Sprite-only software rendering: existing API or own?

Newbie & Debugging Questions
#1

Hello!

I’m working on a rather simple 2D JRPG, but although I’m quite into making pixelart, I’m definitely not looking forward to doing each and every wearable Item in some 20 different poses, 8 different directions (coding a system to somehow arrange the layers according to the current direction), for at least 2 genders und about 4-5 possible body sizes…

So I’m thinking about generating character sprites (and maybe some flora and fauna) from 3D models so I just have to modify some meshes at runtime to achieve those variations.

I don’t want to reinvent the wheel, but also I don’t want to include some 5MB foreign API to just use a small amount of it’s capabilities. I’m a total Newbie when it comes to 3D graphics, so right now I’m totally clueless on what to do: write my own renderer or use an existing one?

Another problem is, is any of the 3D APIs even capable of doing what I need?
Namely:

  • Use software rendering (no prerequisites for users)
  • Return the computed image not in a render panel but as… well, image :stuck_out_tongue:
  • Disable anti-alias and choose between different texture resampling modes (to make it a bit pixelated and thus better fit in with the rest of the game)
  • Add a non anti-aliased outline (same reason)
  • Work with independent body parts without using an overkill scenegraph

I’ld be really grateful if anybody could help me figure out how to best achieve what I’m looking for :slight_smile:

#2

Most games of this variety pre-render. They start from the 3D art but they render it to your 8-direction sprites and then just use those in game.

This emns you cna use all sorts of fancy 3D animation tools that are available at art-time to actually animate the characters, automate capturing the sprite angles, etc.

So the “API” I’d recommend is Maya or 3D Max 8)

#3

Or Blender, which doesn’t cost an arm and a leg. Just has a slightly higher learning curve.

#4

Thanks for the replies!

Pre-rendering sounds at least better than doing it all by hand, but still it’s to limiting for what I had in mind. Adding new animations (and I’m sure over time I’ll add lots of them) would be rather messy, and to be true: 8 directions sounded okay when I still wanted to do everything by hand, but the possibility of infinitely various directions sounds even better… :wink:

(Not to mention that I’m mainly writing this game to polish my programming skills, so a complex coded solution is always preferable to simple non java solutions ;D)

Still, since I already have to lern Blender for a 3D solution, if all fails I can still use pre-rendering, so thanks again for the suggestion!

#5

Then take the whole leap and do your game really in 3D.

In which case you have a number of API options with trade-offs. See my FAQ for my take on most of them.