On Sun, Aug 29, 2021 at 11:40:48AM +0200, j...@dodin.org wrote: > Le 29/08/2021 à 09:20, Vincent Pinon a écrit : > > > I agree the frame number is a more precise indicator than milliseconds, > > more precise certainly not, more friendly may be.
I'd make the claim that frame numbers are more precise in the sense of less potentially ambiguous, and much more difficult to accidentally round up or down a frame somewhere in the pipeline. (With timecodes, Kdenlive's rounding behavior must always be internally consistent and consistent across application versions, as must MLT's, and they must stay consistent with each other, etc.) > But don't forget that when using multitrack from various sources, the frames > are not in sinc... this may be very visible for lipsync >From what I understand this isn't related to frames vs. timecode: if there's a >difference between source video frame rate and Kdenlive profile frame rate, >the in and out points are (were?) written in terms of output frames, not input >frames. The assumption in either case is that Kdenlive/MLT will resample the >input video (as a simple example, if you have a 30 fps source video in a 60 >fps project, each frame will be duplicated once and the duration will stay the >same). Tom
