Browsed by
Month: June 2024

Create a playable game from a single image

Create a playable game from a single image

Google researchers have published a new artificial intelligence model that can take a text prompt, sketch or idea and turn it into a virtual world you can interact with and play.

Named Genie, the virtual world model was trained on gameplay and other videos found online and is currently a research preview. The games are 2D platformer style games.

Genie can be prompted with images it has never seen before, such as real world photographs or sketches, enabling people to interact with their imagined virtual worlds-–essentially acting as a foundation world model. This is possible despite training without any action labels. Instead, Genie is trained from a large dataset of publicly available Internet videos. We focus on videos of 2D platformer games and robotics but our method is general and should work for any type of domain, and is scalable to ever larger Internet datasets. 

Links:

nVidia’s CPU stacks up respectably against the AMD Threadripper 7980X

nVidia’s CPU stacks up respectably against the AMD Threadripper 7980X

More eyes are nVidia’s GH200 Grace Superchip. The GH200 Superchip is the combination of a Grace CPU and a Hopper-based H200 GPU. While the results aren’t definitive, they should raise eyebrows.

In 39 individual Linux-based benchmark tests, Grace (the CPU side) beat the AMD Threadripper 7980X in 17 tests and the 7995WX in 15. It even stacks up well against Intel Sapphire Rapids.

Sapphire Rapids and Threadripper enjoy many advantages. Far more apps are optimized for x86 than for Arm. Threadripper has much more aggressive clock speeds than the efficiency-focused Grace chip and far more L3 cache (7980X has more than double, and the 7995WX more than triple).

But raw performance may not be the only consideration. There is speculation that Grace Hopper is more energy efficient – but there is no real data on TDP outside the whole package requirement of 500w. Also, for tasks that are more GPU bound than CPU bound, the combination may provide higher effective processing power than a faster CPU paired with a discrete GPU. Time will tell.

It shows that nVidia’s Grace entry is not to be ignored. Paired with the Hopper GPUs, it might offer a very viable alternative to x86 stacks – if your software can run on Arm.

Links:

The Driver’s Way helps drivers in Sierra Leone

The Driver’s Way helps drivers in Sierra Leone

In Sierra Leone, there is a step in learning to drive – playing a board game. The game is called The Driver’s Way and is a “Roll and Move” type game where players to roll traffic-light-themed dice and move model cars around a board.  The game aims to teach learners the rules of the road in a more entertaining way than standard textbooks.

Makes me wonder if there’s an opportunity for a board or teen-oriented video game like this for learners – even children to learn how to drive, walk, and bike in shared traffic spaces. Maybe it could be structured like the Oregon Trail or something quite fun.

Two Kinds of Wisdom

Two Kinds of Wisdom

13 Who is wise and understanding among you? Let them show it by their good life, by deeds done in the humility that comes from wisdom. 14 But if you harbor bitter envy and selfish ambition in your hearts, do not boast about it or deny the truth. 15 Such “wisdom” does not come down from heaven but is earthly, unspiritual, demonic. 16 For where you have envy and selfish ambition, there you find disorder and every evil practice.

17 But the wisdom that comes from heaven is first of all pure; then peace-loving, considerate, submissive, full of mercy and good fruit, impartial and sincere. 18 Peacemakers who sow in peace reap a harvest of righteousness.

James 3:13-18
Distracted Clackamas driver launched into orbit

Distracted Clackamas driver launched into orbit

Absolutely amazing this was caught on dash cam, even more amazing they lived. Emergency responders had to deploy a rope system to carry the driver 200 feet up the hill.

Getting closer to synthetic people

Getting closer to synthetic people

Microsoft has released a fascinating new framework for generating lifelike talking faces called VASA-1.

Given a single static image and a speech audio clip, VASA-1 is capable of producing lip movements that are synchronized with the audio and capture a large spectrum of facial nuances and natural head motions.

See more here, read the paper here and here.

Getting worried you’ll be replaced by AI yet? If this gets perfected (it’s not perfect yet, but the results get better and better each year), then you can pretty much get rid of any ‘talking head’ jobs.

This could also be used to fool people on conference calls where video quality would totally render any minor glitches as unnoticeable or easily ignored as just streaming artifacts.

Just slap the CEO’s face into this, set up a conference call with finance via some very easy phishing, and approve that $1m transfer to your Swiss bank account.

Articles: