Monthly Archives: August 2018

Sononym: Finding sound by similarity

This audio browser app has a clever idea, though I wonder if it’d benefit from the kind of rendering that a Google project uses to let researchers visualize thousands of bird sounds via AI:

The primary innovation in Sononym is something called “similarity search”, which enable users to find similar-sounding samples in their sample collection based on any source sound. Essentially, a bit like how Google’s reverse image search works, but with audio.

The initial release focuses strictly on the core functionality of the software. That is, to offer similarity search that work with large collections of samples. Technically, our approach is a combination of feature extraction, machine learning and modern web technologies.

Not entirely dissimilar: Font Map helps you see relationships across more than 750 web fonts.

NewImage

[YouTube]

Absolute witchcraft: AI synthesizes dance moves, entire street scenes

This 💩 is 🍌🍌🍌, B-A-N-A-N-A-S: This Video-to-Video Synthesis tech apparently can take in one dance performance & apply it to a recording of another person to make her match the moves:

It can even semantically replace entire sections of a scene—e.g. backgrounds in a street scene: 

Now please excuse me while I lie down for a bit, as my brain is broken.

NewImageScan

[YouTube]

[YouTube 1 & 2] [Via Tyler Zhu]

Google builds a cloud-based animation studio, creates on the fly

Google showcased its cloud-rendering & collaboration chops by deploying a cloud-based animation studio, enabling a creative team to design & render this short over three days:

To demonstrate what’s possible, we built an animated short over the course of three days.

To do it, we invited some like-minded artists who share our vision to set up a live cloud-based animation studio on the second floor of Moscone Center. These artists worked throughout the three days of the show to model, animate, and render the spot, and deliver a finished short. […]

We used Zync Render, a Renderfarm-as-a-Service running on GCP that can be deployed in minutes, and works with major 3D applications and renderers. The final piece was rendered in V-Ray for Maya.

Zync is able to deploy up to 500 render workers per project, up to a total of 48,000 vCPUs.

Pretty dope—though in my heart, these dabbing robots won’t ever compete with my then-5yo son Finn as a dancing robot:

NewImage

[YouTube 1 & 2]