BLOG:
2022-11-05
So, on the list of cool things I've started to mess around with creating my own library for neural networks. For now I'm playing around with differnt kinds of activation functions I've written. I'm doing it all in rust because it seems like a good way to learn more about the language while doing something interesting with it as well. I'm really trying to avoid using python and pre-built libraries for the moment so we'll see how it goes.
2022-10-17
Figured out gradiant function for modeling contour lines. To represent the *.PLY files I posted about previously I had to take the function and get both partial derivatives to form the gradient. I then found where both partial derivatives are equal to the
desired z-value in the 3d space. This gives me the location of all points along the contour line or level-set.
From there I just have to use OpenGL to make line segments between those points and I have a poly-line thtat approximates the actual contour line.
2022-10-16
Classes have been crazy(in a good way). Have some data visualization going on using OpenGL to render stuff uing *.PLY files which are basicaly csv files to indicate values of points on a mesh (x, y, z, etc). So taking those and then having to setup stuff on each point is somewhat discuraging because it requires so many nested loops that absolutly tank the O(n) time complexity. It's not all bad though, by storing some of the data backinto the file you can precalculate stuff if needed. Adding to the upside I think I can probably leverage some openMP to spilt of the process across multiple cores easily as the Parallel fraction for the processing of vertexs(points) is nothing to scoff at.
Other noteworthy stuff: I've made my way though about ch. 18 in the rust book. Already I can tell that I'm going to want to add some of the methodology for enums(results and options) into my C and even C++ coding if I can. Hopefully I'll be able to scrape out some time this month to rewrite the OpenGL stuff in rust.
2022-02-10
I recently found out about the REALESRGAN(AI image upscaling) project on github. I read through the documentation and decided to give it a try on some videos and images. Low and behold with a little bit of bash scripting it totally works, and I can use my RX580 to handle all the processing. Right now I'm getting around 7.5s per image for rendering time at 1300ish Mhz for the gpu clock. For running video and gif files through it I use ffmpeg to first encode the file into a mp4 then slice it into jpg files for processing. If you want to download my badly written script file, it will be on the links page.
2022-02-06
I've been running some of my servers with FreeBSD jails and using them to run my webservers. I'm still working on figuring out if I want to switch to something else than Nginx(webserver). The next big project I'm taking on is getting a git server up and running. I'm looking at gitea as a option that would let me host my code repos without having to use microsofts github or gitlab.
2022-02-05
Made a pretty cool theme setup for my FreeBSD 13.0 install. I'm running it on a thinkpad T440p with 16GB of memory and dual booted with a install of nixos.
I'll probably upload some screenshots of it later.
