an animated Stable Diffusion/optical flow sequence. These drums FREE as one-liner to output wav.

Layer 1: A prompt for stable diffusion is applied but only to the area marked for infilling.
The prompt was "Mechanization. Artful. Outdoors. Photorealism. Hyperrealistic."
The marking the areas for infilling was (a) detect face locations and mark these or if no faces are detected (b) a selection is made with several randomly sized and placed rectangles blurred together.

Layer 2: An intermediate frame is placed between each SD generated frame. The intermediate frame is generated by performing an optical flow between the previous frame and the current frame and then interpolating.

Layer 3, 4, 5, 6: On each layer, the previous sequence has the same thing as before done to it, doubling the length again. This sequence of layers stretches the original video length of 200 images to a whole lot more (5 doublings is 32x).

The sound was generated
The following is the code for the drums. It requires a sound tool named sox. The command tells sox to create the same audio file used in the video, and name it "mecha_drums.wav":

sox -c2 -n mecha_drums.wav synth brownnoise bandreject 250 0.2071q synth square amod 148 square fmod 49 0 0.5 0.125 synth sine amod 3.5 0 0.5 synth exp fmod 8 0 0.0 0.25 synth triangle amod 7 0 0.125 0.85 synth brownnoise bandreject 150 0.2071q synth square amod 146 sine fmod 49 0 0.125 synth sine fmod 1 0 0.5 0.5 synth exp amod 8 0 0.0 0.25 synth triangle fmod 7 0 0.125 0.35 speed 0.8571428571428572 highpass 150 .9q trim 0 10:00 gain -n