Live data from Hacker News

Ask HN: What are you working on?

news.ycombinator.com

441–450 of 560 posts

Re: Ask HN: What are you working on?

#441

Earlier quoted context omitted.

I've had this thought. I'm intrigued! Strava has a "Strava Local" but it appears to be automated and is, frankly, bad. Would love to see what you come up with!

I think Strava Local is curated, but I agree with the 'frankly, bad' part of your comment. For my home city of London it has a few of the major landmark routes, but doesn't have anything near my house, nor does it have any of the less obvious but more fun local routes. My frustration with Strava is that they have SO much good data, yet they do so little with it.

> My frustration with Strava is that they have SO much good data, yet they do so little with it.

Whereas my frustration is that they have mountains of bad data (and we're sometimes talking egregiously bad here - segment times for runs that are >10,000mph etc.) which they do nothing about. It's a trivial data hygiene problem to note that, e.g., if an activity is marked as "running" and the average speed is 20mph+ over an hour, that's bullshit.

Re: Ask HN: What are you working on?

#442
post #217

Earlier quoted context omitted.

Thanks for letting me know! The homepage doesn't have a link to the pricing page so I didn't think anyone would find that. I'll let you know when it's ready!

I installed the Chrome plugin, and then it prompted me to create an account... so I did. That took me to your dashboard. I loved the little animation with the critter coming to inspect the DOM element, but as of right now the plugin doesn't seem to actually do anything. It doesn't appear to emit any css nor is my css being transmitted to the site. It could well be that you're between iterations, but right now if I di…

Do you mind me asking which site you tried it on?

I haven't had it crash in a while where it produces nothing while snipping. Maybe you found a new way to crash it. What if you try it on this site? (HackerNews as a test)

Also your Chrome version and OS might help me debug it.

Re: Ask HN: What are you working on?

#443

As a side project, I'm working on a route planning app for running. I'm a long distance runner and travel for work a lot, so I frequently find myself planning a run on one of the many existing apps at 5am in a new city whilst jetlagged. My bugbear is that all the existing apps suck. I want a way to say "give me a running route that lasts approximately 10km and is as scenic (greenery, water, mountains, etc) as possibl…

Try imaginerun.com. Imagine run wants to inspire people to start walking or running. The free app creates (three) possible routes via gps based on the distance and pace you set. All routes start and end at your current location, no matter where you are. The app is navigating you, so you can’t get lost at a place you are not familiar with.

> The free app

Although it appears you have to create an ImagineRun account to use it which means this goes straight into the "nope" bucket for me.

Re: Ask HN: What are you working on?

#444

I've been working on a web app to help political campaigns scale and manage an exploding volunteer network. Helping the campaign delegate tasks, helping the volunteers connecting to constituents, and see statistics along the way. I've seen too many campaigns attempt to run hundreds of volunteers, do crazy tricks with Google Sheets (just to have volunteers accidentally delete critical information), and I've been a vol…

Really cool, I've also been kicking around ideas in this area - anywhere I can see something and what type of campaigns are you targeting (local/state/other)? I know it's not the same, but have you seen https://www.victoryguide.us/?

Re: Ask HN: What are you working on?

#445
Currently I'm working on documentation, trying to explain (and sort of figure out for myself) why I spent the last 2 years developing a declarative scripting framework called 'freckles' ( https://freckles.io ).

The code is at a state now that I'm happy enough with (at least to prove the concept), but I realized that it is not as obvious to other people why a thing like it should exist than it has been to me.

I was working as programmer as well as a sys-admin-type years before DevOps became a thing, and I always thought there was something missing. We are automating a lot of stuff, but not up to the 'last mile', there is always some manual requirement install or preparation necessary. And I always thought that once we can wrap every single one of the tasks 'around' a software project into a re-usable, formalized structure, it would make a lot of things easier, and open up lots of possibilities that are hidden from our views for now.

Basically, I think we should be able to create any desired state in a specific computational environment (physical machine, VM, container, filesystem, infrastructure, ...) using a single instruction, and one (optional) dictionary-type structure for arguments. Be it trivial (like make sure a folder exists) or fairly involved (setup a complete Wordpress instance incl. reverse proxy, Letsencrypt cert, firewall, ...).

There are technologies that help with some of the issues I have (Docker, K8s, Terraform, config-management ...), but it feels like they all have a slightly off-target focus and priority for what I have in mind.

I'm still struggling to find the right words (as you can probably see) to explain the what and why, but it's an interesting experience working on basically reverse-engineering something I think is important but can't express (it's like trying to explain a color). Gaining a lot of understanding about the problem too. And the funny thing is, I don't think I could have done all that before I went ahead and actually implemented the/my solution to the problem itself.

Re: Ask HN: What are you working on?

#446
post #242

Combining open-source photogrammetry software into a system that scales reasonably to >100k images (with accelerometer/gyroscope data from the video camera) and >1G vertices in a single, consistent, high-quality mesh. It later computes a texture map based on the images, not smoothing them, but just minimizing seams and using some seam carving and color interpolation to make most seams invisible. I want VR reconstruct…

I was wondering why you choose to go with video data rather than photos. From what I know, for a given feature point, you would rather have pictures with really different point of views. It seems to me video would generate a lot of redundant data, massively increasing processing times for very little gain. What's your experience on this?

Just turn up the light and wave the camera faster, or, if you can't, maybe reduce fps if you need to conserve space at the time. Using only a covering subset of images with consistent redundancy for depth map reconstruction is easier after initial Structure-from-Motion than finding ways to recover from cases of under sampled geometry.

And with video data the initial SfM isn't nearly as expensive as with unstructured picture collections, and it's even cheaper if you get accelerometer and gyroscope data from the camera to initialize the motion recovery part and help feature matching to skip known-not-overlapping views.

You actually only use ~10 to ~40 degree view angles for depth map reconstruction, because of occlusion and texture distortion degrading results at high angles.

Re: Ask HN: What are you working on?

#447

5 years ago, I had a crazy idea that there must be an open source engine that had a really simple learning curve and worked out of the box. The result: https://github.com/typesense/typesense It took me over 3 years to get the nuts and bolts right. It's now production ready and being used in quite a few production environments. Of course, there is more work to do and so it's still actively being worked upon.

Neat, this looks really useful. Was on the lookout for a search engine for my auto-generated documentation pages, this looks like it could fit the bill.

Re: Ask HN: What are you working on?

#448
post #439

Earlier quoted context omitted.

They probably are using drone/car/vehicle to capture the imaginery - at least that's the only reason i can think of - i used to do that back at the university. We recorded a video from a drone, split it into frames, and built the mesh using those frames. When drone is competently flown it creates a nice stripe of overlapping frames - and overlapping stripes(i don't know proper technical english translation for that t…

I'm aiming for software to digest the output of a 300fps tethered camera on a drone. Basically an Apertus Axiom beta with a fancier FPGA to use all 64 lanes of the sensor to not have only 150fps, and being able to attach a 40G QSFP+ transceiver to it for data offloading. Existing software makes hand-picked pictures worthwhile, but I want full-auto behavior. Near-term I want a construction worker to open the case with…

Ok, so it's mostly a hardware issue then, a camera in burst mode continuously shooting at a lower FPS should theoretically work just as well.

But indeed, if that works with a video, that's might be easier to use.

You're talking about imu data, do you do geo-referencing also? It might be difficult having a GNSS fix if your target is inside buildings.

Re: Ask HN: What are you working on?

#449
post #442

Earlier quoted context omitted.

I installed the Chrome plugin, and then it prompted me to create an account... so I did. That took me to your dashboard. I loved the little animation with the critter coming to inspect the DOM element, but as of right now the plugin doesn't seem to actually do anything. It doesn't appear to emit any css nor is my css being transmitted to the site. It could well be that you're between iterations, but right now if I di…

Do you mind me asking which site you tried it on? I haven't had it crash in a while where it produces nothing while snipping. Maybe you found a new way to crash it. What if you try it on this site? (HackerNews as a test) Also your Chrome version and OS might help me debug it.

I ran it against a dev site I have up at localhost:3000... it does seem to work great on HN's homepage, so perhaps the page has to be publically addressable?

Running Chrome 74 on Ubuntu

Can this do media queries? eg if I have 5 responsive ranges, do I have to snip 5 times or is it smart enough to grab all of it?

Post reply on HN