In early August 2026, a new compositing software called “Caddis” was released. This tool for motion designers works with layers, similar to Adobe After Effects. These layers can be placed on the timeline, and their duration can be adjusted. Layer modes are also available. The software offers a separate node network for each layer. This allows users to develop effects and bring content to life.
What Caddis now…
The system is highly versatile, featuring numerous nodes and their respective options. Caddis imports videos, images, image sequences, SVG graphics and audio. Other modules generate objects, text, fields, and point objects. These can be manipulated using nodes for math, fields, and other functions. Matte and tracker functions assist with compositing moving footage. And let’s not forget effects that can be applied to the content.
In an interview, Digital Production spoke with developer Mike Gaynor (LinkedIn | IMDB) about the creation and implementation of Caddis.
DP: Please tell us a little about yourself and your background.
Mike Gaynor: I’m a motion designer and 3D generalist, currently based in New York, originally from San Francisco. II’vebeen doing this for about 15 years – I started in college doing Video Copilot tutorials for fun, then landed an internship at a small studio making SaaS explainer videos and ended up dropping out to work there full time. From there I freelanced at boutique studios around SF, got my first tech job at Uber, then worked across agencies both in-house and freelance, started my own studio, and eventually landed at Shopify, where I’m currently a senior motion lead. Along the way, my work moved from straightforward 2D explainer stuff into more technically complex territory – 3D, live action, data, real-time systems. My more recent Shopify work has been focused on brand storytelling.
DP: What motivated you to develop Caddis?
Mike Gaynor: It started as an experiment, mostly to see if I could. I’ve been using tools like After Effects, Cavalry, Cinema 4D, and Houdini, and I have favourite parts of each. Can I combine those favourite parts into one shape rather than treating them as separate workflows? It turned out to be genuinely enjoyable to build and to use, and that’s what kept me going. Caddis works in layers, like After Effects, but rather than stacking effects, each layer’s contents are a node Graph- closer to Houdini or Blender’s Geometry Nodes, but built specifically for 2D motion work. It hits the right part of my brain – fast enough for straightforward projects, modular enough for experimentation and play.
DP: Imagine you’re pitching Caddis to a motion designer who already knows After Effects, Fusion, Houdini, Nuke, Blender Geometry Nodes or TouchDesigner. What’s your TL;DR?
Mike Gaynor: The short version is that Caddis combines the familiarity of a layer-based motion graphics application with the flexibility of a procedural node Graph. Every layer has its own node Graph, so you can build procedural animations, shapes, deformations, gradients, and effects while still working with a timeline where layers can be reordered, trimmed, duplicated, and animated just as you’d expect.
The idea is to make procedural workflows much more approachable. Instead of forcing you to build an entire project inside one large Graph, each layer contains only the logic relevant to that element. You can expose parameters, drive them procedurally, and connect nodes, while still keeping the overall project organised in a familiar, composition-based workflow.
A simple example is the animated “worm” from the tutorial. It starts as a line, is resampled to add more geometry, deformed with an oscillating field, coloured with a gradient, and animated using normalised time to create a seamless loop. That entire set-up remains procedural, butit’ss wrapped in a workflow that feels much closer to a traditional motion graphics application than a pure node-based system.
Since when are Artists programmers?
DP: Did you develop the project all by yourself, or do you have a team?
Mike Gaynor: The first beta version was something I made entirely by myself, and I’m still the only person developing it. But the beta community has functioned like a distributed QA and product team – bugs, gaps, and improvement ideas have come from users constantly over the past couple of months, and a lot of what shipped exists because someone asked for it. I’m open to others collaborating on it directly as well.
DP: Was AI or any form of vibe coding used in the development of Caddis?
Mike Gaynor: Yes, and I’m upfront about that. I’m not a developer by trade – I’m a motion designer, and I know what this kind of tool needs to do and how it should feel to use. AI let me act on that knowledge directly, but the process was more disciplined than I think the term vibecoding gives credit for.
I started by putting together a “northstar” document that outlined exactly what we were building and the standard it had to meet – that came out of a lot of conversation, no code yet. Once I had a clear picture of the result, I moved to building the app itself with Claude, always working in plan mode: I’d take Claude’s plan and verify it against other AI models before writing anything, to catch gaps or mistakes early. I designed the interface in Figma and used Figma MCP to hand that directly to Claude for implementation.
The future of Caddis
I kept(and am still keeping) a running roadmap, changelog, and bug list throughout. Even with AI doing a lot of the execution, there was a lot for me to learn. This was my first time building a desktop app, and things like code signing, licensing, and compliance were all new territory.
DP: What are the plans for Caddis?
Mike Gaynor: Caddis is still early, and while it’s already pretty capable, there’s a lot I want to add. Right now the focus is on refining what’s already there – UX improvements, better data handling, optimisation – because that’s the foundation for the bigger systems I’m not rushing: physics, particles, maybe 3D eventually.
I’d rather wait until I’m confident I have the right design for those systems first rather than rushing them out. The roadmap is also genuinely shaped by users. I take suggestions seriously, and a lot of what’s already in Caddis came from a Discord message or an email. I think that’s the real strength of an indie tool – it can be actively shaped into what best serves the designers using it without much overhead
