Skip to content

Still Image Conversion

Jump to bottom
George Stoyanov edited this page May 28, 2018 · 4 revisions

Still image export

Picture Type check

First you need to confirm that your video sequence you want to export into images is consisting of only I-frames. You can do that using the ffprobe utility:

ffprobe -v quiet \
        -of csv=p=0 \
        -read_intervals %+5 \
        -select_streams v \
        -show_frames \
        -show_entries frame=pict_type \
         /path/to/input_file.xxx

This command will show you the type of the frames for the first 5 seconds (read_intervals) of the /path/to/input_file.xxx video sequence. If you see only "I" you can proceed if not, then you should convert your sequence using one of the intra-frames codecs supported in FFMPEG like FFV1, etc.. You can find a list of the most popular codecs at this link.

FFV1 conversion

You can convert your video sequence with FFV1 using the following syntax:

ffmpeg -i input_file.xxx \
        -c:v ffv1 \
        -level 3 \
        -threads 8 \
        -coder 1 \
        -context 1 \
        -g 1 \
        -c:a copy output.mkv

Here I am converting the input_file.xxx using FFV1 using GoP (group of pictures) 1, which will ensure that my output video sequence is consisted of only I-frames. The explanation of the rest of the parameters you can find in the table below:

Name FFmpeg argument Valid values Comments
Coder -coder 0, 1, 2 0=Golomb-Rice, 1=Range Coder, 2=Range Coder (with custom state transition table)
Context -context 0, 1 0=small, 1=large
GOP size -g integer >= 1 For archival use, GOP-size should be "1".
Version -level 1, 3 Select which FFV1 version to use.
Threads -threads integer >= 1 The number of threads to use while processing. Adjust this to match how many of your available CPU cores you want to use.
Slices -slices 4, 6, 9, 12, 16, 24, 30 Each frame is split into this number of slices. This affects multithreading performance, as well as filesize: Increasing the number of slices might speed up performance, but also increases the filesize.
Error correction/detection -slicecrc 0, 1 0=off, 1=on. Enabling this option adds CRC information to each slice. This makes it possible for a decoder to detect errors in the bitstream, rather than blindly decoding a broken slice.

You can find more information at the official FFMPEG FFV1 encoding guide

Still Image Export

You can export the video into a single screenshot or multiple screenshots using the following command:

  • ffmpeg -ss 0 -i input.mkv -c copy -vframes 1 -y -f mjpeg first.png - extracts the first frame in a video file. You can define the timestamp at which you want to export the video frame. For example for 50fps video the first frame will be at -ss 0, the second at -ss 0.02, the third -ss 0.04 and so on. For 25fps, it would be 0, 0.04 and 0.08. You can find the duration of the frame dividing 1 by the number of frames per second, which you can find out using the following ffprobe command: ffprobe -v 0 -of csv=p=0 -select_streams 0 -show_entries stream=r_frame_rate input_file.xxx
  • ffmpeg -i input.mkv -c copy -f mjpeg lossless%05d.png - extracts all the frames from the input video and creates png files called lossless00001.png, lossless00002.png and so on, the %05d defines the length of the preceding zeros.

References

FFV1 Encoding Guide
List of Lossless Codecs

Clone this wiki locally