Where it all started
Every collector hits the same wall. The collection grows, the folders multiply, and the gems you saved years ago sink without trace. I was that collector. I was also a software developer with more than twenty years behind me, and this was a problem I could not wait to solve.
-
The concept
File explorers are slow, and they force every file into a single folder. Real content does not work like that. One video belongs under its star, its studio and a dozen other things at once.
So I set myself four rules. Work with the folders and file names collectors already have. Put tagging at the heart of everything, so one video can live in many places at the same time. Give those tags intelligence, starting with inheritance. And never lose a file because it was renamed or moved.
One more, and it was not negotiable: searching had to be instant.
-
Two engines
The first thing I built was not a screen. It was a pair of engines.
The index engine scans your files and pulls out names, keywords and tags. The search engine takes that index and filters, sorts and searches it at a speed no file explorer can touch. Point them at a folder, let them run, and your collection comes back titled, tagged and thumbnailed.
Those two engines still power pornspective today.
-
pornspective v1
Version 1 was about function first and looks second. I chose a quick, familiar technology for the interface so that I could pour my time into the engines.
Then the features came thick and fast. In twelve months it gained admin screens, thumbnail editors, bookmarks, a built-in video player, plugins that fetch information online, and several ways to view your results. It worked, and people were using it.
-
The game changer: pornspective.ai
One job still ate hours: tagging. Computer vision was just emerging, and I wanted it. Bringing it to Windows meant working most of it out for myself.
Nobody offers a ready-made model for this subject, so I trained my own. I wrote helper applications to prepare the training data, fed the model ten thousand images, and held my breath. The results went far beyond anything I had hoped for.
I made it a separate application on purpose. Heavy scanning never slows the organiser down, and the AI can run on a different machine altogether. The two apps swap a results file, or simply talk to each other directly.
-
pornspective.v2
The first interface had always held me back. For version 2 I tore the front end down and rebuilt it from scratch in WPF, leaving the proven engines exactly as they were. It takes longer to build this way, but every control can be shaped around what a collector actually needs.
The result is a new media player with bookmarks and timed tags built in, playlists and compilations, proper tagging controls, and browsers that feel like a streaming service.
This is the pornspective I pictured on day one.
-
Plugins, the television, and what comes next
I have always loved plugin architecture, and Kodi showed me just how far it can take an application. In pornspective, tag information, cover art, content searches and even the AI model are all plugins.
Since then, pornspective.tv has carried the collection into the living room and has become a favourite way for users to enjoy their carefully curated collections. There's always something new around the corner: I'm currently looking at MCP support (which means you can use a local, secure LLM to chat with your collection!) and I'm looking to train more sophisticated models for the AI app. The next release will have a training mode where users can help me by contributing training sets.
Everything you see here has been built by one person with a passion for code and a liking for ... well, you know ... Give it a go, and enjoy.
Give it a go.
A small Patreon subscription gets you the download, your licence key and regular development updates.
- Small subscription through Patreon
- Windows desktop apps
- Standard or portable install
- Your files stay where they are
