A new algorithm for smoother 360 video viewing | Hacker News Reader

A new algorithm for smoother 360 video viewing

code.facebook.com

119 pointsjamesgpearce16 comments

Off topic — I love to read these posts, they remind me how crappy my time was in university where I got to learn a ton of equations with no apparent use case, then I read these articles about the awesome things you could do with math and realize that I didn't take the time to appreciate my education. I would have loved to have an internship in one of these companies to put in practice what I was learning during that time, and not waste my education — specially the calculus classes — making presence in a bank writing mostly irrelevant code with no more requirement than basic math skills.

Disclaimer: I understand the parent and I are hijacking the thread, but...whatevs.

This. x1000. I wish there were more places I could put my four goddamn years of CS knowledge to use. I joke with business people in our team how they could steal my job after maybe 2 weeks of programming bootcamp.

TBH, any 9th grader with reasonable determination can do what I do on everyday basis, while the advanced/cool stuff that I really enjoyed/struggled with in college slowly fades from my brain.

I wish there were a way to somehow reconcile these. I want to make money, yes, but I want to do cool stuff too. I'd be willing to take a not insignificant paycut to work on cool stuff. Unfortunately, it appears that places with more 'interesting' stuff to work on are also those that end up paying more. Just like how colleges that are more expensive to get into are also those that will [more likely] have more money to throw your way.

What gives, people of HN?

Personally, I think it is just a lack of imagination coupled with the true fear of imagination that ultimately leads to technical debt.

For instance, let's say we could automate the process / approval system for management sign-offs for a software development using a blockchain scheme.

Great! You or I could come up with something using open source software and probably have a fun couple of days. In reality, you'd have to train a whole bunch of people and someone will want to integrate with the old system, someone will want to be able to just print and sign and send through the interoffice mail, and then good luck finding affordable technical help because the recruiters won't know how to hire for it, and someone who hasn't done it will want a premium.

Anyway, probably a bad example but creativity is rather punished a lot of time.

> Unfortunately, it appears that places with more 'interesting' stuff to work on are also those that end up paying more.

I don't understand. How is this unfortunate?

unfortunate because even if one couldn't get into a high paying job, one could try getting into an interesting one. Since they're the same, it's both or nothing. Thus it's fortunate for those that get those, but unfortunate for the rest. : )

Academia provides fantastic opportunities for paycut and interesting work. I did that for way longer than I should have. Seriously look for programmer positions in research institutes. Want to write code that studies Protein folding ? Genomic research based on deep learning ? Simulations for physics or biological experiments? done done done.

This is why side projects exist.

Not just home side projects, either. Depending on what you can squeeze in, you should take opportunities to hack together little side things at work. You should do this for your own sanity.

Many times at work the awesome things come from somebody's (secret) side project, which turns out to benefit everybody.

If you do this, you will gradually fill up your resume with cool things that were a result of your own initiative.

Also: side projects are a source of startups.

I recommend doing a few things:

1) Read papers about this stuff, attend conferences about this stuff, and try to implement some of it yourself. For this particular set of techniques, SIGGRAPH is a good conference to attend and their proceedings are full of good papers to read.

2) Make a project that uses some of these techniques on your own. It can be open source, or it can be an app you sell on your favorite app store, or whatever. It doesn't really matter.

3) Look for a new job that uses some of this stuff. You don't have to jump right into working at Facebook or Google. There are smaller companies out there that do this type of stuff.

My story is that I was working for a company that sold large database services to large retailers. It paid pretty well, but was boring as hell. After working for a few years and saving up, I looked for a job doing the stuff I was interested in. I found one. It didn't work out and I quit after 4 months. But the next one! The next one worked out! It was a bunch of image processing stuff. They were a small company and needed someone with some graphics experience. I had never shipped anything graphics related, but I had written a bunch of graphics tools for fun for myself to learn how it all worked. It was enough to get the job. My pay probably stayed the same for 2 years as I went from job to job, but my happiness increased. After 3 or 4 years at the company, I left and started my own company doing graphics work. I did that for about 5 years and eventually decided to go work for one of the big companies that's doing cool stuff with this.

You can do this. It just takes time and effort.

I thought I was unable to learn math, until I learned it outside of a school. One book like Prime Obsession was worth more to me than every moment of high school math. It's unbelievable to me that anyone can learn something devoid of context. I think there's a similar problem in how second languages are taught in schools as well.

Same here. Now I do independent research (on primes) and am learning higher level mathematics myself by reading older edition textbooks. Mathematics is the best toolset to reason about reality. It's a shame that most people do not know or care to partake in it. It's the purest form of intrinsic truth there is in this universe.

I think it's also just useful to learn for the way it makes you stop and examine the world around you, your own decisions, in a new and more analytical light. Besides, if you're into physics, it's the only way; you can't visualize a 10-sphere.

This is pretty incredible! No more need to buy gyro-stabilizers for these cameras (if you are posting to Facebook). I've found the Facebook 360 viewer is more intuitive than other viewers out there.

There are a few different things going on with a shaking camera that makes the video bad: A: Movement between frames B: Slow intra-frame movement C: Fast intra-frame movement

As I'm reading this announcement they have a nice solution for problem A - moving camera. Problem B and C results in jello-effects (assuming rolling shutter) and blurriness, respectively. There are research solutions for all of them (see e.g. "Dolly-cam" from Per-Erik Forssén et al at Linköping University). But for practical purposes you are still much better off if your camera is not shaking all over :)

Look at drones and compact cameras for examples: gimbals and optical IS beats digital stabilization every time.

I have a question, hope someone can shed some light: given that 360 video is recorded, how to make it to display in my webpage? What are the most popular open-source 360 video player? And if I want to make my own video player, what should I start reading at?

To your first question, I think the easiest is to upload to FB, or Theta360 (the makers of the popular cameras, e.g. https://theta360.com/s/dAKje30mhbrEIpEcWC3Hs4YzY) or youtube and then use each site's "embed" feature

Are there no self-hosted js players that could use your own 360 equirectangular video files?

Reply on news.ycombinator.com