XataxataX
LightNews LightNews
xataxatax.bsky.social
XataxataX
@xataxatax.bsky.social
38 followers 34 following 11 posts
Ancient Egyptian God of Frustration. Yarn Surgeon. ザタクサタクスかタクシーがいいです。
Posts Replies Media Videos
xataxatax.bsky.social
Update on VidComp to show how it varies across compression quality and also what happens when frames show drastic differences (specifically not matching.) Varied framerates cause seizure inducing flashing, so I'm skipping those. More #HanamaruYouchien for quick and easy comparison.
HEVC Q26, loss of detail is very minimal. Colorized timeline Intensity Graph at the top. Left Column in descending order: Source video, compressed video, colorized difference. Right Column: Greyscale difference, frame counter HEVC Q40, loss of detail is not highly visually apparent to the naked eye, but readily apparent to the difference calculation. Colorized timeline Intensity Graph at the top, notice taller lines indicating a higher difference between frames. Left Column in descending order: Source video, compressed video, colorized difference. Right Column: Greyscale difference, frame counter 2 Different Episodes Colorized timeline Intensity Graph at the top, note that the beginning (Intro) is extremely similar, but when the actual episode starts the difference is instantly apparent. The multicolored lines show peak pixel difference (the higher color, typically red/magenta/white), and average pixel difference (the lower color, typically green/yellow/blue) Left Column in descending order: Episode 2, Episode 3, colorized difference, see right column description for more information. Right Column: Greyscale difference, this composites the two frames and generates what is functionally a greyscale heatmap of the difference between pixel intensities between the two frames. Because it's ideally designed to be looking at two variations of the same video, it should typically be mostly black, this overlap occurs specifically because the frames are completely different. Frame counter
March 21, 2025 at 8:28 PM
xataxatax.bsky.social
Finally achieved the desired output. When it comes to compressing video, minor quality loss can be hard to notice. I used Python and OpenCV to compare pre and post compression frames to detect quality loss and present the data visually. End credits for #HanamaruYouchien for the test run.
I used Python and OpenCV to analyze the pre and post compression of the video. Timeline intensity and tracker bars at the top, left images are high quality, compressed and colorized diff, the main image on the right is the greyscale difference to show the intensity of pixel differences. This helps find where even minor information is lost, when it may be undetectable to the human eye.
March 21, 2025 at 5:46 PM
xataxatax.bsky.social
What is time anyway?
December 1, 2024 at 3:22 AM
xataxatax.bsky.social
Converting a 2 drive RAID 1 array to a 3 drive RAID 5 array is a slow process.
November 30, 2024 at 7:30 AM