Closed ilia3101 closed 5 years ago
Do you have some information how to do that? (example commands) Now our command is this:
... -vf scale=w=1920:h=1080 ...
for 1920x1080
I was going to try and do this earlier, but qt code is still something I don't understand. But the way to set scaling method it seems is this:
-sws_flags scaling_filter
where scaling filter names are listed here: https://ffmpeg.org/ffmpeg-scaler.html
Thanks, I'll have a look at this.
I think what you don't understand is not Qt, it is my puzzle game attaching all snippets of the ffmpeg command together...
From Qt side, there is only QString( "..." )
, QString( "...%1..." ).arg( variable )
and string.append( "..." )
. I think you understand this :) Most of the other code is standard C++. But please ask, if you don't understand something!
Please test latest commit. sws_flags=sinc is no the only standard. To make it switchable: what options else from the list do you like to be supported? If you want to see the full ffmpeg command, comment in line 2010 of MainWindow.cpp.
Good... and more to test for you. I implemented a combobox in export settings where you can choose between the three types you mentioned above. Hope it works ;-) Edit: bicubic was default. At least I see a difference between bilinear and sinc, so it seems to do something.
thank you!!!
@masc4ii 2880x1080 clips look beautiful upscaled to 3840x1440 with sinc! A big improvement over previous scaling.
I think lanczos should be an option too, I think its good for scaling.
Yes lanczos is really good.
No problem... I'll add it. Edit: done. Edit2: something else?
Can we close, or do we need something more?
I don't know what the current one is but I'd like to be able to choose sinc cubic linear etc. I find when I upscale a small amount with editing software it reduces the detail a lot, so I'd like to do it with sinc filter out of MLV app.
I didn't check that the default filter in mlv app and haven't compared yet but I think it should be an option as sometimes different upscale methods are useful.