/vt/ - Virtual Youtubers

>>65321563
>boomer hands
you can't fool us

>>64938647
Be mad all you want, you are barking up the wrong tree.

That Toaster? All me!

>>50418898
MumAI isn't real, she can't hurt you

my crabby moomers...

>>49312905

>>48716854
>weird eyes
>hairclip
>resolution
a toaster did this..?

>>48090310
>They actually populate in the /streams
As expected, Youtube being a hodgepodge of inconsistencies.
>allow sound-containing webms
isn't that only /gif/ anyway that allows sound? Even in that case might be worth to include actual values. I don't know what the defaults are for libopus but in my experience defaults are usually not exactly good in any capacity. Would have to check in this case though, don't know defaults offhand.
>Crop then scale intentional
Alright then, just seemed a little odd and wasn't really sure if that was an oversight or intentional. Might be worth pointing out the order early to make sure the crop won't get upscaled by accident.
>Padding
I'm not sure if that is sufficiently pointed out in the ffmpeg documentation but the inacurrency should only be affecting the use of positional arguments as input options. i.e. -ss 1234 -i IN_File and should not make a difference (for copy) if used as output options. It's going to make a large difference for the actual encoding process, that much is true...but that's what the pre-cutting is taking care of.

>Constrained vs Constant
This might be a point of divergent approaches.
Situations heavy in movement will have a huge impact on image quality, this is already a thing with just the incoming data from the stream (OBS/NVENC/YT all take a whiff at it) even before we get to put a hand on it.
As we do have a hard filesize limit our goal should be not to hard min-max but to make sure we are using the available limits the best way we can. With that fixed max size we are down to either purely Bitrate based(ABR/VBR) or a hybrid approach with Constrained to make sure are not going to exceed the file size limit. Doesn't really matter how much better the results might look if we can't use them because they are too big.
I don't know your exact testing examples but yes constrained might put an additional limit to quality in some cases over pure constant quality - but that is down making sure we do not hit the filesize limits.
As long as we do not hit the bitrate limit constrained works identically to constant and at least my own tests do show as much. Some tweaking might be possible in setting min/max bitrates but that is going to exceed the scope of a simple soundpostscript for Hooties.
If we go down into the swamps of very long soundposts where bitrates drop to stupid low levels things might look different yet again but at that point you should rather consider if you really need a soundpost that long in the first place.