This update fixes a couple of problems with the screensaver. It now has improved support for multiple monitors (the image will repeat rather than stretch). There was a problem with the screensaver not starting in some cases. This should now be fixed. After upgrading to 4.30 please find and run the 'Catfood Earth Screensaver Settings' shortcut to configure the screensaver.
4.30 also includes version 2022d of the time zone database.
A horror movie made by forcing img2img to watch a Japanese security camera over 24 hours.
The office was empty, so the main influences are sunrise and sunset and then just subtle shifts in lighting. In terms of settings, strength was 0.75 with scale of 6.6 so the animation doesn't look anything like the actual office in question. The prompt was "photo of something unspeakable lurks in the shadows of the office, high quality, cinematic lighting, subtle horror, horrifying, japanese, found footage, sharp focus, 8k, no text". I upscaled with Gigapixel AI and then added some music.
I have an idea for PG&E. We're facing the possibility of rotating power cuts this evening in California due to record heat. The California ISO has a thing called a flex alert where they ask you nicely to reduce power consumption, and I got a text and email earlier today. If the ISO calls for outages PG&E starts flipping off the switch for some blocks, and you can even look up your address to see if you're at risk.
This could be so much better.
Just keep tabs on average power consumption for each block during hot weather. And then instead of a flex alert we need a leaderboard - the blocks which save the most power are safe, and the blocks that do the worst are the first to get shut off.
Another experiment with Stable Diffusion (see my San Francisco skyline video from earlier today). This one uses img2img instead of txt2img. I started with a video of my dog following the Roomba around the house. I dumped all the frames out and then used Stable Diffusion with the strength parameter ramping up from 0.0 (source image preserved) to 1.0 (source image ignored) and a scale of 11.5. The prompt was "illustration of a black labrador being chased by a giant scary roomba trending on deviant art". The frame at the top maybe best captures this concept. I used Gigapixel AI to scale the output back up to 4K resolution and then added the original soundtrack.
Here is a video created by animating the scale parameter in Stable Diffusion:
This uses the prompt "a photo of San Francisco shrouded by fog 8k landscape trending on artstation", seed 56345, dimensions 896x448 and frames that range from scale 0.0 to 21.9. The higher the scale the more literally your prompt should be interpreted.
Interesting at scale 0 Stable Diffusion seems to think San Francisco is a diverse crowd of people protesting something (and it's not wrong):
From there is settles on a foreground quickly, experiments with various numbers of Transamerica Pyramids and then starts throwing in various permutations of the Golden Gate Bridge. It eventually decides that the Transamerica Pyramid belongs in Marin and that the bridge is so iconic that the city is probably filled with red aircraft warnings lights.
I think the versions a third of the way in are the best.
Photo Sorter has been updated to handle some duplicates I've been developing in Google Photos. These are pretty specific rules but might be helpful if you are trying to maintain a local archive from Google Photos via Google Takeout. You can get the latest binary and source from github (or fork away if it's not quite what you need).
The first change is that Photo Sorter now checks for duplicates in the source folder as well as the destination. If two source files have the same date taken and the same filename then the larger file is chosen as the winner and the smaller file is deleted. The filename check ignores anything in parentheses, so 123.jpg is considered to be the same filename as 123(1).jpg. This helps alleviate a fun bug where Google Photos will export via the API a different file that was originally uploaded. I've stopped using the Google Photos API for this reason, and because it will under no circumstances allow you to download a video that is the same quality as the original upload. Crazy edge case Google. Happily Google Takeout still works so I'm stuck doing it slowly and wastefully.
The second change is that if a source duplicate is found using the rules above then it will also be deleted from the destination folder (in order to be replaced by the presumed better version of itself).
Photo Sorter copies some folder full of photos and movies to a different folder with a clean structure and some de-duplication. It's been keeping me sane since 2018. Full instructions here.
I've been experimenting with a four-foot Great Video Maker GR-120QD slider for timelapse photography. So far I'm impressed with the quality for a relatively cheap device but the controller is a little hard to understand. There are various YouTube videos that explain this at length but I find a quick written explanation more helpful (and as much as anything, this post is for me to refer back to quickly!). These instructions will also work for the three-foot GR-80QD.
The first thing to understand is that you need to recalibrate the device every time you power it on. It's a bit of pain but fast once you get the hang of the Konami Cheat Code style UI. Go to settings, Set Start and then hit the left button to start moving. When you get to the desired start point press the center button to stop the slider, long press the center button to save it and then long press again to up a level in the menu. Do the same for the end point at the right side of the slider. More long presses to get back up to the top menu.
Now the fun part. For timelapse you have Interval, Time-lapse, Stop Time, Photo and Auto Loop and the user guide is silent on what any of those might do.
When shooting a timelapse the slider is going to move, stop for a bit and shoot and then move again.
Stop Time is the total amount of time that the slider stops for each shot (seconds).
Time-lapse is the number of seconds to wait after stopping before the shot is taken. This allows the camera a chance to stabilize after being moved along the slider before a shot is taken. The system enforces at least 0.2 seconds but you probably want this a bit longer.
Shutter time is in the hands of the camera.
So if you have a 1 second shutter speed and want to shoot every 4 seconds you would set Stop Time to 4 and time-lapse to something like 1. The slider will then move, wait 1 second, trigger the shutter, wait 3 seconds (the remainder of the 4 second Stop Time) and will then move again.
If Auto Loop is set to No then the system will stop shooting when it reaches the end of the slider, so you can set Photo to a large number and not worry about it. If Yes the camera will go back and forth until it hits the count of photos in the Photo setting.
The last thing to worry about is how far the slider moves between shots. This is a combination of Interval and the slider speed, which you set when you start the time lapse. The minimum interval is 0.1 - I haven't tried anything else. At 70% speed this takes about 45 mins (based on the 4 second stop time) and shoots almost 700 photos which for me is a pretty decent sequence. This part is not a science so you'll need to experiment to get the result you're looking for.
When the settings all look good, hit the center button and you will go to the time lapse screen. Adjust the speed with the up and down arrows and then hit the right or left button to start shooting in that direction.
Updated 2022-08-29 17:43:
Here is some test footage at 60% speed and 0.1 interval. The video starts siding to the left (camera pans to right) at 1/1000s shutter. The last section reverses direction and has a 1s shutter with an ND3 filter to check for any shake. Both sections are shooting every 4 seconds with a 1 second 'Time-lapse' (or stabilization time) setting.