Using math [1] we can determine the maximum video size. Luckily our AV1 encoder allows for setting the output bitrate, so with a target of (13360 - 8000) = 5.360k we are on our way. Note, 60% of our file is just audio! Bad audio ruins the entire experience. If it's listenable, the video at its worst will just be a radio play. This is also funny because on a luxurious DVD or 4K Blu-Ray, a single audio track probably won't exceed 5 GB or 12% of the file. 8k is just as small as you can make something like this, until AI-based audio codecs catch on.

Tried many things for the video encode. It was obvious I needed to reduce the heck out of the 1080p input signal. I first applied hqdn3d as a braindead degrain filter. I did the unthinkable and cropped the source movie. You're seeing the middle 80% of the frame, which works pretty good for most shots. Believe me, for how bad this encode looks, adding another 20% of signal to compress in the same size was even less watchable. I almost ran the output at 1 fps, but going up to the luxurious 3 fps allows your brain to interpolate so much more motion and makes it feel more real and watchable. Also, because of how keyframes and video encoders work, it's not like doubling the frame halves the quality per frame. I scaled the video many times, but ended up at 86x64 or a "64p" video. Yep, that's 64p just like you may know "1080p" or "4K" (2160p). This postage stamp is 0.4% of the source 1080p video. libsvtav1 requires at least 64 pixels for width and height, otherwise I'd do 80x60. I used the "g" parameter to reduce total keyframes and hopefully increase quality. preset 2 is supposed to improve quality (0 is max, slowest).

Please enjoy this bit of cultural heritage! And please agitate for much looser copyright. It shouldn't have taken 95 years for this film to belong to the public.

#publicdomainday2026 #publicdomainday #FreeCulture