video-to-gif

A conversion skill that turns a video clip into multiple GIF variants with different quality and size tradeoffs, then presents a comparison set so the user can visually pick the best result. It fits the common situation where someone wants a GIF from a video but is unsure of the right parameters, since GIF quality trades file size against color accuracy, smoothness, and resolution. Rather than guessing a single output, it sweeps a parameter space and produces a sorted comparison table showing file size, FPS, width, and colors for every variant. FFmpeg and uv are required prerequisites, and gifsicle is optional to enable lossy compression variants.

The default run uses the full preset, generating about 18 variants across three FPS options (10, 15, 20), three widths (480, 640, 800 pixels), and two color counts (128, 256), written to an output directory beside the input. Four presets tune the tradeoff: full for broad exploration, minimal for a quick four-variant comparison, lossy for the smallest files via gifsicle levels, and quality for the best visuals with higher resolution and bayer dithering. Common options let you trim the source with start and end times, redirect the output directory, and override the FPS, width, color, and lossy value lists directly.

Guidance on choosing a result suggests starting from the smallest files and checking acceptability, watching for color banding (raise colors to 256 or use bayer dithering), verifying smoothness (raise FPS to 15 or 20), and confirming clarity (raise width to 640 or 800), noting that around 640px, 15fps, and 256 colors is a common sweet spot for screen recordings. It also uses FFmpeg two-pass palette generation for markedly better quality than single-pass, automatically skips widths larger than the source, and notes that lossy compression can cut file size by 30 to 70% with minimal visual impact at moderate levels.