Google Pixel 8 Pro 18mm f2.8 1/257s ISO17
Ok, technically it's R:Evolution, towering over the end of Market Street trying to make downtown happen again while gazing at a haphazardly decorated utility box.
(Published to the Fediverse as: Revolution #photo #revolution #sanfrancisco Photo of the R:Evolution sculpture towering over the end of Market Street )
I set myself the goal this morning of shipping a vibe coded web application in a day. Welcome to the world, vibevideo.xyz, and apologies if you own Adobe stock.
Vibe Video is a chat based video editor. There are two things that are pretty cool about it. First, it writes its own code. When you ask it to make an edit it uses GPT 4.1 to write a function for the edit and then runs that function. That's the Vibe Video part. It's a single page JavaScript application but the core functionality is hallucinated at runtime.
Second, it runs locally. It's based on ffmpeg.wasm, which I knew nothing about before this morning. The site is a Vite/Svelte SPA deployed to a Cloudflare site with no backend except the OpenAI API. Your videos never leave your computer, and all the editing happens in the browser thanks to WebAssembly. I hadn't touched Vite or Svelte before today either.
Both of those things are crazy. There are some limitations - as the great saying goes, "If you are not embarrassed by the first version of your product, you’ve launched too late.". Right now you need to bring your own OpenAI API Key. Maybe I'll add some free usage when I get a grip on how much it costs to run. It's fast when you don't need to reencode - so trimming the video is quick. but changing resolution or speed will take some time. And there is a 2GB limit on file sizes, a limitation of WebAssembly. My hope is to iterate and improve, so kick the tires and let me know what else it needs.
To build this I used Visual Studio Code with GitHub Copilot in agent mode using Claude 3.7. The only code that's hand crafted is changing GPT 4 Turbo to GPT 4.1 (Claude refused to admit that 4.1 exists) and repeatedly removing a call to a non-existent file size property. While I didn't have to write any significant code I did need to coax the LLM - an experience halfway between being a product manager and pair programming with a buddy who is a much faster typist and won't allow you to touch the keyboard. There was a fair bit of troubleshooting involved as well, some with the environment setup and some with runtime errors. Having the experience to help guide Copilot through these helped a lot.
I have been experimenting with a lot of different tools over the past few weeks. Cursor and Windsurf both worked well for me, Replit was pretty good as well. Cline impressed me with debugging an app after building it. V0 and Lovable struggled with my test use case. I used Copilot for this personal experiment as I already have a license and wanted to see how far it would go. All of these tools are going to continue to improve (and/or merge). The current state of the art is a big deal already. A few weeks ago the statement that a good chunk of Y Combinator startups were writing 95%+ of their code with AI sounded like bullshit. Now I think they might be sandbagging.
One bad smell is that the experience gets worse as you add code. The start of the project was unbelievably easy, but as features were added Copilot started to bog down. This is unsurprising as the more you load up the context window for an LLM the more it struggles with the details. Again, this will just improve over time as models and agentic implementations improve. Good architecture is also going to matter - and is no different here from freeing regular developers from needing to keep too much in their head at the same time.
(Published to the Fediverse as: Vibe Coding a Vibe Video Editor #code #ai #vibe #openai #llm #anthropic How I built a video editing service in a day with GitHub Copilot and ffmpeg.wasm )
Google Pixel 8 Pro 18mm f2.8 1/450s ISO120
A carp begs for food in Spreckels Lake, Golden Gate Park.
(Published to the Fediverse as: Begging Carp #photo #carp #goldengatepark Photo of a carp begging for a snack in Spreckels Lake, Golden Gate Park, California )
Sunset Dunes is the newest park in San Francisco, created in the November 2024 election by a controversial ballot measure. It's a two mile stretch of the Great Highway that was often closed due to drifting sand. The road was also closed to traffic during the pandemic, a move popular enough to get enough support to convert it to a full time park. Many people are mad with this development due to traffic spilling onto neighboring streets and increased commute times, to the extent that there is a recall campaign against the local supervisor. I'm bored of recall campaigns and NIMBYs and voted in favor of the new park.
The video below is a hyperlapse of the two miles from the Golden Gate Park to Sloat Boulevard through Sunset Dunes and then the two miles back on Ocean Beach.
Not much has changed since the pandemic experience. This isn't surprising as the ballot measure didn't add any funding for the park. The two sides of the original Great Highway are divided into a slow lane and a fast lane which makes sense. There are a few street murals and art installations dotted along the road which are fine but seem beside the point. I hope this remains as a safe space to cycle, scoot, board and stroll.
Hike starts at: 37.767321, -122.509288. View in Google Earth.
(Hike Map)
(Published to the Fediverse as: Sunset Dunes #hike #greathighway #sunsetdunes #sanfrancisco #video #hikevid #sfdogwalk #oceanbeach #ggp #map Four mile hike down the Sunset Dunes park and then up Ocean Beach in San Francisco, California.</span> )
Spring has sprung (09:02 UTC, March 20, 2025) for the Northern Hemisphere and Autumn is here if you are south of the Equator. Rendered in Catfood Earth.
(Published to the Fediverse as: Vernal (Spring) Equinox 2025 #code #earth #catfood #equinox #spring #vernal #autumn The exact moment (09:02 UTC, March 20, 2025) of the Spring Equinox in Catfood Earth. )
Google recently bricked my digital photo frame, so I set out to build a new one. I did this in two parts - a Raspberry Pi to display the photos, and a C# script to preprocess them. The second part is optional but worth it.
The display side of the project turned out to be way easier than I thought. There is a utility called fbi that will run a slideshow to a HDMI monitor. Create a boot SD card from the lite version of Raspberry PI OS, copy over your photos and then run:
sudo apt-get update sudo apt-get -y install fbi
You can then connect a monitor and test that images are displaying as expected.
Create a file called launch.sh with the following:
#!/bin/sh # launch.sh # run photos sleep 2m fbi -T 1 -noverbose -mode 1366x768-30 -t 60 -u -blend 1500 -a /home/rob/photos/*.jpg
-T 1 uses the first display, -noverbose just shows the photos, -mode depends on your monitor and is likely safe to omit, -t 60 changes the image ever sixty seconds, -u displays in a random order, -blend 1500 cross fades for 1.5 seconds between images, -a is auto zoom and the path at the end is where ever you copied your photos to.
The sleep 2m command allows the system to complete booting and bring up the login prompt. Without this the photos might start first and then the login shell ends up on top, which is pretty boring.
Make the script executable (chmod 755 launch.sh) and then edit your crontab:
sudo crontab -e
Add the following:
@reboot sh /home/rob/launch.sh >/home/rob/logs/cronlog 2>&1 30 22 * * * /sbin/shutdown -h now
The first line runs the launch.sh script at startup and sends any output to a log file (adjust paths to your system). The second line will shut down the Pi at 10:30pm every day. I use this so I can control when it's running with a smart plug - the plug turns everything on at 7am and off at 10:45pm, and the shutdown prevents the Pi from getting in a bad state from having power removed unceremoniously. If you want it running all the time just omit this line.
Reboot and you should have a working digital photo frame. Fbi will do a reasonable job with your photos, but I wanted something better.
The C# script below preprocesses all my photos to make the best use of the frame. I'm using an old 1366x768 TV so I want all the photos at that resolution and 16x9 aspect ratio. Many of my photos are 4x3, or 3x4, or 9x16, or something else entirely. I don't want any black borders so cropping will be involved.
Cropping is handled by detecting faces and then trying to keep as many in frame as possible. This uses FaceAiSharp.
Photos are processed in two stages. The first stage just crops horizontal photos to fit the screen. Any vertical photos are saved in a list, and then paired using image embeddings (CLIP using this code). My implementation pairs the most similar photos - it would be easy to do the most different as well.
Here's the code. I run this with C# 9 in VSCode on Windows. You'll probably want to change the input and output folders, and possibly the output resolution as well:
(Published to the Fediverse as: Building a Digital Photo Frame #code #c# #raspberrypi #ai #ml How to create a Raspberry Pi based digital photo frame with face aware cropping and AI image pairing. )
Microsoft is killing Skype in May 2025. I haven't used it for a few years, but I'm going to shed a tear or two.
It's hard to overestimate how important Skype was a little over 20 years ago. I had recently moved to San Francisco and made a lot of phone calls to friends and family back in the UK. There was this thing called a landline, and as well as local service you had to choose a long distance provider with all sorts of complicated tariffs and fees. Cell phones were their own nightmare and even for SMS you paid per text message. Incomprehensibly, ring tones were on their way to being a multi-billion dollar business. My current phone has been on vibrate for its entire life and I travel the world with free high speed Internet included in my plan. Skype marked the beginning of this transition using P2P to open VoIP and video calls to the masses. I even had a cordless phone that made local calls over the phone line but seamlessly switched to Skype for long distance.
Skype's P2P stack could be used for more than phone calls. I built a business to backup your computer over Skype. It's one of the best things I've ever worked on. The Internet was slow for most people. and so you could do a local backup on LAN and then send incrementals P2P via Skype whenever your computer was idle. We also did superior local backups with full history, encryption and locked file support when most competitors choked on PST files. Skype were kind enough to include us as a premium extra in the client. We did some paid search and PR, but the lion's share of our customer acquisition came from this placement.
My current job involves many Teams calls, but before that we operated on hundreds of Skype channels. Before that work was driven by email and conference calls. I'm ambivalent about this shift. Email has some advantages, and hours of video conferences are way more draining than the voice bridge alternative. Teams brilliantly combines channels and chats that look very similar but work completely differently adding an entirely new mental load to the workday. Skype used to display a special cat emoji if you held down three keys at once. I know which philosophy I prefer.
(Published to the Fediverse as: Bye Skype #etc #skype #microsoft Memories of Skype, including building a backup business on it, that mysterious cat emoji, and a landline with Skype built in. We'll miss you! )
I want to do two things with my digital photos. First, keep them safe, especially all those precious memories of random parking meters and unfathomable HVAC mechanisms. Second, enjoy looking at the small subset that are precious family moments. I'm not a special snowflake, these basic requirements should represent a large and competitive market. Unfortunately I'm not holding my breath for much in the way of consumer friendly regulation for the next four years in the US.
Google is making some changes to their Photos API next month, which amount to "Get the fuck out of our Photos API". They're a polite organization so they phrase it a little differently: "We're excited to see the creative solutions developers will build using the new Picker API and the updated Library API.". The developer documentation is a little more pointed: "If your app relies on accessing the user's entire library, you may need to re-evaluate your app or consider alternative approaches."
I have an Aura Carver Mat, a nice digital photo frame that I synced to a shared Google Photos album. Easy for me to add photos, way too easy for the kids to add photos - a fantastic device. As of next month it's ewaste though. I'm not going to upload photos slowly through some picker API like an animal. I'm going to end up building something complicated out of a Raspberry Pi (adds to actuarially unrealistic to do list).
This change doesn't impact backup, because that was already broken. For a while Google Photos nicely integrated with Google Drive and I ended up with a local copy of everything that I could then backup through other means. I'm never going to trust any one company to look after important files and so my philosophy is to backup twice online and once to an external hard drive that lives in a fire safe. (At one point I even built a backup company based on this model).
Google killed the Drive integration and so I MacGyvered together an apps script based solution that used the Google Photos API. This revealed to me that the Photos API would not return location information. Even worse it was impossible to get the full resolution version of a video to download. So it's not like I was in love with the API before this most recent change.
My current approach is a mix of sad and awesome. The sad part is that I use Google Takeout once a month to get an archive of all my online photos. Thankfully this still works. The awesome - I wrote this photo sorter tool that takes the messy download and organizes it by year and month. And I also wrote a volume shadow copy tool that lets you backup a drive without getting hung up on locked files. Those pieces get my photos safely to an external drive, and I upload to Amazon Photos too (the third leg of my backup stool).
Update - fixed with a Raspberry Pi!
(Published to the Fediverse as: Google Photos killed my Aura Frame #etc #google #photos #aura #backup Google is changing their Photos API so my Aura frame is ewaste. Thoughts on using and backing up digital photos. )
(Published to the Fediverse as: Simple Perceptron #code #ml #perceptron #wml Python notebook illustrating a scratch perceptron implementation as well as an sklearn example. )
