AI Image Colorizing, Sodium Reduction Science and Strategies, Spanish Civil War, More: Sunday ResearchBuzz, October 23, 2022


Ars Technica: AI tool colorizes black-and-white photos automatically. “A Swedish machine learning researcher named Emil Wallner has released a free web tool… that automatically colorizes black-and-white photos using AI. After uploading a photo, users can choose a color filter or refine the colors using a written text description.” Spent half an hour playing with this. Excellent work.

Institute for the Advancement of Food and Nutrition Sciences: Sodium Reduction Science And Strategies Database. “With heightened concerns about sodium intake and the limited impact of public health campaigns, strategies for reducing sodium in the nation’s food supply are increasingly important. A new searchable database of over 280 scientific papers looks at flavor, functional and physical modifications to lower sodium intakes.”

The Guardian: Virtual Spanish civil war museum aims to cut through political divide. “The Virtual Museum of the Spanish Civil War, an online history centre that has been almost a decade in the making, may chronicle and examine a conflict that ended 83 years ago but its aims could not be more timely. As last Wednesday’s senate session demonstrated, there is still precious little consensus over the 1936-39 war and how to deal with its bitter legacy.”


The Mainichi: Yahoo Japan to require users’ cell phone numbers to post comments on news stories . “Yahoo Japan Corp. will make it mandatory for users to provide their cell phone numbers when posting in the comments section of its news distribution service in order to deter inappropriate postings, the IT firm has announced.”


CNET: Getting Political Text Messages on Your Phone? Here’s How to Block Them. “If you’re wondering how the organization got your number in the first place, it’s because all states allow access to voter data for election purposes — so if you’re a registered voter, your information is on file. Here’s how to stop unwanted political texts on your phone.”


Times of India: How an Andhra village learnt to talk in American accent. “Students at this government school in Bendapudi village in Andhra Pradesh are learning to speak English the American way. Along the way, they have become internet celebrities.”

Discover Music: The Digital Music Revolution: From The MP3 To Music-Is-Free. “She may not think so herself, but Suzanne Vega just might be the most influential figure in the past three decades of music. That’s because ‘Tom’s Diner’ was the very first song to be digitized when a German engineer, Karlheinz Brandenburg, unveiled a new audio compression tool in the early 90s, pointing towards the future of digital music.”


Jerusalem Post: Iranian hackers claim to have obtained files of Iran’s ‘dirty nuclear projects’ . “Iranian hacker group ‘Black Reward’ announced over the weekend that it had successfully hacked the internal email system of Iran’s Nuclear Power Production and Development Company and that it was releasing 50GB of files to the web.”

New York Times Magazine: She Was Killed by the Police. Why Were Her Bones in a Museum?. “Katricia Dotson’s remains were studied, disputed, displayed and litigated. Lost in the controversy was the life of an American girl and her family.”


Globe and Mail: Social media emerges as primary fan-engagement platform for sports fans. “Once a minefield of complications for sports teams and players, social media has matured into a key enabler of fan experiences and connections. And when it comes to online engagement, Toronto teams tend to punch above their weight class.”

UCLA: There’s a lot of hate in the world. UCLA’s scholars are asking why and what can be done. “UCLA is launching the Initiative to Study Hate, an ambitious social impact project that brings together a broad consortium of scholars to understand and ultimately mitigate hate in its multiple forms. Supported by a $3 million gift from an anonymous donor, researchers will undertake 23 projects this year. The three-year pilot spans topics that examine the neurobiology of hate, the impact of social media hate speech on kids, the dehumanization of unhoused individuals, racial discrimination in health care settings and more.” Good morning, Internet…

Do you like ResearchBuzz? Does it help you out? Please consider supporting it on Patreon. Not interested in commitment? Perhaps you’d buy me an iced tea. I love your comments, I love your site suggestions, and I love you. Feel free to comment on the blog, or @ResearchBuzz on Twitter. Thanks!

Categories: morningbuzz

5 replies »

  1. That AI colorization is great! Running old B&W family photos — which I’ve seen HUNDREDS of times– through it is really eye-opening, and in some cases quite moving. As the Ars Tecnica article mentions, I’m not sure how comfortable I am with having private stuff on the cloud, but it’s thrilling just to see it in action.

    (I do have Photoshop but not on my Chromebook,so I can’t use PS’s colorization filter while on the road.)

    • One thing which I was curious about: with human features and maybe common objects like iconic buildings and, oh, say, Ticonderoga pencils, it’s pretty easy for the software to guesstimate the true colors… but how well would it work with things like fabrics? The sample I tried was my sister-in-law’s old high school cheerleader team photo. Interestingly, the AI guessed the school’s colors to be either bright red or royal blue — a different color scheme for each cheerleader — although it game them identical solid-black uniforms. 😵‍💫

  2. Have you tried editing the text description that the AI uses to color the picture? Hold your mouse over the image to see it and click on the pencil icon on the right side of the screen to edit it. It can change your colors a lot. I’ve found putting in a brief description of the light conditions — sunny, overcast, early morning, etc — can go a long way to getting the colors right.

    • Huh – I saw those text descriptions but didn’t realize they were editable. (It kinda surprised me… I was submitting mostly old family photos, people sitting on sofas, sometimes sleeping on a sofa, and it was weird how often the word “tenant” showed up as a keyword!) The only fine-tuning I tried was really brute-force: just rotating among the predefined filters. Thanks!

Leave a Reply