So, if you know one frame, you just have to store the difference with the next one, and re-compute it on the fly when playing, and then do the same for the next frame, and so on.

Instead, video files are compressed, using the fact that in a real video, the next frame remains close to the previous one. For a one minute HD video at 30 frames per second and 8 bit color depth, the file size would be more than 10 Gb. You can take a series of JPGs and concatenate them into a video file, for instance, using FFmpeg:įfmpeg -framerate 30 -i frame_%03d.jpg -c:v libx264 -pix_fmt yuv420p output.mp4īut storing each individual frame of a video is an insanely inefficient solution. # A Quick Introduction to Video Encodingīasically, a video is a sequence of still images, called frames. To get an understanding of the principle, one must know a little bit about video encoding. I found that amazing, and wanted to try it myself. And he plays with that to merge several pictures in evocating ways.Įxtracted from Avant l'effondrement du Mont Blanc (2020) He encodes videos in such a way that he loses data without corrupting the file, resulting in mesmerizing video artifacts. In the video, the French artist Jacques Perconte exhibits his work, that relies on data moshing. This process can lead to some very cool and unique glitch effects that have been used to various degrees within different mediums to create an interesting and unique effect.I once saw a very intriguing YouTube video: (), on (), a show produced by the Franco-German television channel () and dedicated to emergent forms of art. What results is a new video moving based on another videos directions or an image from the same video where the pixels go in directions they’re not suppose to.

What datamoshing does is it gets rid of the new full picture frames and instead only keeps the frames that tell the pixels where to go. This process allows for much smaller file sizes for videos as well as allowing datamoshing. So what many formats do is simply tell the old pixels on the screen where to go to make the new picture instead of creating a whole new picture.

If you think about it a lot of the frames in a video are just a very similar picture with slight differences. So what the developers of modern video formats did was only have full pictures when absolutely necessary. Most videos very between 24 and 60 frames per second and as you can imagine having 60 pictures for only one second of a video would take up a huge amount of space. To achieve this goal they have developed some clever techniques that can look very cool when they don’t work as they should. Modern video formats have been designed in such a way as to minimize the storage they take up while maximizing things like resolution and frame rate.

Videos for datamoshing