On 20190921 08:50:20, stan via users wrote:
On Sat, 21 Sep 2019 13:59:23 +0800
Turritopsis Dohrnii Teo En Ming <teo.en.ming.smartph...@gmail.com>
wrote:

What I mean is combining multiple smaller 4K video clips into a
bigger, single 4K video and at the same time, allowing me to add
layers of watermarks into the 4K video.

Disclaimer:  haven't done this, not an expert

If you are concatenating several segments of 4k video, that should be
fast because there is no encoding or decoding.  It would be best to do
this as a separate step, creating the final concatenated video as an output.
untested ffmpeg command to concatenate all the files

ffmpeg -i [list of input files] -codec copy [concatenated output file]

yields a single file of a name you choose containing all the input files
in the order you list them

Adding the watermark is the piece that takes so long, because the
video has to be decoded, every frame filtered to add the watermark, and
re-encoded before writing to the output.

FWIW I was playing mpeg/mjpeg/avi for watermark and video, two channels, and cross fading or cross squeezing with live video for branding and news room work back around 2005 live for HD with Matrox DSX cards on Windows. Back then the Matrox people were unable to make nearly enough performance out of their cards with Linux so they abandoned the effort. That may have changed, especially with Windows 10 putting itself in the way of clean honest video work at every step of the way. So this is just a benchmark of what WAS possible. nVidia based systems were doing approximately as well back then. In calculating what you want to do and how to do it be sure to calculate bus usage into the picture. With all the program I had authored was doing bus transfers became an issue as did, it turns out, the bus transfers between CPU chips in a multiple CPU system. Everything you can think of needs to be figured out with simple arithmetic to make sure you are not demanding more out of your system than it can deliver. It's not just a software issue. It's GPU cycles, CPU cycles, OS overhead, bus speeds, CPU to CPU bus speeds, and probably other more exotic issues such as page sizes that matter.

Apple and Windows are the places with the really good software for this last I knew, which is some time ago. I'm retired now. Linux at the time was extremely anemic in comparison. You MAY have to embrace one of the devils.

{o.o}
_______________________________________________
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org

Reply via email to