florians

“> Claude: Semble surfaced things grep missed — here are the additions to the earlier answer.”

Nice!

What I want are precise and tight bounding boxes. Why is this so difficult?

Same

Curious about background removal with BiRefNet. Would you consider it the best model currently available? What other options exist that are popular but not as good?

Love the vibe!

The problem is that most sites implement dark mode wrong with too much contrast. It doesn’t work if you make it “black”. It’s more about dimming the lights.

Reading dark pages will never be as good as reading in a well lit room. But there are ways to make it work.

This. Thanks for illustrating how little is required for a basic dark mode support.

It depends on the environment. Of course, reading HN in a dark room will never be pleasant. But looking into a bright white flashlight can really hurt. Good dark modes are not simply black/white but make use of the right font weight and color to reduce glare.

And soon even your oven will require an account with subscription so that the built-in camera can notify you when the bread is browning and it asks for permission to lower the temperature

I enjoyed reading the article and found it interesting. Had no prior interest in the topic but it was an entertaining read. Last time I used binary and calculated with it was in high school. Didn’t know about UNIX signals and now I understand how processes are terminating.

This is quite genius. Instead of wasting time with responsive design, grid and flex layout complexities, this solutions gets the job done.

That’s how I would have approached it in Flash back in the days when 800x600 was all you had to worry about.

Noteworthy: Contributions by Claude

I find it amusing, yet sad, that some here expect a podcast to exclusively be a source of information where every second delivers bite sized facts. What about entertainment? What about engaging with a topic for hours and eventually learn something that‘s not a fact, but a new perspective?

This is entertaining. I learned Turbo Pascal in high school.

What I like from watching it again: the aspect of structured programming.

It’s quite refreshing to see a language that doesn’t rely so much on brackets.

It even got away without syntax highlighting by using all uppercase REPEAT, BEGIN, END or capitalising function calls.

Thanks for sharing!

Ah, I can imagine why the illustration weren’t as clear. Especially if you don’t speak German it might be difficult to catch mistakes. I added new examples.

But as you mention: starting by observing one’s own habits and preferences and then coming up with small instructions is all it takes. Feels like piping in Unix where each process is a prompt.

Peanuts if you build a business on top of the SDK.

Sweet API design!

Especially the design principles are a collection of best practices. Part of it feels like reading Apple’s HIG or Material Design Guidelines. Part of it Refactoring UI. It’s what you learn (or wish you would learn) in design school. Or from a good mentor. Or acquire through years of experience and reading.

Curious: How certain are you that this is indeed the original system prompt, without further hallucination?

Very good points. Especially the glyph-specific features you mention (a, g, …) are definitely of my interest and the origin of the project. I think I‘m only at a point where I arrived at a basic foundation that hopefully will enable me to do the things you mention. Thanks for your insights!

I’m glad you enjoyed reading.

1. Adding previews for the weights/styles is already on my list :) 2. I initially starting the project when trying to pair fonts and stress test them with my own content. Using the tool to find and save fonts for real projects hasn’t been the focus, but the data will surely end up in something you’re looking for. 3. File size, yes, that’s something I can add. Large file size might even be a hint for very large character set?

I’ve started to implement a way to parse local fonts via drag & drop (no actual upload required). This will enable to create a personal database and eventually be a foundation for a shared, larger database. But before getting into this, I want to improve the measurements first and I’m happy about some suggestions that have been made here already!

Thanks for your detailed feedback. Your suggestion for measuring the weight more precisely is what I’d been thinking but didn’t figure out how to do it. Introducing a third color is such a simple way to do it! I’ll look into this.

I will reconsider updating the histograms to reflect the current filtered subset. I tried this in the past but ended up with cases with very little results and thus small visible areas in the histograms that often required to reset filters. But I agree that this would better reflect the filtered data.

Thank you.