- This topic has 0 replies, 1 voice, and was last updated 1 month, 3 weeks ago by
.
-
Posts
-
Hi everyone,
About a week ago, I started digging a bit deeper into “vibe coding” out of curiosity after watching a video about it. The idea of building working software relatively quickly with the help of AI sounded interesting enough that I wanted to try it myself. So I ended up getting a Pro subscription for Anthropic’s Claude, among other things, and gave Claude Code a spin.
For a test project, I picked something I’d been meaning to do for a while anyway: a GUI for printer profiling that brings together several ArgyllCMS tools like targen, printtarg, chartread, colprof, and profcheck into a single interface and structures the whole process.
The result is a macOS app called “ChromIQ” (name courtesy of ChatGPT). I also generated some of the graphical assets using AI, specifically Google’s Nano Banana and Adobe Firefly.
Quick disclaimer: this all came together in a pretty short time, so it’s not fully polished yet. I’ve effectively spent about a week on it.
Originally, the plan was to build a cross-platform app for macOS, Windows, and Linux. I started out that way, but ran into issues fairly quickly, wasn’t happy with how the app behaved, and couldn’t properly test on Windows due to lack of hardware. So I scrapped that approach and started over, this time focusing specifically on macOS. I’m much happier with the current result.
How it works:
The app guides you step by step through the profiling process:
Create test chart
You can generate automatically optimized charts tailored to paper size and measurement device, so as many patches as possible fit efficiently on a sheet. Alternatively, there’s a manual mode where you can set all parameters yourself. A preview is shown right away.Print
You can print the chart directly from the app. I still need to double-check whether color management is reliably disabled here, and improve the overall rendering a bit.Measure
The measurement process is visually guided. It shows which row to scan and advances automatically. This works pretty well so far.Build profile
In the simplest case, you just click “Build Profile.” If you know what you’re doing, you can tweak additional parameters.Check quality and refine if needed
Profcheck runs in the background. After that, you get an evaluation based on DeltaE values.
If the result isn’t good enough, the app offers a guided refinement process: it identifies problematic patches (e.g., DeltaE > 2, threshold adjustable), highlights the corresponding measurement strips, and takes you back to the measurement step. There, the relevant areas are rescanned automatically.
So far I’ve only tested this with smaller targets; larger ones might still need some optimization.You can see a video of me showcasing the measurement and refinement process right here:
At this point, the entire workflow runs without touching the command line, which was basically the main goal.
As for the role of AI:
A large part of the structure, many code sections, and even debugging steps were developed in close collaboration with Claude and ChatGPT. I worked iteratively—defining requirements, generating code, testing, fixing, and feeding it back in. Without these tools, I definitely wouldn’t have pulled this off in that timeframe. At the same time, you do notice that results need to be checked carefully, especially with things like the print pipeline or color management, where small errors can have big consequences.I also got a lot of helpful feedback and support from GitHub user Soul-Traveller (aka Knut), the author of the Argyll Printer Profiler Script, especially when it came to tracking down bugs. Big thanks for that!
If anyone feels like taking a look or trying it out:
https://github.com/itsab1989/ChromIQ
Feedback is very welcome. So if you happen to have a ColorMunki lying around, aren’t put off by all the technical details, and feel like giving it a try, I’d really appreciate it 😊
Best regards,
Sebastian
