Hi,
I know I can add a custom ffmpeg command line, but I don't know which lines the default XDCAM node uses.
People that have NVIDIA GPU can get a lot of improvements over the normal encoding if they use -hwaccel cuvid before "input".
Other people may also get many speed improvements using dxva2, so using -hwaccel dxva2.
I think you should add a setting in the XDCAM50 node called "use hardware acceleration" and enable users to use cuvid or dxva2.
I'd like to add a custom ffmpeg command line that is different, though.
I'd like to add a custom ffmpeg node that has the very same settings as these in the screen (default XDCAM50 node) but also -hwaccel cuvid:
Sure, I can do that, but it wouldn't automatically crop.
Let's say I have a 720x480 flagged 4:3 footage; I should crop 60 top and 60 bottom and the upscale to 1920x1080, but my command line is not gonna do that. I don't know how this is handled by the XDCAM node and I don't know how the "analyze" is handled in the XDCAM node (I mean, for progressive footages and interlaced footages). I could add a bunch of conditional node and do it using Avisynth, but I don't see it as a very practical solution.
In other words, if someone can provide the exact same ffmpeg command line the XDCAM node uses, I can just put -hwaccel dxva2 in front of it and everything is gonna work flawlessly.
Remember that the inbuilt encoders has LOTS of "conditional nodes" to decide how to construct a proper command line to make the source fit your settings. There is no easy or practical way around this. But, luckily for you, FFAStrans will write the exact ffmpeg commands used in a log file if you enable this in the workflow properties -> "special" -> "log to files". Just submit a file and search for the string "_Simple_MediaOutput executing:" in the log file. Here's a full example of an 4:1:1, lower field, NTSC SD 4:3 file coverted to XDCAM-HD422i50:
