F*ck paying for tools, I'll build my own

The past year or so I have lost the joy of illustration. For most of my illustrations, I used to use Adobe Illustrator. It’s a great tool, but as a hobbyist, the cost of it started to outweigh the benefits.

The single-app plan runs close to 40 euros a month if you want the freedom to cancel whenever. That’s close to 500 euros. Pay a year up front, and you’ll still spend over 300 euros per year.

01 Other tools are kinda shit

Maybe an exaggeration. Inkscape is actually kinda good, and it’s free. I use it a lot for my pen plotter, Axidraw Axidraw is a writing and drawing machine. I’ve got the V3/A3 and it is amazing. Source has an amazing plugin for it. But Inkscape has some very odd choices baked into the UI and always messes up my way of working. Most of my illustration work is built on repeating patterns, and I’ve built a workaround in Adobe Illustrator for it. That didn’t really translate well to Inkscape and those patterns slowed the application down to nearly unusable.

I’ve also tried Affinity. It’s not bad at all. It’s currently free and has several useful features. But I fear it won’t be free forever and then I’d be stuck with another workflow that I eventually have to unlearn. Also the pattern tool in Affinity didn’t work. It just didn’t work at all.

I realized using shitty tools took the fun out of illustration for me. I spend most of my time figuring out how to do the same thing that I already knew how to.

02 How hard can it be?

Building Quiet Tarot earlier this year changed how I look at design and design tools. With Claude and other code assistants, I can not only build a solution, but quickly create a tool to solve my specific problem.

For Quiet Tarot, I built several tools to make my life easier. One to manage copy Quiet Tarot has over 55k words of copy across its content at scale, and one to track my product testing. These tools were simple JSON editors, so building my own SVG / vector tools was a bit of a stretch.

So I went in with a pretty yolo mindset: just start and see where it takes me. Claude advised against it, for it being highly complex and unrealistic to build. I’ve made a quick UI sketch in TLDraw TLDraw is one of my favorite design tools. It’s simple, fast and perfect for throwaway designs. Source and wrote a quick PRD. I based the requirements on specifically the tools I use in my workflow and descoped all the rest.

03 Done and feature-complete in one week

I broke the PRD up into 25 phases, each building on top of the previous phase. Each phase came with its own design brief and scope. I’ve learned that scoping in Claude is amazing. Tight design briefs for each phase allowed me to build each step quite quickly as it built upon each previous step.

The first 7 phases were really just getting a working MVP: canvas and pan/zoom, the pencil tool, pattern fills with global registration so hatching lines up across separate shapes, undo/redo, and save-straight-to-disk.

Everything after that was refinement. A chunk of phases went into the basics: export, layers, guides and snapping, selection tools, arrange tools, etc. Another chunk went into improving stability and making life easier. Most of the initial phases went quite fast. I did a minor iteration in TLDraw to cobble some screenshots together to sketch what I wanted.

The original scope expanded from 16 to 25 phases, as a few features turned out to be quite useful in my workflow (and it was quite easy to build). Multi-artboard support, templates, type support, even a Live Trace tool to auto-vectorize a scanned sketch into editable paths. I also decided to wrap the whole thing into a standalone desktop app, so I don’t need a browser tab open just to draw.

I built it almost entirely in Claude Code in the evening hours over the course of a week using mostly Sonnet with a bit of Opus mixed in, on a Pro subscription. The biggest bottleneck was working around the daily limit.

04 Pattern control and exporting videos turn out to be sick

The app, which I’ve named Omen, is actually really good. In particular, the pattern tool is exactly what I wanted, and I haven’t seen any other app that comes even close. The pattern tool is fast and high-performance. Where Inkscape and Affinity choked on patterns, Omen handles them with ease. Additionally, you can control the pattern set to make sure the cross hatches always match as you adjust size, roughness and randomness.

Beyond patterns, it’s pretty much everything I use when drawing: pen and pencil tools, redraw-edge, line and star tools, a vector eraser, pathfinder add and subtract, offset path, Shape Builder for combining shapes by dragging across them.

The list continues with multi-artboard documents with templates for new files, type support, mirror and anchor tools, proportional stroke scaling, vector brushes and art brushes, even a warp tool. It’s pretty feature complete.

One thing that wasn’t even part of the original plan: because every action gets logged into a persistent history file, the app can replay an entire illustration from blank canvas to finished piece and export that as a compressed video. I built it mostly to share process on social media, but there’s something satisfying about watching the playback.

05 Why you should be building your own tools

I have built the perfect tool, for me. It has everything I need. And even though a few things could be better, The pencil tool is not that great. That took a long time to get sorta right. Particularly the “feel” of the pencil isn’t as good as Illustrator, Affinity, or Inkscape. It is usable, but lacks refinement. Type and the Live Trace tool is also kinda bare bones. It works, but it doesn’t have the same level of control as Illustrator’s Live Trace. Omen is still better and tailored perfectly to what I need. The cool thing about building your own tool is that you can always improve or change it later. If I get annoyed by something in the future or if it needs to be different, I can just add or change it.

So where from here? I thought about open sourcing Omen. I’m not sure I’d be able to maintain it properly, as my life is busy as it is. I might bring Omen into a more generic shape so that I can open it to the public later. I don’t want to change Omen to a general purpose tool though, because adding more irrelevant features for me, bogs down my own workflow again.

To be continued, below is a sneak peek of my next project.