It's true that superb pixel art is harder to pull off than superb 3D models because of greater self-imposed limitations. However, it's also true that if you just want your art to be passable, it's easier to achieve that via low-res pixel art than via amateur 3D modeling. IOW, pixel art has both a lower barrier to entry and a higher skill cap.

If you want your 3D art to passable, you download a few models from online for a few hundred dollars and tweak the bone animations.

IE: What Archer does with all of its 3D models of cars it sticks into the show.

Which is exactly my point: if we're resorting to external assets, there are infinitely more spritesheets available online for the low, low price of free, because making decent sprites is such a comparatively small effort that even artists themselves attach less value to the task.

(Which still isn't to devalue the effort that it takes to make good-looking sprites; I've tried, and damn do I suck at it.)

That's not useful at all.

You need a baseline of art, and then you need to tweak the animations to match your video game. A model may not have a "double-jump" animation but you can easily add one through bone manipulations with any 3d model.

But if a sprite-sheet were missing the double-jump animation? You're either doing it yourself (hard pixel art style) or hiring someone else to do it.

But for me some pixel artists miss (for me) a large part of the artistic value of pixel art: restricting complexity of the scene -- not just portraying things in blocks. The blocks aren't important at all, in fact they're pretty ugly if you're not a great pixel artist (hence the natural wish to find better upscaling algorithms). It's often overlooked the CRTs those art were made for were quite blurry and round, alleviating the blocky look.

That's the right reason to defend using modern screens to their full capability with a certain simplified-art style.

I mentioned BlazBlue specifically because of their combination of 3D Art and 2D Art. The 2D art is as you said: very simplified.

http://i.imgur.com/mDlQYar.gif

Count the number of colors on Jin Kisaragi. The smooth, defined lines going down "strong", hand-crafted pixel art lines. With massive sprites built for the high-definition era to boot.

The 3D art gets a bit intricate and busy, but the 2D Sprites.

The question that immediately springs to my mind is "what did the 3D reference image look like". Rotoscoping (the animator's term for 'tracing reference footage') is an old, old process, almost as old as animation itself. You can do it well, by using the reference footage as a base to get the hard parts of the 3D motion down then putting that away and pushing the drawings, or you can do it poorly, by just tracing it and not really doing much to it otherwise.

I would bet that if you superimposed that BlazBlue animation over the 3D reference, there would be a huge difference - that looks like it's had a lot of human thought put into it.

(And then go look at Arc's latest game, Guilty Gear Xrd. Which is presented entirely 3D with some models that have had a LOT of love applied to them.)