Just some personal thoughts and a few tips:
hayasidist wrote:I use Moho to render out png sequence. It's a lossless codec.
This is what I always do too, and it's pretty much the industry practice for animation, 2D or 3D. The reason is 1. quality, 2. performance, 3. flexibility.
For quality, PNG is a mathematically lossless codec, meaning it's compressed but with no adverse effect on the quality of the image. This also means, if you have an embedded alpha channel, you won't get noisy edges typical with a lossy codec.
By rendering to image sequence, as opposed to a movie file, the computer can work more efficiently since it doesn't need to hold the entire movie file in memory before saving it. It also means you can use distributed rendering which can really speed things up (although that's not currently a native Moho feature; you'll need a render controller for that.)
As for flexibility, with image sequences, if you crash, you can resume the render where you left off; or if you change something in a portion of the animation or have a bad frame or two, you only need to re-render the new frames, not the whole thing. With a movie file, it's typical to re-render the whole thing because it gets messy trying to render tiny movie segments to 'patch' the animation in editing.
I also like to render to image sequences for compositing. With Moho's excellent Layer Comps system, it's much easier to work with image sequences than with a bunch of movie files, especially if you want to process or manipulate the layer comps in the compositing program (like AE or Fusion.)
For the editor, I do prefer working with movie files. Typically, I render the final composited scenes as a movie file for editing. But, as mentioned earlier, many editing programs will accept images sequences and treat them as 'clips'. I think working with a movie file in editing is more efficient, though.
When I was making the HLF animation, I used Virtual Dub to batch process image sequences to movie files. In fact, I used this system to render two versions of each clip--a low-res compressed version of the clips for editing, and a high-res version of the clips for final output. Once I had the Virtual Dub presets setup, it went pretty quickly and smoothly. The different versions of the files shared names so, when I was ready to output my final edit, I just swapped the files for the editor.
Lately, I've been using either AE's batch render or Vegas' Batch Render Script to output movies from Moho rendered frames. I think you can do the same with Premiere but I haven't used it for batch rendering myself. (Might have to look into that.)
The codec I've been using it MagicYUV, which so far has been working well for me. It's a modern lossless codec that can write to AVI or MOV. It's a commercial codec but I think it's still selling for about $14.
Last night, I noticed that Vegas now has native ProRes support (called MAGIX ProRes). I'm going to give that a shot this weekend.
At work, I typically render image sequences from Moho and, whether I'm compositing or not, I use AE to output the frames to a .mov using the free Avid DNxHD codec. (Editorial is done in another area.) Personally, I don't think DNxHD looks all that great compared to ProRes--to me, DNxHD seems to have AA problems with intense colors and contrast which I don't see with ProRes, and sometimes I need to compensate by reducing intensity and blurring edges slightly. But I should point out that neither of these two is lossless unless you use the uncompressed mode. At home, I haven't seen this problem with MagicYUV (probably because it's a lossless codec).