Disney published this paper about using AI to digitally age and de-age actors in a fraction of the time it usually takes for normal frame-by-frame manual aging techniques used today.
FRAN (which stands for face re-aging network) is a neural network that was trained using a large database containing pairs of randomly generated synthetic faces at varying ages, which bypasses the need to otherwise find thousands of images of real people at different (documented) ages that depict the same facial expression, pose, lighting, and background. Using synthetically generated training data is a method that’s been utilized for things like training self-driving cars to handle situations that aren’t easily reproducible.
The age changes are then added/merge onto the face. It appears this approach fixes a lot of the issues common in this kind of approach: facial identity loss, poor resolution, and unstable results across subsequent video frames. It does have some issues with greying hair and aging very young actors, but produces results better than techniques used just a few years ago (not that the bar was very hard to beat).
The Markup reports that tax filing services like H&R Block, TaxAct, and TaxSlayer have been quietly sending their financial information to Facebook.
These services use Meta Pixel (also known as the Facebook retargeting pixel) – a snippet of code inserted into the back end of your website to track visitor activity. It works by loading a small library of functions which can use whenever a site visitor takes an action (called an event) that they want to track (called a conversion). Tracked conversions appear in the Ads Manager where they can be used to measure the effectiveness.
It turns out the information sent included users’ income, filing status, refund amounts, and other information. The team that found this behavior has a github repo that shows the actual data sent if you want the gory details.
Intuit’s TurboTax also uses Meta Pixel – but they do not appear to send financial information to Meta.
I have a nice MSI MPG Z690 Carbon WIFI motherboard with my shiny new Intel 12th gen i9-12900k processor. Recently, I tried to upgrade my 1TB Samsung 960 EVO M.2 with a 2TB Samsung 970 EVO Plus M.2; but ran into a hitch. When I booted from the Windows 10 installation USB, the NVME drive would not show up in the list of drives for installation. Running Windows repair tools didn’t help.
Hmmm. I tried running the installer repair tools – but it would give a unhelpful errors and no drives would appear. Even though I had 1 NVME drive and a standard old SATA drive as well.
This is probably because you have Intel Rapid Storage Technology enabled. If yes, I would suggest keep using it, because it will let you get as much performance as possible, from you NVMe drives, on Windows 11. You need to run W11 installer (with VMD enabled) and the load Intel drivers. It will make your M.2 drive appear in the Windows installer. https://download.msi.com/dvr_exe/mb/intel_rst_19.0.zip Remember that the drive will never be visible in BIOS when using VMD. It is normal.
If you rather don’t want to use VMD (not recommended) – disable the Intel Rapid Storage (or Intel RAID – don’t remember how it’s called in BIOS).
When I went into BIOS, I noticed VMD (RAID) was indeed enabled in my BIOS because I had a set of RAID 5 drives on my previous installation. For an experiment, I turned off VMD (RAID) in BIOS, booted from my Win10 installer USB, and sure enough I could see my NVME and other drives during Windows installation. If I turned VMD back on, the drives would disappear.
The solution came from something I should have realized earlier. The Windows 10 installer (and apparently Windows 11) doesn’t know about my fancy Z690 chipset storage devices. I needed to download the Intel RAID storage controller drivers, unzip them on my Windows installation USB drive, and then manually load those drivers at the drive selection page during Windows installation.
So, here’s the two solutions (but solution 1 is best)
Solution 1: This method includes full Intel storage controller and RAID support. It’s your best bet for full perforrmance and full functionality.
Turn the system off and install the NVME M.2 drive
Boot to BIOS, turn VMD (RAID) on
Save BIOS settings and boot off the Win10 install USB stick
When you get to the drive selection page in the Windows installer there will be no drives. Click the ‘Load Driver’ button.
Browse to the USB stick, select the directory where you unzipped the Intel RST drivers. Be sure to point to the proper sub-directory with the floppy version of the driver files: <unzip root>\VMD\f6vmdflpy-x64\
You should see at least two Intel devices listed. You don’t need to select anything, just hit ok and Windows will load the drivers
You’ll be returned to the drive selection page and you should see your NVME, RAID, and other drives listed!
Pick the boot drive you want to install too, and hit OK.
Windows 10 will install and you should boot normally after that. You should see all your drives – including any RAID sets you already had.
Be sure to run Windows Update and download/install the latest Intel RAID drivers.
Solution 2: No RAID support later without completely reinstalling Windows. It’s also likely you will not be getting the full performance of your NVMe or other drives either.
Create the USB Windows 10 installation stick.
Turn the system off and install the NVME M.2 drive.
Boot into BIOS, and turn VMD (RAID) off
Reboot from the windows installer USB stick.
Install Windows like normal. Turning off VMD will let you see all your attached drives during the installation phase (all but any RAID drives) and install Win10 on any of them.
You will not be able to use hardware RAID of your motherboard unless you completely re-install Windows. If you turn VMD/RAID on later, the system will ‘lose’ the NVME drive and refuse to boot. If you set it up with the RAID controllers, then the Windows bootloader apparently makes sure the drivers for the VMD device are always loaded. This is why it’s recommended to use solution 1, because it lets you use RAID later if you want.
Any CS/CEE program worth its salt uses the Computer Architecture book from Hennessy and Patterson. Besides the operating system chapter, the chapter on memory was one of my favorites. Enter Branch Education that created this really excellent video on how computer memory works.
Definitely worth the 35 minutes. It starts slow, but around section 8 it starts going fast and furiously through a lot of different concepts. Each builds on the previous, so you may need to pause and really make sure you grock what they said.
I have a Brother DCP-L2540DW combined scanner, copier, and laser printer unit. It’s been a great little unit I picked up for about $100 on a Black Friday sale that has cheap cartridges and loads of features for the occasional printing/copy job. On a side note, laser printers are a FAR better option than inkjet printers if you only occasionally print something. Now that good laser printers are below $100 and their toner is far more stable than inkjet printers that dry up and clog if not used regularly, they’re really the way to go for occasional everyday printing.
Install and click on the Brother Utilities provided with the printer:
Click ControlCenter4 to open the Control Center 4 app.
On the Scan top menu, click the button for the type of scan that you would like to perform. For a multi-page PDF, you want to select File. A new dialog box will open:
Be sure to change the File Type to PDF and then select the folder you want to save the document.
Load the upper tray with the pages you want to scan. I’ve scanned up to 40 pages simultaneously this way without issue.
Press the Scan button and the bottom and it will scan all the pages and combine them into a single PDF that is located in the folder set in the Scan Location directory you specified.
If you have enabled the ‘Show settings dialog before scan’ option, then you will be prompted to choose your desired settings and then click Start Scanning.
If you are using the Continuous Scanning option, you will need to load the subsequent pages either on the flatbed or in the Automatic Document Feeder (ADF) if available and then click Continue. Once you have scanned all pages, click Finish.
AI generated art has caught fire. Learning how to generate the command line prompts to generate the art is still a work of trial and error. But some folks are helping you learn by giving some example prompts to help you learn what works and what doesn’t.
All the below items were 100% auto-generated and included on the page. It looks like people are exploring and sharing different prompts to generate different kinds of art.
One of the first things you’ll run into is that you’re not going to be able to generate any images at 512×512 or larger if you have a graphics card with 8GB of VRAM or less. Even smaller if you only have 4GB of VRAM. The first/easiest method is to limit the output image size. There is also an option that splits the model into 4 parts and loads each separately (thought it will take longer), or using a more optimized/compressed set of trained model data.
So how do you do that if you have an older graphics card that only has 4GB or 8GB of VRAM? TingTingin has some tips at the end of his installation video if you are using a card with 8GB of VRAM (nVidia 3070’s for example).
Summary (at 15:45): Modify your txt2img.py and add the line ‘model.half()’ after model = instantiate_from_config(config.model) in the load_model_from_config() function.
Reset your graphics driver – Win + CTRL + SHIFT + B
Remembering a wifi password you entered forever ago: c:\> netsh wlan show profile < shows a list of wifi profiles you have connected to > c:\> netsh wlan show profile <wifiProfileName> key=clear < shows the wifi password you last entered >
Message every single windows system on the network c:\> msg * <message>
Hackertyper – Hit F11 to fullscreen your browser and then look like a hacker. https://hackertyper.net/
FakeUpdate – go to the website, get on your friend’s computer, load the update screen for that OS, hit F11 to go fullscreen, and then see how long they’ll sit there before resetting. https://fakeupdate.net/
Andrew Ng is one of the biggest names in AI. He makes a few predictions, and I thought the article had some good observations.
His current big focus is using AI in manufacturing. Andrew Ng founded Landing AI in 2017. His focus was primarily consulting, but after working on many customer projects, Ng and Landing AI developed a new toolkit and playbook for making AI work in manufacturing and industrial automation. This led to Landing AI and the development of a data-centric approach to AI.
“In consumer software, you can build one monolithic AI system to serve a hundred million or a billion users, and truly get a lot of value in that way,” he said. “But in manufacturing, every plant makes something different. So every manufacturing plant needs a custom AI system that is trained on their data.”
The challenge that many companies in the AI world face is how to help 10,000 manufacturing plants build 10,000 customer systems. In short – scale.
In manufacturing, there is often no big data to go by. The data for manufacturing different products is unique. Their first observation was to see it makes more sense to keep the models relatively fixed while focusing on quality data to fine-tune the models rather than continuing to push for marginal improvements in the models.
This uniqueness of data also means there is almost never enough images of faults or cases to train models. The only way out of this dilemma is to build tools that empower customers to build their own models and let product experts engineer the data and express their domain knowledge. Ng and Landing AI do that through Landing Lens, which enables domain experts to express their knowledge with data labeling instead of constantly tweaking the models.
