Twitter needs every penny. With millions of dollars in allegedly unpaid rent and bills, plus $13 billion owed to the lenders who financed its takeover, there’s “a lot more work to do” if the company wants to avoid bankruptcy, Musk said last month, Wired wrote on the subject.

Twitter recently auctioned off about $1.5 million worth of furniture and equipment from its San Francisco headquarters, including odds and ends like keyboards and USB keys. But the company has left shiny assets worth tens and perhaps hundreds of thousands of dollars gathering dust in the homes of its former employees.

Some people fired or laid off by Musk are perplexed as to why Twitter hasn’t bothered to collect their company laptops, the latest problem in a takeover characterized by botched product launches, abrupt policy changes and delayed payrolls.

Eric Fronhofer, a California software engineer fired in November after confronting Musk via Twitter, says he hasn’t heard a word about bringing back his company’s 2021 Apple MacBook Pro M1 laptop. “It’s still sitting in the closet,” he says. Like the laptops of thousands of remote Twitter employees Musk has fired or let resign since early November, his has been digitally locked, rendering it useless.

Restored versions of his model can still be bought for about $1,000, and new ones are twice as expensive. Fronhofer feels no obligation to Musk and is in no hurry to return the machine. “I’m happy to let it sit there and be a brick,” he says.

Some former Twitter employees have told co-workers they sent the equipment back after approaching the company about prepaid shipping boxes. Others within the past few days have received generic emails asking them to fill out a “Twitter Device Collection Survey,” many say. But four of the five who have spoken to WIRED have not received the email themselves and are still eyeing Musk’s property.

In typical productions, digital re-aging of faces in video involves a complex 2D painting workflow that often requires a skilled artist to perform frame-by-frame manual work that can take days to accomplish. This tool changes that by first by tackling the problem of collecting longitudinal training data for learning to re-age faces over extended periods of time.

“Finally, our new face re-aging network (FRAN) incorporates simple and intuitive mechanisms that provides artists with localized control and creative freedom to direct and fine-tune the re-aging effect, a feature that is largely important in real production pipelines and often overlooked in related research work”,

said the researchers.

The research team demonstrates how such a longitudinal dataset can be made by leveraging current state-of-the-art in facial re-aging technology.