Budabellly

example of that! https://news.ycombinator.com/item?id=19916205

For me, it's fidelity of human facial animation that has a long way to come. There are teams doing a great job of it [1], but the labor/skill required to bring one high hyperreal facial rig to a game or movie seems insane. I think companies pursuing AR/VR applications will lead here.

ML creative tools stand to automate a lot of this imo.

[1] random example: https://snapperstech.com/

It's basically an after effects vector animation for each letter of the alphabet[1], which is rendered as SVG (JSON really) with bodymovin[2], and played back (maybe with lottie?) with key bindings.

[1] https://animography.net/products/mobilo [2] https://aescripts.com/bodymovin/

Super cool. I really appreciate gadgets like this to add a little flare to existing elements on the page.

This would be really awesome to combine with animography somehow: https://codepen.io/airnan/details/JbOrbV

Probably not an advisable strategy because there are many ways it would adversely affect certain groups, but as a thought experiment, what if a pre-requisite for future COVID stimulus checks in the U.S. (individuals or businesses) was taking a covid test or antibody test?

This discussion thread has me wondering if there's a good opportunity between the ubiquity of modern cameras (drones/satellites/phones) + computer vision to reduce the cost of those "total count" surveys you mentioned.

Not that I was prepping for an interview but a couple questions I asked myself...

* How could you use social media to establish the size of the pidgeon population in NYC?

* Design a system for tracking Bluefin Tuna populations with an unlimited satellite imagery budget.

* How many drones would it take to count the total number of wolves in Yellowstone in 1 week?

* In safari destinations in Africa like the Maasai Mara, what % of megafauna are photographed annually?

Makes you wonder if all of our facial recognition tech could find a better use in ecology.

In my opinion, the argument companies really should be hiding behind is the cost of labor for a given geo. It's got cost of living baked into it, is sufficiently handwavy, yet still makes it a little more clear that you're compensated based on market supply/demand and little else.

It is worth mentioning that it's fairly common to use FFmpeg via gstreamer in the form of plugin.

https://packages.debian.org/stretch/gstreamer1.0-libav

I don't think that Loon has all of its eggs in the balloon basket. They're interested in satellites as well:

https://www.theverge.com/2019/1/31/18200879/alphabet-project...

I was thinking less about penalization or individual application decisions, and more about aggregate or anonymized datasets for institutions and governing bodies (College Board, Common App, etc.) to inform their policies.

And yeah, you could certainly lie, but I feel that's true for a lot of fields on college apps? Perhaps it could be reasonable to have your school validate or sponsor the answer?

It's well researched that private tutoring (among many other socio-economic factors) gives you a huge advantage on these tests. Having one of these bodies come out and say by how much could be a wake up call, and maybe spur more decisions like this one in the original post. Totally agree with your points, but also think tutoring != independent studying, the latter of which could not effectively be reported.

Is there any way that it could make sense to require reporting the # of hours a student spends in SAT/ACT tutoring, whether it be private, through your school, or other?

As a required field on the common app, this could help in making admissions decisions. There are so many reasons using standardized tests for admissions isn't fair across the board, but perhaps understanding how much prep went into the test would provide useful data for admissions offices, and also at the macroacademic (lol) layer.

Built a simple WebGL landing page for an older iOS project where you can wear President's faces in AR. Most of the work actually enough went into optimizing load time... still some room to improve.

https://cloan.me

Certainly, there are lots of apps leveraging the TrueDepth sensor on iPhones that reach an even higher level of fidelity as far is pose estimation is concerned.

The limiting factor for accuracy in a lot of these technologies is the actual rigging process of the characters, probably because that is very difficult to standardize or generalize across different geometries, art styles, animation drivers, 2D vs 3D, etc.

This would make for an excellent after effects (or any post-production software for that matter) plugin. Would highly recommend going this route for the "Unscreen Pro" version that is yet to be released.

Separately, because I'd bet the makers are reading, are there any plans to offer the segmentation models or APIs locally? Was looking for this for the remove.bg product as well.

Dunbar's number seems relevant for this line of thinking: https://en.wikipedia.org/wiki/Dunbar%27s_number

I'm really curious to see how this would work on humans.

It'd be great to see quality level similar to the Obama 3D scan available to consumers: https://www.youtube.com/watch?v=gpMOe7O88cY

I've been hobbying around with Bellus3D's face app leveraging the iPhone's depth camera: https://www.bellus3d.com/faceapp/iphone

This is quite meaningful for external graphics solutions, PCI peripheral devices, and other modular pc components.

I've been wondering recently, (how) do identity protection or fraud protection software services actually work?

I feel like someone actually doing that right would be a big deal but haven't heard anything special about the market leaders.

OG post: https://www.linkedin.com/feed/update/urn:li:activity:6462686...