[FFmpeg-user] systematically encoding using x264

[Frank Tetzel]

> I am looking for specific guidelines to encode many  sequences of
> image to a video using x264 encoder. The image resolution and frame
> rate could be arbitrary (some sequences are 256x256 others could be
> 4kx4k or more), usually with images 1:1 aspect ratio. I would like to
> create multiple resolution/bitrate videos from these sequences which
> could be served on multiple devices as well as downloaded. The
> encoding time is not a very big concern but quality is a huge
> concern, as these sequences have scientific content. 
> 
> I am seeking some pointers on this and if there is someone out there
> who may be interested in collaborating that will be great. I'll be
> willing to open a set of sequences which could be used for tests,
> benchmarks, quality comparisons etc.

Hi,

so you want to create slideshows with different quality settings
including different resolutions?

This page describes how to create slideshows:
https://trac.ffmpeg.org/wiki/Create%20a%20video%20slideshow%20from%20images

more information about the image format:
http://ffmpeg.org/ffmpeg-formats.html#image2-2
important:
"The size, the pixel format, and the format of each image must be the
same for all the files in the sequence."
If that's not the case you have to convert the images beforehand, e.g.
with imagemagick.

The scale video filter is described here:
http://ffmpeg.org/ffmpeg-filters.html#scale-1

An example could be:
ffmpeg -r 1/5 -i img%03d.png \
-vf 'scale=256:-1' \
-c:v libx264 -preset veryslow -tune stillimage -r 30 out.mp4

As you want to append sequences with different framerates, you need to
concatenate them. Something like that:

ffmpeg -r 1/5 -i first/%03d.png -r 1/10 -i second/%03d.png \
-filter_complex '[0:v] scale=256:-1 [v1]; \
[1:v] scale=256:-1 [v2]; [v1][v2] concat, fps=30' out.mp4

Regards,
Frank.