pornspective
patreon
SDK, Faster Thumbnails and what's next...
🕑 Added 2022-10-15 21:34:25 +0000 UTCHi Guys,
Thought I'd update you with what I've been working on this last couple of weeks. As you know I've now released v2 in .NET 6.0 and at the same time did some housekeeping with the code. However this was just enough to get a good solid build and update some libraries.
Since then I wanted to continue work on the plugins and bring these over to the v2. Whilst I was doing it I've been commenting on my code in the plugin base library, as I plan to release this part of the source code in an SDK. This will allow others with .NET skills to contribute to creating plugins for the application. I'm hoping a starter for the SDK will be in the next release, including a testing project to ensure plugins are written well enough for v2 to use. So if you fancy having a go at writing plugins, then watch this space!
During the review of the plugin code, I decided that I might be best to now bring AI into the main code base (it was previously on v1 branch). At the same time I've been updating all my namespaces in the code to really tidy things perfectly. So I now have a single codebase that builds v2, AI and the SDK. Happy days.
Onto some really good news. A few have asked me about the thumbnailing engine in the application which is used by both v2 and AI. Its always worked well but I've also been conscious that the way each app calls the engine is through a mechanism which could be improved. Well, now I've enhanced the thumbnailiing code with a much better way to call and use the engine. It results in a MUCH FASTER! thumbnails. Early tests are showing this is 3 or 4 times faster than the existing one, and I'm experimenting with parallel coding to use more CPU cores to get even faster speeds. This is obviously going to improve scanning location times and more importantly the AI tool which relies heavily on lots of thumbnails being generated! I'm hoping to get this out to you guys in next two weeks.
What next after that?
Well I plan to move the AI to WPF like v2 is and then properly onto .NET 6.0. I need to replace the remote access between AI and v2 though as Microsoft have stopped WCF in .NET 6+, instead I'll need to use another mechanism. So a little bit of work there. Finally I've noticed that the AI model I use now has newer, faster and more accurate detection so I'll certainly look to update this in the next two or three months.
As always thanks for the support guys!
