Featured Animator: Latent Space - Developing a Workflow For 2D AI Animation With a Specific Art Style and Character Design

Characters From an Animated Short, featured in a
two part behind the scenes video by Latent Space.

Recently, I've been exploring how to create a 2D animation workflow that allows me to animate with my own art and characters, while incorporating generative AI into the process to do the 'heavy lifting' (basically all the traditionally time consuming stuff like the animation itself).

If you've seen my previous post, Reallusion's Cartoon Animator Versus PixVerse Image to Video and Lip Sync AI - Battle of the Talking Head Avatars, you may have noticed I used green screen (a.k.a. Chroma keying) with my AI output.

This was something I did because, if you want to animate a particularly long scene, especially with lip sync, you need to string a number of generations together. This is easier to do with characters if they don't have any background in the generation - which may not stay as consistent as the character across generations.

To cut a long story short, YouTube, today, randomly served up a video, Digital Diary : Reviving Animation with AI, Transforming 3D into 2D Cartoon Art, to me by animation channel Latent Space, who is primarily a 3D animator and digital artist but recently started experimenting with 2D AI animation.

That particular video was more about using AI to turn 3D characters into a specific 2D art style and is well worth watching for context but it was Part 2, embeded below, that really caught my attention.

Not only will you see a really professionally created cartoon, right at the start, you'll also see the AI incorporated workflow that I was leaning towards, but Latent Space has actually put into practice, as a proof of concept. Watch the video then read my notes below.

Digital Diary. Part 2: Reviving Animation with AI, Transforming 3D into 2D Cartoon Art.

Notable Highlights

• The animated short at the start doesn't stand out as AI animated. If you have a really discerning eye there are some small giveaways but otherwise it could pass for a completely human animated short.
  
  
• As noted in the video, animating all the characters separately allows for more control and flexibility when animating the scene as a whole. It's easy to swap out one character's performance without losing a great performance from your other characters. It also gives you a library of reusable performances of each character for future projects.
  
  
• Having all the separate components is also beneficial for adding further effects in software like After Effects or your Video Editor.
  
  
• The lip sync is done by trying to time the audio with the mouth movements of the clips rather than using actual lip syncing. The technique works fairly well as a interim technique until AI finally learns to lip sync simplistic 2D characters more accurately. Currently AI often won't even recognise a stylised human character and will refuse to even try lip sync it.

If you're looking for a 2D animation workflow that still gives you a lot of control Latent Space's two videos are a very good starting point and may well be the future of 2D animation where the vision, art style, and character design needs to be very specific.

o---o--- ---o--- o---

Did you find this article useful?
Subscribe to my newsletter and get the
latest articles delivered to your inbox.