I took a vid from youtube. (I had it already for older project).

Extracted all ~5600 frames.

Converted it from full colorRaster (BMP, xy, pixels, whatever) to 1-bit. (black and white).

Then to SVG which is Vector (Lines with angles, filled geometric shapes)

Then from Vector BACK to BMP but VERY tiny. (28×64 – that’s 28 across by 64 down: A VERY SMALL smartphone!)

THEN back to SVG (making new vectors from tiny ones, which got rid of noise by pushing lines together (got rid of gaps).

THEN from Vector back UP in size.

So that’s my short code to remind me of my process so that I’ll kow what I did a few years from now.

===

Compare the two: Look how much is missing from the one color (black blobs) video compared to this.No colors. No textures. No familiar parts. And yet, our brains can extrapolate a LOT even with lot of missing information.Shannon’s Information theory goes through this in detail. You can understand a message because information is redundant. You get the same information many different ways.So this allows us to reconstruct even when half the keys are broken.https://www.youtube.com/watch?v=-JlBozCbV1Q

====

I’m going to make a side-by-side (original flanked by two versions). I didn’t expect it to go this far but I was really happy at finally understanding thresholding and how significant it is as well as bridging the rastor to vector to rastor divide. (discrete vs continuous in math)

====

Making vector movies from random video. Testing lower limits of information redundancy. Importance of Threshold . Shrinking down to 28×64 for cleaner vectors with some loss of detail.Used Inkskape, potrace, ffmpeg, Excel, batch files, SVG2PNG, and lots of expanding and shrinking and expanding again, from bmp to svg back to bmp back to svg back to bmp, all at different scales and resolutions.Comparing 3 – Vector (SVG) movie, Computer Vision, minimal information, boy meltdown in car

====

There we go. I REALLY picked a STUPID video for this experiment but I didn’t know I’d be doing it on and off for days trying different ideas. But I’m proud of the result anyway.

https://www.youtube.com/watch?v=cgnwylmZYX8

=====

“Information can never be created or destroyed”… supposedly.

I’m testing its limits in my own way.’

—–

Also, I’ve been fascinated by Claude Shannon’s information theory, particularly redundancy and compression, for my whole adult life.

So this relates to: How can a single “wink” contain so much information?

It’s more than just “sign and signal”. There’s a massive compression and decompression of information happening in us all of the time and it’s damn amazing.

——