filifa
/
slitscan
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

add readme

This commit is contained in:
filifa 2024-10-20 21:47:56 -05:00
parent 8e6d0e6672
commit 68e688a3bb
1 changed files with 47 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

47
README.md Normal file
View File

@ -0,0 +1,47 @@
# slitscan
slitscan is a script inspired by [this Mastodon
post](https://fed.qaz.red/@elithebearded/113070315634312301). It takes a video
file as input and allows you to "transpose" the frames of the video to view
along the height or width axes instead of the "time" axis you normally watch a
video along.
## Usage
Say you have a video file `foo.mkv`. To transpose along the height axis and save
to `bar.mkv`, run
```
python slitscan.py foo.mkv bar.mkv --height
```
Similarly, to transpose along the width axis, run
```
python slitscan.py foo.mkv bar.mkv --width
```
## Details
### Uh... what does this do?
Check the linked post for another explanation and examples.
To elaborate, you can think of a video as a 3D array of pixels. Say the axes of
the array are the frame, height, and width, in that order. Then the entry (24,
500, 900) gives you the pixel in the 499th row and 899th column of the 23rd
frame of the video (we're 0-indexed). You can imagine "watching" a video as
just iterating through the frame axis of that array and displaying each 2D
slice.
This program transposes that 3D array so you're instead iterating over the
height or width axis and displaying *those* 2D slices. For instance, if you
transpose along the height axis, then the first frame of the output video
consists of the top row of pixels of every single frame of the original video.
The second frame is then the second row of every single original frame, and so
on. Instead of viewing each original frame one by one, it's like you're viewing
them all at once, but only one horizontal line of pixels at a time.
### But why?
The author of that Mastodon post mentioned that he wrote a bash script to do
this that took a *a couple of days* to run for a short clip. I wanted to see if
I could make it faster (and I did - this runs on the order of a few *minutes*
for similarly short clips). It might be possible to make this script even
faster, but I couldn't figure out how without the script using too much memory.
### Is this even useful?
I would love to find out that it is!