If you want to create games you should use every framework that helps you reach your goals. Building a engine from the ground up to make another game is just silly in 99% of the cases, the other 1% is a AAA+ title that needs every inch of juice to pull out those insane graphics.

Either way, LibGDX, unity or unreal you need to learn to code. LibGDX requires the most coding, its no engine but a framework, it grants the most freedom as well. It also ports to android for free. You can buy the additional framework to port to IOS.

Unity has a complete 3D interface and you can drag and drop stuff onto the scene (game world) then you can add premade code snippets to add functionality. But if you need something specific for your game you need to code it yourself. Unity is definitely easy, i had some kind of buggy FPS with my own 3D gameworld up and running in a couple of hours.

Not sure about unreal engine 4, i have dabbled in UDK and it is a nice engine you need to learn unreal script which should not be that hard i suppose.

If anyone could tell me the differences and pros/cons of using eclipse java only and using the engines (c++ or c# I assume) I would be very grateful.

You need to do much less when you use an engine or library. The tools and technology are already written and debugged. They give you a community, sometimes a very large community, that can answer questions and provide sample code and demos.

If you rely entirely on engines to do the work you may never learn about the internal workings. This can limit your ability to advance later in your career, good professionals need to understand the internal workings and the "how" behind complex systems.

Since vehicle analogies are so common, let's put one here. Would you rather re-invent the internal combustion engine, try to figure out how to engineer your own crankshaft and piston rods, or would you rather just go down to the hardware store and buy a prebuilt $100 engine to mount on your skateboard? Using a prebuilt engine won't teach you the internal mechanics of the engine, but it will help you complete your project much faster. Different engines and libraries provide different things, so compare them and try several.

If you're looking for a bigger list of engines and libraries, the mobile forum FAQ has a list of the most common cross-platform game engines.

Note that mobile development is usually not a good idea for beginners. You already have the complexity of developing a game, along with the complexity of learning how to program. Throwing the complexity of remote debugging and mobile devices into the mix just makes a more painful learning curve.

Note that mobile development is usually not a good idea for beginners. You already have the complexity of developing a game, along with the complexity of learning how to program. Throwing the complexity of remote debugging and mobile devices into the mix just makes a more painful learning curve.

Well you dropped the last one on me.

If mobile development isn't a good idea what would you recommend me?

Learn languages and just build game prototypes (like 2d sidescrollers) for pc and then move to more complex stuff?

Also does it even make sense to work alone? I mean i'm learning to code but i'll still need models, textures and a lot of other things.

Yes, usually it is easier to develop your first few applications on a desktop PC. It is much easier. You don't need to deal with concerns about remote debugging, remote connections, deploying data to the device, the constrained environment with limited memory and reduced processing power, and passing data between the host and guest operating systems.

I recommend you wait until after you are already comfortable with development then add in the complexity of mobile devices. It can be a few months, maybe 4 or 6, or if you don't program much maybe a year or two. Just some time to be familiar with development before adding in another level of complexity.

That doesn't mean you cannot do it. Some people jump right in trying to everything right at the beginning. Most people prefer a more shallow learning curve.