Original courier noise exercise

Input: courier-envelope-motion.mp4 from the linked authored animation study
No AI model or commercial plugin was run

ffmpeg -i courier-envelope-motion.mp4 -vf "noise=alls=18:allf=t+u:all_seed=42" -an -c:v libx264 -preset medium -crf 18 -pix_fmt yuv420p -movflags +faststart denoise-courier-noisy.mp4
ffmpeg -i denoise-courier-noisy.mp4 -vf "hqdn3d=4:3:6:4.5" -an -c:v libx264 -preset medium -crf 18 -pix_fmt yuv420p -movflags +faststart denoise-courier-filtered.mp4

Run in a separate folder and keep your original input. The filtered output includes a second lossy encode. These values illustrate a controlled exercise, not universal production settings. The three-panel comparison scales each source to 480 by 270 with label bands; full-frame PNG stills use zero-based frame 48 at two seconds.