arglborps wrote: ↑Mon Dec 28, 2020 4:20 pm
How do you deal with a case where you have two characters interacting in a scene, e.g. one walks towards the camera in a scene, and you want to render out both characters separately so you can do the compositing and lighting in AE (or in my case Apple Motion)?
It depends on the level of interaction in the scene. For example, I believe the Thighsander and Norge shots from King Julien (on my 2017 demo reel,) were rendered as a single Layer Comp. Technically, I
could have rendered them separately , but it wasn't necessary because the lighting effects are pretty basic here. Plus, because the characters are holding hands, it was just easier deal with this way.
When I need explicitly separate the lighting and shadow elements for each character, then I'll create separate Layer Comps for each character. I'll do this so I can get a clean alpha of each character to work with. When I need to be especially particular with a lighting effect, I'll break out passes for some of the body parts. For example, on my 2019 reel, if you look at Po's left hand when he's up in the sky, you can see a subtel rim effect on the inside of his hand to separate it from the arm and body. To achieve that, I set an extra Layer Comp for just his hand (paw?) and used its alpha channel to create the 'inside light' effect.
As you might have gathered by now, the Layer Comps window and Moho Exporter are some of my best friends.
Normally I just try to render out every character at the largest needed size and place them in 3D space in Motion and do my compositing – resolution is sufficient in most cases, but if, say, one of them walks towards the camera from full body shot to ultra closeup – how do you handle this? Render the characters in 4k so you can zoom in on them in AE? But 4k would take forever to render, right?
I typically do the 3D positioning and movements inside Moho, especially if the characters are interacting with each other or with animated props. This can be done by moving the groups and/or camera in z-space. This usually doesn't include camera movements unless I really need to see a parallax effect that I feel is easy enough to do in Moho.
Most of the time, I do my camera motions in After Effects since I'm doing the final compositing there anyway. This way, I maintain more interactivity and have more options at the end of the process, and probably will not need to go back to Moho or other animation programs I may have used to create my layers.
That said, I'll probably still set up non-rendering (i.e., set to 'Do Not Render' in Group Settings) proxy version of the environment in Moho for visual reference. This way, I can accurately place the characters for my Layer Comps. By setting it to 'Do Not Render', I can prevent accidentally rendering it even when it's visible in a Layer Comp.
But if I'm animating a really
complicated camera motion, I might assemble the environment and char layers in a
dedicated 3D animation program (in my case, LightWave or Maya) and animate the camera there. For advanced camera choreography, it's just faster and easier to animate the camera in a 3D program because there are better camera rigging options available. Then, I'll render out my 3D passes for my compositing program.
(In short, regardless of where creating my animations,
everything eventually ends up in compositing. At my workplace, this is usually in After Effects, and sometimes Nuke; at my home studio, this is either in After Effects or Fusion.)
As for rendering 4k in Moho, render time is not usually a problem for me because when I'm using Layer Comps, the background pixels are mostly empty (transparent.) Of course, that's not likely for background art renders, but since I'm compositing in another program anyway, I just never bother rendering 4k backgrounds from Moho. It's not necessary, and it's more efficient to simply assemble the 4k BG images in your compositing program and render from there with your camera's final framing and resolution.
Come to think of it, I had to do exactly that a couple of weeks ago at work: I rendered 4 characters from Moho in a single Layer Comp called 'Char' at 4k. For this Layer Comp, I did not animate the camera because I was going to do that in Ae. 'Char' rendered quickly because most of the 'background' was transparent pixels. I didn't bother rendering from Moho 1.) that would take a hideously long time, and 2.) a single frame still image in Ae would serve the same purpose.
(The exception is when I've created the background art within Moho, in which case, I'm rendering a single frame from Moho, not an animation, so it's probably quick render too. The point is, when you're compositing, you can optimize your render times from Moho by rendering fewer and simpler element. This way, if you need to change or fix an animation, you only need to render the affected Layer Comps and frame range, not the entire thing again. BTW, this is one of
many reasons why I never render a movie file from Moho. It's more efficient to render frames from any animation program (2D or 3D,) and leave the movie file output for the compositing program.)
Finally, after assembling my 'char' layer and background elements in Ae, I animated my camera in 3D. This final render will go quickly Ae because I'm not rendering a 4k image, I'm rendering only what the camera sees at either HD (720p) or FHD (1080p) res.
Hope this helps and that I wasn't too 'rambly'. (I'm known for that.)