Blender Breakdown – Recreating Into the Spiderverse’s Camera
CodyWinch writes:
Hello BlenderNation!
In this video I breakdown how I recreated the unique frame rate mixing camera animation from Spiderman Into the Spiderverse in Blender 2.80.
Mixing a character animated on 2s with a camera animated on 1s presents an interesting issue. The character looks jittery due to the mismatching frame rates. In the video I show how I went about finding a solution and in the end provide a free addon to automate the process.
Awesome and inspiring, as always with your videos !
Thanks a lot to share your findings on the spiderverse, to me it’s a milestone in the way we think about 3D animation , there is a lot to learn and experiment from that movie !
What a cool video man!!! Still need to watch this film :)
Thanks for a great video, Cody!
Just a comment/question… with the jittering Suzanne, could it be that they, in Spiderman, rendered him separately with a shadowcatcher and then overlayed the animation of him over a render of the environment?
I can’t know for sure how they did it, but there are a few reasons why I don’t think they did that and why I wouldn’t go that route.
1. It’s not only the shadow he is casting that changes during the hold frame. The light being cast on him changes. If he was perfectly static during the hold frame the lighting would not change on him.
2. If you were to render a separate shadowcatcher/reflection pass and then comp it in you would have a mismatch in the layers. During the hold frames the environment render is still moving and your shadow/reflection pass would not match the geometry of those frames. While it might not be much if you have a large camera move it would be pretty noticable imo.
3. Adding more render passes just adds complexity that is unneeded. Just imagine instead of 1 character moving around. You have multiple characters at different frame rates and different hold frames. You would need a shadow/reflection pass for each character/object you have on hold frames. With the addon its much simpler to just bake an empty for each thing that has its own unique hold frames.
I’m inclined to think this was done during compositing. A normal pass will allow for a relight of any element rendered with a normal channel. Frame rates can easily be manipulated in comp.
3D camera data can be imported into a 3D compositing environment and rendered elements can be constrained to the imported camera (what you script is doing-brilliant btw). The shadow passes can be treated in the same way.
I’m not saying this is how they accomplished this, I’m sure they used different techniques depending on what the shot required. Looking at it from a compositor’s point of view, I would lean towards compositing techniques just because it would allow for more flexibility.
Great work and I’m glad I found this video, I’m working on a show open that will incorporate some of these techniques. I’m a pro Maya/Nuke artist but I’ve been pecking around in Blender since 2.8 was released. Look forward to viewing more of your videos.