so your current workflow is like: A/V decoder-> Watermark-> H264 Encode?
If so, we need to get rid of the A/V decoder and Watermark, so the only way is to encode using a commandline processor or custom proc.
But when we do this, we loose all the smart audio mapping and color space conversions and such that ffastrans calculates in the encoder processors.
So the question for you is if the INPUT files do always have the same audio configuration (e.g. 8 mono tracks or 2x4 stereo tracks) and what exactly is your watermark image, is it a full frame with same height/width as the video or is it something smaller? -If smaller, where you want to place it, top left?
Thanks.
The vídeo are múltiple resolucions and meny varius codecs.
I think A/V previos is necesary.... but weatermark and h264 in only procesos its not posible in h264. Really?
The CMD procesos not control error and not like to control the workflow.
taurojo wrote: ↑Sun Feb 20, 2022 10:21 am
... but weatermark and h264 in only procesos its not posible in h264. Really?
Well it all is actually done in a single process but when the A/V decoder is used, there is one step for "indexing" added. On top of that, your current method of watermark (builtin=avisynth) is unfortunately massively slowing down the encoding process. This is because the used avisynth image overlay filter has a software design issue, making it slow to overlay the pictures and therefore slowing down the whole encoding process.
The fastest you can get is to get rid of the A/V decoder one way or the other.
Maybe @FranceBB can help here and suggest an avisynth filter that's (mucH) faster than the one we use currently by default?