ascii-video

Decode the source video with ffmpeg, sample per-frame luminance and edges, map brightness values to characters from a density palette, and composite them onto a pixel canvas with pre-rasterized font bitmaps. Pipe the resulting RGB frames back to ffmpeg for H.264 encoding.

Analyze the audio with SciPy FFT to extract frequency bands, RMS energy, spectral centroid, and beat onsets per frame. Drive value field generators, particle systems, and beat-reactive shaders from those features, then encode frames to MP4 with the audio muxed in.

No GPU is required. The pipeline runs on CPU using NumPy-vectorized effect math and Pillow font rasterization, with concurrent.futures splitting frame ranges across worker processes. A 1080p 24fps frame renders in roughly 100-200ms.

ASCII characters are small bright marks on a black background, so most frame pixels are dark and linear brightness multipliers clip highlights. Use percentile-based adaptive tonemapping with gamma around 0.75 instead of multiplying the canvas by a constant.

Setting stderr=subprocess.PIPE on a long-running ffmpeg encode deadlocks once the stderr buffer fills at about 64KB. Redirect stderr to a log file instead and keep only stdin open for the raw RGB frame stream.

Not all fonts render every Unicode glyph, and unsupported characters silently produce blank cells. Validate each palette at initialization by rendering every character and dropping any that produce empty bitmaps.