"Better" is very subjective. The important question is, do you have the budget and resources to make Cuboidz into a 2.5D game similar to the game you posted? It's definitely a [i]lot[/i] more asset-hungry than the existing version. If you're a solo dev, you're adding months, if not years, to the schedule to try to create that 3D content yourself. If you contract it out, it's definitely going to cost a lot more than a few simple sprites will cost. Would the game be fundamentally different, gameplay-wise, or would it be purely an aesthetic change? If purely aesthetic, you can just write it as pluggable back-ends, do active development on the pure 2D back-end, and if budgetary concerns change focus efforts onto the 3D backend. That way, you don't inflate your schedule only to realize, months in, that you lack the resources to complete it.

Budget I have, yes, since Unity3D has a free version that I've tinkered with and love. I've made some small games in Unity before, but never anything as detailed as what I'm thinking of.

Resources wouldn't be hard to create at all. I would basically just take the 2D images I've created, and transform them into 3D. The tiles would be made of cubes and it would be easy to make the textures tileable, since I've already sort of created them. The only tricky part I see is with unwrapping, but Blender actually makes a fairly easy process of that!

I am solo-deving this, so it will take a lot more time for sure! But I'm willing to spend the time on it in order to make it a good game.

The gameplay will be very similar, but I have several ideas for elements to implement that wouldn't be possible in the 2D version. For example, if you've played Super Paper Mario, you're probably familiar with the flip into 3D mode. I plan on a similar thing, but it will just change the view. There will be multiple views one can use that will reveal certain paths previously unseen, which may contain collectibles/powerups as well as secret exit paths leading to new realms in the game world.

Another reason I'm thinking about 3D is how much more fun it would be for the player, since they'll be able to see the depths in all of the levels. I want it to be completely immersive, and very detailed. I realize this will take a lot of time, but I'm fully ready to take on the task. :)

Resources wouldn't be hard to create at all. I would basically just take the 2D images I've created, and transform them into 3D.

Easier said than done. Especially in light of...

I want it to be completely immersive, and very detailed.

Just the simple fact that you think making the 3D assets is a simple matter of taking the 2D assets and "transform[ing] them into 3D" makes me cringe just a little bit. After all, the assets in the Rochard game you linked to are so very much [i]not[/i] made of simple cubes mapped with tiled textures. If you are aiming for something more like a 2.5D-ish Minecraft look, though, then you should be fine doing your graphics that way; but then that sort of eliminates the "very detailed" aspect of things, since low-def cubes tiled with textures is almost the anti-thesis of "very detailed".

Just the simple fact that you think making the 3D assets is a simple matter of taking the 2D assets and "transform[ing] them into 3D" makes me cringe just a little bit

I have done 3D modelling and texturing before, so I know that transforming the 2D into 3D won't be the easiest, but since I know how to mark seams and unwrap with Blender, it shouldn't be too terribly difficult. When I said "transform them into 3D", I realized fully what I would have to do in order to achieve such a thing. Make the "tiles" in Blender consisting of cubes, mark seams, unwrap, save uv map out, load up GIMP, redesign the textures, export, load back in Blender, apply and save out. All in all, probably a week's worth of work, depending on my texturing skills. :P

...the assets in the Rochard game you linked to are so very much not made of simple cubes mapped with tiled textures.

It will be a cube-based tile engine, and because of that, using normal and bump maps will help to add in some more details on top of the original textures. Things like trees and fences may be a bit tricky, but I'm more than willing to take on the challenge. :)

Basically, what I had in mind is sort of a 2.5D style of "cube tilesets", which I can use to create levels much in the same way as using 2D tilesets or spritesheets in GameMaker. Performance-wise, I'll have to figure out some way to merge the objects into 1 for drawing, but that's not a process that scares me. What scares me the most is the art, and making it look good. :P

Another thing I can do is design the bigger in chunks of the level in Blender, and Unity will treat it all as one object. So things like hills, backdrop landscapes and maybe even foreground objects can all be modeled inside of Blender and be treated as one object later on, rather than have a bunch of separate objects.

I have decided to finish Cuboidz in GameMaker. I realized that some of my ideas can still work fairly well, I'll just have to forget the changing-view idea. That can be saved for another game. :)

Sometimes finishing a project you are a fair way into can be better for future projects :)

Indeed, that's what I thought as well! I would love to make a 3D version of Cuboidz someday, but for now 2D will work just fine. :)

FleBlanc had some good points which also helped in my decision, despite my original stubborness :P