New user testing out Photostructure as I need something the family can use and understand. This is the perfect tool:-) Have been looking for the error on the forum, however not able to find something similar.
Have been running the Windows version 1.1 for a short time, and it works perfect. However, after reading on the newer versions, I would like to use features like geotag.
It looks like everything is installed with no error, and I can redo the install with no errors. Both the initial SW part, the choco install ffmpeg jpegtran and the npm install --global yarn npm
I get an error on the final “./start.sh” command, and not sure how to proceed, as there is no log I can find.
Copy of the error, also showing the git clone part if information is needed. Error is in the last step below.
Home@Home01 MINGW64 /c/photostructure
$ dir
photostructure-for-servers
Home@Home01 MINGW64 /c/photostructure
$ cd photostructure-for-servers
Home@Home01 MINGW64 /c/photostructure/photostructure-for-servers (main)
$ git checkout beta
Switched to a new branch ‘beta’
branch ‘beta’ set up to track ‘origin/beta’.
Home@Home01 MINGW64 /c/photostructure/photostructure-for-servers (beta)
$ ./start.sh
No local changes to save
Already up to date.
Dependency installation failed.
Please refer to “Removed due to only 2 links for new users”, points to the install artickle.
You can also visit “removed due to 2 links for new users” for help.
I haven’t seen this error before—ideally I’ll be able to set up one of my windows test machines just like yours is currently to reproduce the issue and investigate how to address the problem.
Are you using the Git for Windows bash shell?
What version of Node.js did you install? Did you use the installer, or chocolatey? (I’d use the installer from the website, and then make sure you tick the “Automatically install the necessary tools“ option during installation): PhotoStructure | PhotoStructure for Node
Hi, thanks for reaching out.
I have just been through the following again, just to be sure.
Removed everything, even git and node.js
On a “clean” Windows 11 installed
Installed Git, version 2.46.0 64-bit version for Windows.: No errors.
Installed “node.js”: Version 20.16 as installer from Node.js — Run JavaScript Everywhere. Did ensure “PATH” got installed and choose “Automatically install the necessary tools.”. Installed with no errors
Run “choco install ffmpeg jpegtran” in a powershell. Will not install due to a checksum error. Run “choco install ffmpeg jpegtran --ignore-checksums” and it installs as expected with no errors
Open a git bash window
Run " npm install --global yarn npm": No errors. 14 packages
As mentioned, the Windows version worked, however I would love some of the new features (geotagging), hence I installed the node version. Not sure about the timeframe for Windows, but maybe your (probably limited) spare time is better used on the Windows version.
To be clear: photostructure doesn’t do geotagging (or any tagging for that matter), at least not yet. What the current alpha version does is do a reverse-geocoding to convert GPS coordinates into country, state, city. But this requires that your pictures have already been geotagged outside of photostructure. I apologize if you already knew that, just want to make sure you’re not dissapointed when you get it working.
I just started scanning my 50’000 photos/videos with Photostructue Docker on my Raspberry Pi 5. I’m very curious how many hours (days?) it will take
And you’re right: it took me about one hour to configure everything with Docker. This is also thanks to the fact that I already use Immich with Docker.
It will take some time First time on the pictures, some days however afterwards the updates are fast, like 10-20min.
The videos are more complicated. I created a separate photostructure server, just to thest the videos and the time for transcoding my 1614 videos on a older 4 core Intel laptop was way to long, like close to a week.
I then looked into the possibility to reduce this, and I found that most of my videofiles were in MP4 and MOV format. Both formats (at least mine) were ready for streaming, hence do not need transcoding. However, photostructure used a lot of time transcoding on all videos.
I can suggest to avoid transcoding videos, by setting “transcodeVideos = false” in the setting.toml file. However, make some test first. As mentioned, I created a seperate photostructure server and tested on a few video files.
Some formats do not get imported into photostructure, like MKV/MST, some get imported however do not work. Of my 1600 videofiles, I had to convert 300+ from MKV into MP4 as photostructure do not reconise them. I had to convert 50+ MPEG files, as they were not streamable (as photostructure is set to not transcoding them). This is easily done using FFMPEG.
All this just as a suggestion, will depend a lot on your videofiles.
Matthew’s specific contribution is about PhotoStructure on a Raspberry Pi. The software recognizes it runs on Pi and automatically disables several time-consuming features (including the video transcoding you mentioned).
Extract from this article:
Here’s a list of the features disabled by PS_QUICK_SYNC_MODE:
validateJpegImages
validateRawImages
validateVideos
transcodeVideos
jpegMinimized
jpegProgressive
sharpen
useImageHashes
enableSiblingInference
With these time-consuming features disabled, my Pi 5 can scan/transcode up to 2000 assets per hour, which is quite good. The speed when browsing through the pictures/videos is surprisingly good. So a Pi 5 is a perfect recommendation for a private Photo Server. The power consumption of less than 5W is also unbeatable.
