cpcdoy

Multiple data layers have to exist per country, and both have to be fetchable straight from the browser, since the site is static with no backend (everything is computed in your browser).

You need both accurate heights and outlines at ground level for buildings, forest, etc. OSM has outlines nearly everywhere, but height data can be missing. Without a height, it'll be inaccurate to answer "will that roof block the Sun", which is the whole point of the app. France and Spain both publish real heights, so they went first.

You also need good terrain elevation data: with the Sun only a few degrees up, a hill can hide it completely, and your own ground height counts too. Inaccurate terrain data can flatten hilltops for example, which I've seen in many cases.

So, since I want things to be accurate: it's one integration per country, and plenty of countries don't publish heights openly.

I'll be gradually adding more and more countries. I've already started working on the 2027 eclipse (https://eclipse2027times.com/), so by then it should have even more supported countries.

Hope it's useful. For what it's worth the variation within Germany is small: about 88% coverage in Cologne and Düsseldorf, 89% around Stuttgart, down to 85% in Berlin and Hamburg. So it isn't worth driving for a couple of points. The most important thing to check for will be cloud coverage at this point. Enjoy it.

Ah I see, I think I didn't make it clear enough that for the 3D viz there's only France and Spain that are currently supported today. I'll change the phrasing to make it clearer. I'm planning on adding more countries and features before the 12th.

Thanks for the feedback!

Oh nice, you were in the path of totality, it must have been incredible to watch.

Today the eclipse viewpoint simulation already renders parks actually but only in France for now. I'm planning on adding more countries as well.

Thanks!

You should see it from Germany but it'll only be a partial eclipse. If you're talking about the eclipse of the 11th August 1999, that one had the totality crossing Germany, so the upcoming one will be less spectacular I think.

I've also updated the website to add more info about German cities, hope it helps!

Thank you for your feedback.

You can try the prompts that I share in the article with any model you'd like. Here's the link: https://docs.google.com/spreadsheets/d/1QLQwPnKM_RKglueteCdD...

Which HN tag do you think fits better in this case?

Abwab.ai | MLOps / DevOps Engineer | Riyadh, Saudi Arabia | GLOBAL REMOTE | Full-time / Part-time

We're democratizing SME financing in MENA by building the Moody's for micro-financing. Our team comes from Revolut, Nvidia, Ubisoft, Bloomberg, OakNorth, and other leading companies. We're looking for an MLOps wizard to help us scale our ML infrastructure and deploy state-of-the-art models (LLMs, VLMs, SLMs, etc) in production. Our team is spread globally across Saudi Arabia, Egypt, France and more!

Tech: Python, vLLM, GCP, K8s, Docker, Terraform, Ray

Apply here: https://revolutpeople.com/abwab/public/careers/position/seni...

Or contact me directly: chady@abwab.ai

Abwab.ai | Multiple Roles | Remote | Full-time | https://abwab.ai/teaser

Abwab.ai is democratizing financing for Small and Medium-sized Enterprises (SMEs) by disrupting the underwriting tech stack. We're reinventing credit rating for micro-financing, creating the infrastructure for AI-based lending in MENA and globally. Our team comprises tech enthusiasts from around the globe with experience from companies like Revolut, Bloomberg, Nvidia, Ubisoft, OakNorth, and more.

We're hiring for:

- Senior Machine Learning Engineer

- Product Manager

Apply at: https://abwab.ai/hiring or send me an email at: chady@abwab.ai

I've been working on optimizing inference time of a BERT-based multilingual NLP model in Rust, check it out here: https://github.com/cpcdoy/rust-sbert

For example, you're reading a technical article and the answer to a question is a complex diagram or a graph that could easily fit in one card but would take way too long to explain in pure text.

Also, another suggestion would be to enable some kind of formatting of the text in the cards, in case you're trying to remember steps to a recipe, algorithm or something else.

Really good job again, love the extension :)

Really cool extension!

A suggestion would be to enable adding pictures in the Answer part of the card