in earlier GSoC applications. For some reason we didn’t in 2014, but we did have “Android Demos”. With the experience we’ve had so far with our newest example game project, I would love to do others like it. We could have an overarching “Example Game” category and think of 3-4 suitable projects.
As mentioned, improving the assets pipeline is a major priority. The FBX Importer would definitely make for a great project. @Eirenliel and @tort32 are there any students among you in your team?
Heh, sorry for dragging you into the GSoC discussion with no warning Search for “Google Summer of Code” and you can find out more. I’m just talking about making your FBX Importer a possible GSoC project, i.e. a student would get paid to work on it. A core developer/contributor would probably mentor it, but the student and mentor would be collaborating with you. If it’s confusing don’t worry, more will be explained later
I think, improving assets pipeline can be a great project and awesome improvement to jME. We had some problems with organizing assets in our game and developed some tools for it. FBX importing is not the only problem, there are a lot of tasks that can be supplied.
In case of formats support it will be awesome to have support of substances from Substance designer.
Substance could be a really nice addition.
Though it doesn’t just need a loader, we need to account for the “dynamic” nature of it in the engine itself, and maybe provide some UI as they have in UE4 and Unity.
If it comes to life I’ll contact the guys from Allegorithmics, and se what we can work out.
Oh it comes out that the discussions that followed my post cover pretty much anything I would’ve want to ask and the ideas list you created too. I will have to think about an idea if I have one.
→ For commercial models it is quite common to simply have a all in one animation, plus a readme that say frame 0-101 walk 102-200 run ect. A programmatical way to split animations for jme would be nice. (plus maybee gui in sdk for this?)
→ Generally all model loaders do not allow to add a custom material hook, instead it is always necessary to do a later iteration step over the whole model, replacing the materials with the own ones. (becomes an issue when the jme material losses data already, one example are some blender material settings that can be read by the loader but do not map to general jme materials)
→ apart from that some features that apparently seem to work in unity does not work when going the .blend file way.
Maybee its possible to implement at least a few more commonly used ones.