C++ – Recommended file formats and graphics libraries for importing 3D model into OpenGL/C++ project

In XNA, penalty of trying to do anything on your own (load data formats in particular) is much higher than living with imperfections of the ones supported right off the shelf, at least that's what the documentation looks like, and what i heard from someone who bended the content pipeline dearly to support what he needs (runtime geometry generation) and whom i trust to know what he is doing. So i guess .X and .FBX are pretty much gonna be it. They both support skeletal animation, and i think also vertex morping. I'm not exactly sure what features exactly Blender is able to export into them, but i can look into it if you like.

Blender 2.5 is going to finally have good Collada support (current one doesn't have armatures and is broken in multitude of ways), and Autodesk provides a good tool for Collada <-> FBX conversion.

If you look around hobby development community, you'll see that Id software data formats are quite popular. On the one hand the MD2 and MD3 vertex morphing character animation formats, MD5 skeletal animation format, whose needs are obviously covered by the above, and on the other hand the BSP file format, which describes game levels.

One thing that speaks for BSP is that a good editor is available (GTKRadiant and derivatives). This is an efficient format which can cull hidden geometry. However, the runtime cost of building custom geometry in XNA is significant, and with the limits of these formats are quite modest at 64k vertices and 64k polygons in the whole game level, which is something you can just dump onto the graphics card nowadays and no performance drop will be noticeable. So you don't gain any benefit from supporting it directly.

I doubt there are any powerful, modern, available formats and tools for culling-optimized level design, whose performance benefits would translate into XNA.

A couple of links you might want to check out, if you haven't already:

3D World Studio content pipeline

Quake port

All unmaintained and out of date, no idea how optimized or not.

It seems like all you'd need to do is write an animation loop that draws the models and updates repeatedly rather than actually program the code to draw every little thing.

Everything that happens in a computer does so because a program of some form tells it exactly what to do. The letters that this message is composed of only appear because your web-browser of choice downloaded this file via TCP/IP over an HTTP protocol, decoded its UTF-8-encoded text, interpreted that text as defined by the XML, HTML, JavaScript, and so forth standards, and then displayed the visible portion as defined by the Unicode standard for text layout and in accord with HTML et al, using the displaying and windowing abilities of your OS or window manager or whatever.

Every single part of that operation, from the downloading of the file to its display, is governed by a piece of code. Every pixel you are looking at on the screen is where it is because some code put it there.

HTML alone doesn't mean anything. You cannot just take an HTML file and blast it to the screen. Some code must interpret it. You can interpret HTML as a text file, but if you do, it loses all formatting, and you get to see all of the tags. A web browsers can interpret it as proper HTML, in which case you get to see the formatting. But in every case, the meaning of the HTML file is determined by how it is used.

The "draws the model" part of your proposed algorithm must be done by someone. If you don't write that code, then you must be using a library or some other system that will cause the model to appear. And what does that library do? How does it cause the model to appear?

A model, like an HTML web page, is meaningless by itself. Or to put it another way, your algorithm can be boiled down to this:

1. Animate the model.
2. ????
3. Profit!

You're missing a key component: how to actually interpret the model and cause it to appear on the screen. OpenGL/D3D/a software rasterizer/etc is vital for that task.