Use -qscale:v to control quality

Use -qscale:v (or the alias -q:v) as an output option.

• Normal range for JPEG is 2-31 with 31 being the worst quality.
• The scale is linear with double the qscale being roughly half the bitrate.
• Recommend trying values of 2-5.
• You can use a value of 1 but you must add the -qmin 1 output option (because the default is -qmin 2).

To output a series of images:

ffmpeg -i input.mp4 -qscale:v 2 output_%03d.jpg

See the image muxer documentation for more options involving image outputs.

To output a single image at ~60 seconds duration:

ffmpeg -ss 60 -i input.mp4 -qscale:v 4 -frames:v 1 output.jpg

To continuously overwrite/update/save to a single image

Use -update 1 image muxer option. Example for once per second from a live streaming input:

ffmpeg -i rtmp://input.foo -q:v 4 -r 1 -update 1 output.jpg

Also see

• FFmpeg FAQ: How do I encode movie to single pictures?
• FFmpeg Wiki: Create a thumbnail image every X seconds of the video

Output the images in a lossless format such as PNG:

mkdir stills
ffmpeg -i my-film.mp4 -vsync 0 -f image2 stills/my-film-%06d.png

Then use another program (where you can more precisely specify quality, subsampling and DCT method – e.g. GIMP) to convert the PNGs you want to JPEG.

It is possible to obtain slightly sharper images in JPEG format this way than is possible with -qmin 1 -q:v 1 and outputting as JPEG directly from ffmpeg.

If you want to extract only the key frames (which are likely to be of higher quality post-edit) you can use something like this:

ffmpeg -skip_frame nokey -i my-film.mp4 -vsync 0 -f image2 stills/my-film-%06d.png

The -vsync 0 parameter avoids needing to specify the frame rate with -r and means all frames in the input file are treated as, um, a frame.