/mpv/ general. Open source, cross-platform media player.

File: HLv1qmj.png (3KB, 534x196px) Image search: [Google]

3KB, 534x196px

Still no shinchiro build... The page says it was updated but there is nothing new. Sigh.

>>61270877
>Noisy audio like broken audio
Oh it seems there is some problems with audio and will have to wait for the fix.

>>61270916
>windows in 2017 with abusive update auto-install
>don't want audio problems

>>61271157
You think its related to Win 10? Yeah its probably is. I dont understand why people use it, its garbage.

What is default scale radius for ewa_lanczoss in mpv?

>>61271176
There were a lot of audio issues with recent M$ updates. I don't really know if it's the case here. However it was a massive pain for the users to rollback and most of them had to wait for another updates.

linked mkv support, chapters, volume control, settings menu, and doom9 unbanning when?

>>61271960
>linked mkv support
What is dis!

>>61271960
eyes surgery, brain tumor ablation when? no money?

>>61271911
>What is default scale radius for ewa_lanczoss in mpv?
3
https://github.com/mpv-player/mpv/blob/master/video/out/filter_kernels.c#L364

>>61272041
Thank you! 3 = 3 taps, right?

Few questions
does it support android/ios?
Why not just vlc?

>>61272085
>does it support android/ios?
It does. This PR will btfo mx player:
https://github.com/mpv-player/mpv/pull/4580

>>61272085
I think it has Android version. Better than vlc quality wise and no bloat.

>>61272099
How do I play a video and let it play audio in background when I move to another app?

>>61272154
What am I looking at?

>>61272094
this one? https://play.google.com/store/apps/details?id=is.xyz.mpv

>Noisy audio like broken audio #4582
REEEEE somewhere answer him!

Which is correct

audio-device='wasapi'

or

audio-device=wasapi

?

Shinciro release the build! Its probably Win 10 issue! AAAAA~

Are there any good config file templates for low-end systems?

Wow. haasnsoft is pretty good for low quality low res videos. Good alternative to nnedi3 if you dont have resources for it.

>>61272763
Yes much better than ewa_lanczos + SSimSuperRes (use them only for really good 720p videos to 1080p screen)

Woooooooooooooooh!
https://github.com/bjin/mpv-prescalers/commits/ravu

>>61272298
Prototype for segment list/editor

>>61272850
>Yes much better than ewa_lanczos + SSimSuperRes (use them only for really good 720p videos to 1080p screen)
I agree! haasnsoft is actually pretty good for 720p -> 1080p anime too. haasn is a smart guy, this upscaler suits almost ALL anime needs!
>>61272871
Oh my! Why bjin is so cool?

>>61272871
Its his original work, right?

File: ravu.jpg (49KB, 800x640px) Image search: [Google]

49KB, 800x640px

>>61272871
>>61273053
Basically superxbr with ringing

File: superxbr.jpg (44KB, 800x640px) Image search: [Google]

44KB, 800x640px

>>61273205

>>61272083
It means radius 3. Define “taps”

Fix the damn audio already.

>>61273205
Did you read the comment he wrote?

>>61272154
>_<

>>61273205
>>61273215
Could you use 720p or 1080p screenshot from a movie or youtube video to compare them with less than 2x upscale?

>>61273543
Stop being like that. You know full exactly what it means. Lots of people call it taps.

>>61273205
>>61273215
RAVU is inspired by RAISR (Rapid and Accurate Image Super
Resolution). It adopts the core ideas of RAISR but did the
interpolating in a way similar to Super-xBR.

Don't use it, at least not in the current form. It's still a WIP,
and insanely slow at this moment.

The weights are trained with only a few random wallpaper/screenshots
files that I could find in local disk. They probably won't work well
for all kind of videos.

>>61273608
>You know full exactly what it means.
No I don't, which is why I'm asking

>>61273205
>>61273215
>>61273622
Looks pretty good IMO especially considering its early untrained WIP.

File: 55062635.jpg (31KB, 499x451px) Image search: [Google]

31KB, 499x451px

>wake UP
>still no new build
ADWJDAO:IJDamowhdo

>>61273547
No you.
Opensource brotha!

>>61273547
What the fuck is wrong with audio? I didnt see any changes to the code? Wat happened?

>>61273679
> ao_wasapi: drop use of AF_FORMAT_S24
this commit might break wasapi

>>61273704
FUGG.

>>61273547
fix your OS

>>61273601
The only difference I noticed compared to ewa_lanczos is there is almost no aliasing with ravu.

It is bad that the frame timings(red) are all red?

>>61273811
No, red=good
Its only logical

>>61273811
Yes, robin...

Anyone miss NGUfag?

>2017
>Choose your CNN poison!
>RAVU, NGU, NGU-AA, FSRCNN

>>61274657
No!

>>61274675
>NGU
Never ever. Not like it matters anyway.

The new build is out.

>>61275333
FUCK YEAH

File: 49602154.jpg (49KB, 438x440px) Image search: [Google]

49KB, 438x440px

>>61275333
Thank you shinchiro! If not for your builds id never switch to mpv from mpc-hc+madVR+xysubfilter combo!

What are pros and cons of blend-subtitles?

>>61275492
mpv = MPC-HC + madVR + XySubFilter + ReClock.
Really makes you think.

>>61275619
Impressive!

File: test.webm (2MB, 1280x720px) Image search: [Google]

2MB, 1280x720px

>>61275333
It broke nnedi3 for me.
I'm using the lastes version of nnedi3 too.
Specs:
RX 480 8GB
i5 6600
16GB RAM

*last version

>>61275714
lots of haasn's commit this week so I'm not surprised if it broke something. You can try dxinterop if it works for you

>>61275714
update nnedi3. Last update was 2 days to adapt new changes in mpv

>>61275933
I'm using that update

File: 1479039307504.jpg (77KB, 409x409px) Image search: [Google]

77KB, 409x409px

>>61270828
>no ngu

File: werks for me - Copy.jpg (454KB, 1920x1080px) Image search: [Google]

454KB, 1920x1080px

>>61275714
Werks for me. Latest everything. Win7, dxinterop. Check your opengl-shader chain, they changed expressions. My old one didnt work.

>>61276119
Example of the new expressions?

>>61272154
I feel like you're going a bit crazy over the features. Why don't you have a public repo for this btw?

Just tested new nnedi3 with haasn improvements vs old one. 32 neurons. 720p ->
1080p.

Old frame times: 12250
New frame times: 11600

Nice :)

>>61276151
opengl-shader
opengl-shader

https://github.com/mpv-player/mpv/commit/91583fccac85880ad2199f50463c095560791d56

>>61276265
it got reverted back to opengl-shaders

>>61270828
>tried it
>it's ass.

I used IINA on macOS but now I'm forced to use Windows. Is there any decent frontend for mpv?

File: 1460093428567.jpg (28KB, 250x250px) Image search: [Google]

28KB, 250x250px

>>61276429
>Is there any decent frontend for mpv?
mpv

>>61276435
Well, is there any decent frontend for the mpv frontend for mpv?

>>61276448
mpv

>>61276448
mpv

>>61276455
>>61276463
Well, is there any decent frontend for the mpv frontend for the mpv frontend for mpv?

>>61276429
There are front-ends but they are hard to build on windows and pure and utter shit.
I'm afraid IINA is pretty much the only decent mpv front-end in existence.

>>61276483
No

>>61276483
mpv

>>61276500
>>61276523
Aight thanks, I'll use mpv. Sounds cool.

>>61276533
shit wrong board

Why is this objectively better than MPC-HC?

>>61276574
Why not.

>>61276933
It doesn't offer any notable features and can't play audio files.

>>61276964
then how am I playing audio files with mpv right now?
also it has youtube-dl support unlike mpc-hc

Did another test with haasn improvements/updated nnedi3 64. 720p -> 1080p.

Old: 26850
New: 23770

Nice-nice!

When is mpv v0.26.0 going to be released?

>>61277045
JUST WAIT™

What's your opinion about ravu and mpv-fsrcnn?
Do you think they will beat ewa_lanczossharp or nnedi3?

>>61277020
Which GPU?

>>61277164
>Do you think they will beat ewa_lanczossharp or nnedi3?
ewa_lanczossharp yes, nnedi3 maybe.
mpv-fsrcnn is dead I think.

And another test of haasn/nnedi3 improvements. nnedi3 32 neurons 8x4. 720p -> 1080p.
Old: 12100
New: 11450
Thats average render times. Both configs identical. Tested on the same source.
>>61277181
AMD RX 470.

>>61277211
Will ravu be faster than nnedi3 (16, 32, 64)?

>>61277244
Can someone calculate the benefit in %? Im stupid.

>>61277244
You should include FPS for your tests too.
There are a lot or 50 and 60 FPS videos on youtube.

>>61277244 Anon here.
I dont know what the fuck is going on, but i tested ewa_lanczossharp in the same scenario.

Old: 910
New: 3200

What the heck :^)

>>61277322
no amount of meme shaders can make youtube videos not eyecancer

>>61277360
Which GPU? Is it a laptop?

>>61277360
ewa_lanczos is slower for me as well (~5%, nvidia)

>>61277366
It's a streaming service! What do you think? Good quality is expensive!

>>61277376
AMD RX 470. Regular desktop GPU.

>>61277398
new nnedi3 is also slower
old - 17700
new - 18200

File: 94597060.png (3KB, 236x176px) Image search: [Google]

3KB, 236x176px

>>61277360 Im here again. Ookay... The last comparison was current/latest build as of today vs build from 11.06. I just made a comparison to 1 week old build 01.07 and here are the results.

Old: 4150
New: 3200

So its either there were some major changes since 11.06 build that killed performance, or stats.lua as of june shows incorrect stats.

910 -> 4150 -> 3200. H-how come...

How exactly do you train a neural network?

>>61277520
New stats.lua is probably the culprit.

>>61277558
stats.lua doesn't produce any stats or information. It just formats and displays them. The values are all provided by mpv.

>>61277544
Like a pokemon?

If I use KrigBilateral, what should I use in cscale?

File: am i retarded - Copy.jpg (534KB, 1891x622px) Image search: [Google]

534KB, 1891x622px

>>61277520
>>61277558
>>61277568
Here what i get. Maybe im retarded and watch in the wrong place?

Wasn't haasn improvements nvidia only?

>>61277708
I dont think so. It is just that he doesnt have AMD GPU and cant test it. Im defenitely getting huge speed boost with latest build compared to 1 week ago, but they both much slower that junes build if im not being retarded and looking in the right place in >>61277679

Well the way mpv measures frame timings changed significantly.
Also, haasn made some optimizations that have a chance to backfire on certain hardware/drivers.

>>61277679
Did you check the detailed stats to see what's taking so long?

>>61275714
What did you use to record your desktop?

File: 2017-07-08_23-02-40.webm (2MB, 1172x690px) Image search: [Google]

2MB, 1172x690px

This (aspect ratio constraining) was such a goddamn pain in the ass to figure out properly (I'm shit at this sort of "geometry" maths), but I had to do it. And it's swell.

>>61276206
Answer for both statements; because I'm doing this for myself. When it's :ok_hand: I will repo it.
Deeper reasons lie within my views on sharing code and getting things in return, but those are boring sad topics.
However, I can condense it thus; I want to make this so good few if any will have anything to add or complain about (yes, unrealistic goal, but at least feature wise) - I do not want to make this a "group effort" until after I've decided yes, this is it. I'm just doing me now.

File: 1476562281836.jpg (46KB, 444x453px) Image search: [Google]

46KB, 444x453px

>>61277809
RELEASE THE FR*CKIN' SCRIPT!!!!!!!!!!!!!!!!!!!!!!!!11

File: help - Copy.jpg (709KB, 2234x824px) Image search: [Google]

709KB, 2234x824px

>>61277758
>>61277758
Both scalers themselves are much faster with haasn changes. Too bad that the June build doesnt have this detailed page functionality. >>61277679 Is as much info as i can get.

>>61277797
AMD's ReLive.
It seems this anon >>61276265 is right, now you have to use this expression for the shaders to work:
Example:
# Shaders
opengl-shader="C:\Program Files\mpv\shaders\KrigBilateral.glsl"
opengl-shader="C:\Program Files\mpv\shaders\SSimDownscaler.glsl"

nnedi3 still gives me a blue screen tho.

>>61277882
>nnedi3 still gives me a blue screen tho.
You need TODAYS mpv build and latest shader files. I had this blue screen error before TODAYS build came out.

>>61277809
>Deeper reasons lie within my views on sharing code and getting things in return, but those are boring sad topics.
You're okay with benefiting from using free software like mpv though?

>>61277903
Have you read my previous post?
I download the last shinchiro build+ the last nnedi3 shaders from 2 days ago.

>>61278004
Date of your build.

>>61278004
*downloaded

>>61277882
Why not opengl-shader="~~/shaders/KrigBilateral.glsl"?

>>61277809
>Deeper reasons lie within my views on sharing code and getting things in return, but those are boring sad topics.
You are boring and sad.

File: 342342243.png (170KB, 1920x1080px) Image search: [Google]

170KB, 1920x1080px

>>61278016
Here's the SHA-1 of the build(the last shinchiro, the one from 6 hours ago) I'm currently using:
SHA1: 8120E7D4E9F73FF86F853CFB2F347C385CD0968A

>>61278136
Thats the one im using. It fixed blue screen issue for me. Maybe you installed it incorrectly or fucked up your config or shader files. Try disabling every shader except for nnedi and see if problem persists.

>>61278206
As you can see in >>61275714 I was only using nnedi3(luma)+opengl-hq

>>61278278
Fresh "Install" everything into new folder. I had blue screen with 01.07 build and 2 day old nnedi3 shaders because the shaders were optimized for new things that were implemented only in todays build, so updating fixed everything. I see no other reason for blue screen other than your mpv.exe is out of date or nnedi shader is out of date. Or maybe its a bug but i doubt it.

>>61278344
New needi3 is opengl4 only now.

File: 1494585673745.jpg (98KB, 1280x720px) Image search: [Google]

98KB, 1280x720px

>>61278344
I already told you I downloaded the last shinchiro+the last nnedi3 shaders, every time I install a new version of mpv, I run uninstall.bat, delete the mpv folder, run ccleaner to fix erros in the registry, create a new mpv folder and unzip the new version there and run install.bat.

>>61278386
I have a RX 480 with the last drivers so it shouldn't be a problem.

>>61278427
Where is the log file?

>>61278427
Well then sucks to be you. Maybe its Win10 problem. Im on Win7 and have RX 470. No problems.

profile=opengl-hq
scale=bilinear
cscale=bilinear
dscale=bilinear
correct-downscaling
alang=jpn
slang=eng
--fs
no-border
osd-font=calibri
osd-font-size=30
screenshot-jpeg-quality=100
screenshot-png-compression=9
video-sync=display-resample
interpolation
tscale=oversample

What do you think?

>>61278654
low quality bait but I'll r8 anyway
>scale=bilinear
retarded, even worse than bicubic
>cscale=bilinear
see above
>dscale=bilinear
see above
>osd-font=calibri

>screenshot-jpeg-quality=100
t. poorfag with 500 MB hard drive

Is video-sync=display-resample worth it?

>>61278875
yes; https://github.com/mpv-player/mpv/wiki/Display-synchronization

"Basically,--video-sync=display-resamplemakes video playback as perfect as possible, especially as the video framerate and display framerate get very close (eg. 60 Hz for both). In such a case, and assuming the file is not broken and the hardware is not too slow, it should guaranteezerodropped or duplicated frames. In essence, “perfect” playback"

File: tfw angry.png (29KB, 633x758px) Image search: [Google]

29KB, 633x758px

>tfw Dailymotion changed their shit again so youtube-dl doesn't work anymore

>>61278910
>tfw youtube-dl package on Ubuntu is so outdated fucking Twitch doesn't work

>>61278910
Yeah lets just shit here, complaining and weeping.
Meanwhile, actually useful people already reported and fixed it 2 days ago.
People like you are fucking cancer.

>>61277211
>mpv-fsrcnn is dead I think.
Not yet, I managed to achieve some things:

1. Getting it to actually learn shit again, by swapping out the momentum gradient descent optimizer for the adaptive adam optimizer. Not only does this learn way faster but it also always at least finds a local minimum without “stalling” forever on saddles or overshooting/undershooting. Now I can train neural nets again, faster than before

2. Swapping out PSNR for (MS-)SSIM. This actually seems to significantly change the way the network's intermediate results look, interested in seeing if the final results are any different.

Ideally I'd like the ability to change the upscaling ratio, label size and padding - but unfortunately this thing fucks up its label calculations so hard that it's effectively hard-coded to 11x11 labels / 3x upscaling ratio, and I haven't been able to find all the places in the code that need fixing to make it correctly calculate the padding for other sizes

>>61277679
Press ‘2’

>>61279048
I pressed 2 there >>61277874 Cant press 2 in June's build.

>>61277905
typical closed source hypocrite

File: Not really my motto but a recurring thought.png (647KB, 467x1125px) Image search: [Google]

647KB, 467x1125px

>>61277905
I never meant money, but like I said, boring and sad topics. Here goes.

For background starters; after doing and offering "so much" for others and realizing how little I'm offered, I've grown sensitive about my personal I/O ratio. I'm not asking 1:1 exchanges but yees, I'm a bad person who doesn't want to be just an output.
Example: say I offer a boat trip for pals without expecting anything in return (it's fun anyways), but was I ever offered such things? Subjective as fuck but I'll spare you all the elaborations on this.

I know my mindset is "faulty", but because of the IO thing, I dislike sharing my code in general.
Since I haven't had successful personal open-source projects and almost all of my private projects are created to serve my needs, I don't get satisfaction from sharing something I spent effort on and others could take and modify - even if they made it greater, at worst case (for me) they get fame for standing upon my shoulders. I'm not that nice of a person anymore, if I ever even was.
For now, I just want to be able to call this mine, own it, then release the thing when it's so complete that I believe it'll stand up to github scrutiny.

Some checks:
Contributing to an OS project? Works.
Starting an OS project (solo or with a group)? Works.
Open-sourcing something in-progress that someone "could" grab, improve/complete, claim as theirs/profit from? Nope, although the fear of exploitation is near irrational.
Releasing just the end result right now (blob/obfuscated)? Works (the thought of which made me realize the underlying reason).
Open-sourcing a "finished" project, worthy of public gazing? Yes.

So yes, I'm fine with enjoying mpv and other OS projects without sharing things I've coded up.
We can discuss specifics if you want to, but be aware I know this mindset is "wrong" - you don't have to convince me of that. Attribute it to selfishness and bad experiences.

>>61278103
True - which is also why I'm doing this.

Haasnsoft vs lanczossharp?

>>61279158
NGU

File: 1495139693990.jpg (309KB, 1745x1080px) Image search: [Google]

309KB, 1745x1080px

>>61279165

>>61279072
Well that screenshot confirms the new build is fater on your hardware as well.

The only thing that sort of strikes me as odd is the fact that the “interpolation” step seems to alternate between taking very long and being very quick, that seems like it's either a driver bug in the implemention of query objects, or maybe the GPU is constantly changing between high and low performance mode? (doubtful)

Probably a driver / measurement bug. But if you look at your earlier screenshot from june, you have the same high 2ms measurements for the “present” step, so it's not like anything has changed.

If anything, your confusion is due to the fact that mpv groups the results differently now. (“render” and “present” are no longer separate)

>>61279105
Sounds like excuses. You will never release this shit.
>>61279158
haasnsoft is good for anime of various sources (shit quality and good/okay). lanczossharp is good for good/high quality source and not only anime.

>>61279105
So to summarize.. you don't want to share your work because you fear people using it without contributing back, and therefore you're content to just use mpv without contributing back?

I... see. Very... enlightening

>>61279193
What about ewa_hanning and ewa_ginseng?

>>61279189
Please explain which is faster in >>61277679 And where i should look to compare them both? The very last build is much faster for ewa_lanczos scalers compared to July 1 build, but comparing it with June build due to different stats.lua is very confusing. For example, NNEDI3 in Junes build is SLOWER than in todays build. So apparently i thought i were looking in the right place. Please help me with my confusion!

>>61279264
For that one it seems like the left one is faster, but with *how much* it's faster it seems more so the case that your scaler settings aren't getting applied or something.

800μs total time for both scale+cscale+debanding+everything else is unrealistic, the one on the right matches what I'd expect. You may have to check the verbose logs to see if those shaders were actually all running in the june build. But if it persists, then bisect. You could also specifically test 19b0b1df51c1b^, that one has the new measurement code but the old polar algorithm.

>>61279105
You are a poor fuck, really.

Why is it always the people contributing so little/next to nothing that are so concerned about getting enough "back" from their absolutely minor shit?

Is there a way for me to free-size an mpv window? Damn thing doesn't obey xdotool or wmctrl commands for shit.

>>61279105
>I know my mindset is "faulty", but because of the IO thing, I dislike sharing my code in general.
I think you overestimate yourself a little here... there were never more than 1 or 2 anons even asking about your shit.
The last big anticipated shit was the mpv configurator guy. After he released the initial version about... 1 person used it? And then nobody ever again. Even though a lot of people wanted and requested it in the threads.

>even if they made it greater, at worst case (for me) they get fame for standing upon my shoulders. I'm not that nice of a person anymore, if I ever even was.
Just this one sentence perfectly reveals your mindset.
I can't put into words what I'm thinking about you right now. You are disgusting.

>>61278985
>he doesn't get his shit from his package manager whenever possible
It's like you absolutely want unstable bloat.

>>61275714 anon here, it looks like the problem is the angle backend.
It works with dxinterop but with worse performance than the last build.

>>61279105
>Attribute it to selfishness and bad experiences.
No. Your are not getting away with this because of "bad experiences". That would be a fine and easy excuse for your character deficits, wouldn't it.
You are just selfish and arrogant. That's it. Pretty simple.

File: Ryushika.png (426KB, 454x603px) Image search: [Google]

426KB, 454x603px

>>61279193
I will, one day. But, as I said in another response, a pessimist doesn't get disappointed. Start working on your own mpv encoder script to surpass me and my lies! Not like you'll even have any competition for dependency-free encoder scripts. At worst you have to beat a three-year-old YAD-dependent one.

>>61279240
Pretty much, though you're missing the "unfinished" part. And who's to say I'm not contributing to mpv? (Well no, not that I really am, other than the eventual release of the script). I already pointed out the hypocrisy myself as well; having a jab at it doesn't really make me realize my errors, repent or release the script incomplete/ie. any faster, if that was your aim.
But glad I could clear that up for you! Understanding is great and all.

>>61279379
Calm down, I'm holding the script hostage for money or some shit like that. It's just not done yet.
See; as long as it's minor shit, I won't bother with releasing it.

>>61279426
I'm not overestimating myself; exactly for that reason I don't want to release stuff that no-one uses - which is why I'm making it so damn good in the meanwhile.
It's not really about "revealing" my mindset - in that post I *explain* my mindset. I have no qualms about stating it: I do not want others to gain from my personal work (not as in using), as long as it's a thing of mine.

>>61279486
"Bad experiences" is what my character resulted in - I'm not so arrogant as to say the world is the problem!
But yes, you're correct. So where does that lead us next?

I can stop teasing the project here if that's so disgusting, but there's at least one anon I don't want to think that I've dropped it entirely. So just purge it from your minds and be pleasantly surprised at it on the eventual day. In the end, it's just a tool I started making for myself; book starts, book ends.

>>61279379
>Why is it always the people contributing so little/next to nothing that are so concerned about getting enough "back" from their absolutely minor shit?
Ego problems. Why do you think so many people develop closed source hobby software? Especially windows/doom9 devs...

File: analysis - Copy.jpg (416KB, 836x1097px) Image search: [Google]

416KB, 836x1097px

>>61279356
Please analyze this picture. Fresh "installs" with the config below:

opengl-backend=dxinterop
profile=opengl-hq
scale=ewa_lanczossharp
cscale=ewa_lanczos

For June build used the last compatible stats.lua script that shows render times. On July builds used the most up to date one.

>>61279589
chuuni edgelord: the post

How old are you? 15? 16?

>>61279638
1513+1880+330 = 3723 avg for june 1st

compare that to 3901 avg and 2991 avg and consider the fact that the old stats system didn't record all passes. (I assume one the 170μs are probably lost in one of the passes that didn't get recorded before)

overall seems fine to me, performance stayed the same between june and july 1st and improved between july 1st and 8th

>>61279589
>So where does that lead us next?
Well, maybe be underage somewhere else and come back when your grew into a reasonable adult?

>everybody talks about js, react, vuejs, node, ...
>go to big company github area
>python everywhere
>go to 4chan, /g/ says python is shit

File: analysis 2 - Copy.jpg (406KB, 828x1096px) Image search: [Google]

406KB, 828x1096px

>>61279700
>1513+1880+330 = 3723 avg for june 1st
Oh, so new average fresh is a summarize of all old averages? But then how >>61277679 June build is so much faster? 2277 vs 3187. It was tested with different config but the config was the same along all of the builds. Pic related is nnedi3 64 neurons comparison.

>>61279879
can you stop posting the fucking summary? it tells you nothing

>>61279935
Excuse me?

>>61279971
You're excused

File: analysis 3 - Copy.jpg (589KB, 826x1408px) Image search: [Google]

589KB, 826x1408px

>>61279700
>>61279999
Nice quads and thanks, but whats your point? I cant see passes with June's build anyway. Im only interested in comparison of July builds to the June one. Or am i missing something? A comparison with my full config: 2232 vs 4048 vs 3144.

File: 1474559648136.jpg (394KB, 1024x768px) Image search: [Google]

394KB, 1024x768px

>>61277645
bump

File: analysis 4 - Copy.jpg (309KB, 1262x722px) Image search: [Google]

309KB, 1262x722px

>>61280078
>>61279999
>>61279700
Pic related is from the last taste. Is there anything abnormal about it? Can it be that old stats.lua didnt count interpolation pass? But even without interpolation pass its 2277 (June) vs 2078 (July 8) and July 1 is MUCH slower than June - 2700!

>>61277645
shouldn't matter

>>61280284
>shouldn't matter
But it still shows in >>61280249 page like this? Why?

>>61280299
that screenshot is not using krigbilateral

>>61280249
look unless you're either going to bisect the commit or start actually posting verbose logs (like I told you to) you're not going to get any further. I still strongly doubt your settings weren't actually being applied the same way. I'm off

File: hmmm.jpg (464KB, 1486x793px) Image search: [Google]

464KB, 1486x793px

>>61280284
>>61280299
>>61280358
I mean pic related. And it affects performance too!

File: [UTW]_Fate_Zero_-_Creditless_OP2_[BD][h264-1080p_FLAC][DB6BA8BA].mkv 00_01_07.692.png (1MB, 1280x720px) Image search: [Google]

1MB, 1280x720px

>>61280358

>RiCON - hanna: so basically svp's mpv fork?
>hanna - RiCON: yes except with more RESTful and less RM-RFul
SVP! This one? -> svp-team.com

>>61280358
>look unless you're either going to bisect the commit
Im too stupid for that.
>start actually posting verbose logs (like I told you to)
You mean as in log-file=log.txt?
>I still strongly doubt your settings weren't actually being applied the same way
I checked logs for errors and everything. They look identical.
>I'm off
Well sorry im concerned about mpv's performance and bugs! I just want the software to be absolute best media player!

>>61280436
SVP is proprietary SHIT.

>>61280376
the timings disappeared when i updated stats.lua

File: what the fck.jpg (481KB, 1472x819px) Image search: [Google]

481KB, 1472x819px

>>61280376
>>61280358
WTF bicubic is just slightly faster than ewa_lanczos in this scenario!

>>61280547
What timings? I have the last one.

>>61280487
I know that! That's why I am quite confused.
Are they working on a SVP fork or is it a different thing?

>>61279263
>ewa_hanning
Haasnsoft is ewa_hanning with a higher radius and more blur. ewa_ginseng got more upscaling artifacts than ewa_lanczos and ewa_hanning.

>>61280689
I thought you were off.

>>61280689
So ewa_hanning is better for anime than ewa_lanczossharp?

>>61280575
updating mpv fixed it

>>61280817
I have the latest one.

>>61280358
I think i figured it out. In June's build, when i turn interpolation ON, render times get reduced by around 800. While in July's builds turning interpolation ON adds ~800. With interpolation OFF all results look correct: June build is on par with July 1 build and July 8 build is like 30% faster than both. So, is it stats.lua bug? Or is it some kind of weird mechanic?

File: yay.jpg (348KB, 1747x441px) Image search: [Google]

348KB, 1747x441px

>>61281074
>>61280358
Pic related.

File: 05316936.gif (34KB, 360x280px) Image search: [Google]

34KB, 360x280px

haasn, you bully, even if i were autistic with my pictures at least we can see that your recent changes improved performance in ewa_lanczos's like on 30%! And at least 10% with nnedi3. Good job. B-baka.

>>61281183
Fuck that fag, he abandoned fsrcnn

File: 2017-07-08 22_27_05-Greenshot.png (2MB, 1366x768px) Image search: [Google]

2MB, 1366x768px

What I'm doing wrong?

>>61281571
Did he abandon it because of RAVU?

File: 2017-07-08 22_27_16-Greenshot.png (2MB, 1366x768px) Image search: [Google]

2MB, 1366x768px

>>61281592

>>61281592
Your watching shitty BD rip.

File: 2017-07-08 22_23_46-Greenshot.png (17KB, 1366x768px) Image search: [Google]

17KB, 1366x768px

>>61281609

File: out.jpg (627KB, 3440x1400px) Image search: [Google]

627KB, 3440x1400px

What's going on here? on the left of the afterburner graph are files that play fine no dropped frames .003-.007 VSync Jitter. On the right is UTWs Carnival Phantasm dropping every other frame and a VSync Jitter higher than the ratio.
I've tried multiple files from the torrent and they all do the same too.

>>61281609
>>61281592
Specs?

>>61281609
Your shitty notebook can't handle high quality playback.

>>61281603
I don't think RAVU is able to surpass the quality of nnedi3.

>>61281627
try comment the

video-sync

>>61281592
not using NGU

File: 1471542984559.png (461KB, 870x722px) Image search: [Google]

461KB, 870x722px

>>61281676

File: 2017-07-08 22_39_15-Greenshot.png (210KB, 1366x768px) Image search: [Google]

210KB, 1366x768px

>>61281660

File: 2017-07-08 22_39_46-Greenshot.png (204KB, 1366x768px) Image search: [Google]

204KB, 1366x768px

>>61281694

>>61281694
try using hwdec
and possibly angle or dxinterop

>>61281694
Disable opengl-hq, I don't want to sound rude but you'll never be able to run it with that shitty celeron,

>>61281627
deband=no or failing that remove opengl-hq

Haasn, I showed you my pics. Answer me.

>>61281731
Yeah, that solved it.

>>61281731
celery nigger
https://www.youtube.com/watch?v=XqY2wuvlruo#t=20s

>>61281649
Nevermind I was retarded and forgot to change hwdec back to no after testing something.

File: 1470922112550.jpg (121KB, 485x603px) Image search: [Google]

121KB, 485x603px

bump

What is more important?
video sync+display-resample or profile=opengl-hq

>>61283419
If you want better picture profile=opengl-hq
If you want smoother playback video sync+display-resample

>can't cache beyond 10 seconds on some youtube videos
Is there a way to fix this?

>>61283781
you can set

--cache-secs

to a larger value, but I don't think there's currently a way to use the cache as you normally would
https://github.com/mpv-player/mpv/wiki/FAQ#Why_does_the_cache_not_work_for_Twitch_HLS_sites_and_some_youtube_links

>>61283836
I've tried playing with the cache size options but nothing changed.
Increasing that time there does help though. Thanks.

just updated to latest shinchiro from an old version.

what's the new vo-performance? i had a brief look in the git manual but can't find anything

>>61271176
Most likely because they do not know how easy it is to install on USB and basically boot off that for everything except video games.

>>61280376
Hmm you're right, because it still has to correct for the chroma offset (which krigbilateral seemingly just ignores).

>>61280561
Because it's not actually upscaling. The frame is already upscaled (by krigbilateral), this is just correcting the chroma offset.

(Try `bicubic_fast` instead of `bicubic`, btw)

>>61277882
Apparently you can't separate items in a list with "," now, you have to use ":" (Unix) or ";" (Windows).
Also just use "opengl-shaders". "opengl-shader" is just an alias for the append option.

>>61284706
> Apparently you can't separate items in a list with "," now, you have to use ":" (Unix) or ";" (Windows).
Is it documented somewhere in the manual? I dont see it

File: listoptions.png (14KB, 888x471px) Image search: [Google]

14KB, 888x471px

>>61284828

>>61279034
interesting! so does mssim help getting rid of the ringing artifacts? or in which way does the look differ?

>>61270828
Can I use SVP with this?

>>61285486
if you want artifacting and to ruin anything you watch then use mpc-hc or vlc don't taint mpv with that garbage

>>61284526
>>61280561
For me this chroma offset correction when using krigbilateral happens only with h.264 files, never with vp9 or HEVC.

>>61285510
K

WTF happened to the newest build?

Can't use multiple shaders now. I thought this garbage was rolled back the last patch?

>>61285521
Interesting, seems like the VP9/HEVC decoders don't set the chroma location, it gets left as “unknown” which in mpv apparently means “do nothing”?

>>61285919
>>61284828

>>61285926
Why was it changed? Any specific reason? Or just aesthetics/fucking around with people?

>>61284706
>Also just use "opengl-shaders". "opengl-shader" is just an alias for the append option.
Using shader is easier for testing anyway, it's easier to comment a line than delete the shader from the shaders list

>https://github.com/mpv-player/mpv/issues/4586
What the fuck! I am beginning to understand that kind of gibberish. OMG! Mpv transformed me into a monster!

[REEEEEEE]
profile-desc=cond:get('container-fps', 24) < 25 and get('packet-video-bitrate', math.huge) < 2000
opengl-shaders="~~/shaders/SSimSuperRes.glsl;~~/shaders/adaptive-sharpen.glsl"

How do I make this more adaptive? Currently this profile is used when my video bit rate is below 2000 kbps threshold. This works in general. However in some videos, the bitrate range from 3000 to 1000 depending on the scenes. Is there a way to automatically change shaders scene by scene based on bit rate?

>why use this in first place
Cause high bit rate video slows down when I use my shaders. On "low" bit rate, it works out fine, computationally.

>>61279034
I want to try it right meow!

>>61286100
And this doesn't seem to work always either. Not sure what's causing it. Works on some videos but not on others.

>>61286182
Did you write inverse REEEEEEE profile?

>>61286100
Your rules says that if fps 24 or less than 25 then something something. What are you doing, lol?
>"~~/shaders/SSimSuperRes.glsl;~~/shaders/adaptive-sharpen.glsl"
I heard you like sharpeners so we added sharpener on a sharpener. KEK.

>>61285486
Yes. SVP devs even say it has better performance than with mpchc because of vapoursynth. They provide windows binaries but only for Pro users

>>61279589
This is stupid, you're benefitting much more from all the existing OSS than you could ever have been wronged of.
It's distateful to post teasers without any code, in hopes of having people rely on YOU rather than your code, but that's a really shitty mindset.
You'd fit better among the madVR crowd.

Also there are other functional scripts than the YAD one.

>>61286239
Trying to cannibalize as I have no idea where property values or what the mpv specific functions are.

Intent is get the fps of video, check if its 24 fps or lower and check if video-bitrate is less than 2000 or not.

Problem I discovered after posting this (i was just thinking about how to go about doing this) are as follows. I have no idea what valid passes are for the get functions. I assume its (variable, maxvalue or something. the math.huge seems to be wrong too, so I don't know where to look up the proper attributes.

Also the shaders thing is something copy pasted from earlier, wasn't thinking about it too much but thanks for pointing out the redundancy.

Where's the mpv functions/attributes list lookup? mpv source code itself?

>>61286343
>check if its 24 fps or lower and check if video-bitrate is less than 2000 or not.
But why though? If you want different profile for, lets say, 4K content then measure width. For example

profile-desc=cond:get('width') <= 720
opengl-shader="~~/shaders/nnedi3-nns64-win8x4-all.hook"

>>61286343
>>61286390
For fps use

profile-desc=cond:get('container-fps', 0) <= 24

>>61286390
width doesn't always do justice when determining whether or not the video will be choppy. Most accurate indication for me has been video bit rate and fps.

if I have a 4k video with 1Mbps then I can add shaders without choppiness.

If I have a 720p with 5Mbps bit rate, then it will get choppy when I add my shaders.

So width doesn't workout quite well.

>>61286343
Bitrate is only going to affect decoding, right? With swdec a higher btirate shouldn't result in higher GPU usage, is that correct?

>>61286425
Logically yes,I think so. FPS and resolution is all that matters for GPU.

>>61286425
No idea, but when the scene switches to high bit rate, drops/delays happen. Its the pattern I've noticed, so thats why I was trying to find a "adaptive" solution that disables/enables shaders based on bit rate.

File: color conversion q.jpg (466KB, 1640x797px) Image search: [Google]

466KB, 1640x797px

Why color conversion step uses so much resources when using nnedi3? Pic.

File: color conversion qq.jpg (347KB, 1157x702px) Image search: [Google]

347KB, 1157x702px

>>61286492
Here is without nnedi3. But now interpolation averages are HUGE.

>>61270828
Good concept, but I suggest these.
>open I one window
>setup installer

>>61286492
Because you made the video larger, so there's more to convert

>>61286100
Found out the packet-video-bitrate is wrong or rather its always returning a 0 value. I even tried video-bitrate. same shit.

God dammit.

Anyone know what property value I should look for getting video bit rate?

>>61286692
I understand but... i mean... not 30 times bigger!

>>61286779
In theory it should take 4x as long, possibly slightly more if it exceeds some sort of cache size or scales nonlinearly.

I don't know why in your second graph it's fluctuating between 500 and 100 all the time.

>>61287011
I think stats.lua measures it incorrectly. Do you have any idea about the interpolation stats?

my niggas is there a scaling algorithm that looks nicer than spline36 without being quite as heavy as lanczos

>>61287228
No.

>>61287228
NGU :^)

>>61287260
NGU is 10x slower than spline36. :D

>>61287320
I can use NGU low even on Intel HD 400 (celeron)

>>61287338
I can use NGU even on my phone!

>>61287338
But it's still slower than spline36.
Spline36 is really fast so a 10x slower thing is still fast. ;)

FSE on Win 7 yay or nay?

>>61286303
There's much talk about benefiting off of OSS, but judging by these responses nobody is inclined to actually help develop this thing or try and convince him that hey, it's fine, nobody will exploit you.
(Also, which scripts? I only know of Zehkul's convert_script.lua [mpv, but yad] and occivink's encode.lua [ffmpeg])

>>61279589
As I said, don't mind the others too much. Teasing the thing without code/concrete release date/whatever is pretty blueballing, but I gain more from the script than telling you're twisted in the head. I'd help you code this thing if I knew lua, and I know there are other people who would. It's fine, you can change your ways, g is just known to be autistic about OSS. You can code without releasing everything even if you use OSS, but teasing a finished? product and holding onto it will get people's ire on you. Those who went against you wouldn't even use the script, they're just mad over your disgrace on the OSS community. (I just wish I had been here to manipul- guide you sooner, because the others kinda fucked up in their justice-ness.)

>>61287370
Phones are computers now so it's not that surprising. ;)
However you need a phone with win7/win8.1/win10 installed! :D

>>61287394
What the hell does somebody "exploiting" him even mean? If it's about stealing attribution or whatever, point me to *any* script to which this has happeend.

There is also this script https://github.com/lvml/mpv-plugin-excerpt which also relies on ffmpeg, but I wouldn't call it an external dependency considering that mpv already depends on it, and encoding with mpv just forwards it to ffmpeg.

I think there is a bug in new stats.lua. When with stats on the screen and pressing "2", going into detailed view i get giant bumb on total number on some passes (color conversion, scalers, etc). If i stay on advanced the bumped total number wont drop down much and will stay high forever, but if i press back to simple view "1" then total number drops after few secs as it should.

Is it a good "medium" (current iGPU) mpv config?
profile=opengl-hq
cscale=bilinear
scale=bicubic
opengl-shaders="~~/shaders/SSimSuperRes.glsl"

>>61287574
>bicubic
try spline16 instead

>>61287594
OK! SSimSuperRes became slower in the last revisions, for a good reason but I had to reduce other things to use it with iGPU.

File: total number.jpg (625KB, 2002x784px) Image search: [Google]

625KB, 2002x784px

>>61287558
This is what im talking about.

File: 96151313.jpg (25KB, 400x311px) Image search: [Google]

25KB, 400x311px

I see a lot of old madvr users from doom9 switching to mpv. Top fucking kek.

>>61287748
Madshi lost interest for madVR and his delphi/cpp madProducts become less popular and his revenue decreased a lot. Every developers will know this kind of feeling once in their life. He is working on a video only ReClock alternative but it seems to be harder than expected.

>>61287796
I like madshi but i wish he made madvr open source. Just imagine what amazing piece of software we could get in the end with contribution of thousands people.

What value

opengl-dwmflush

uses by default? Im on Win7. How do i enable fullscreen exclusive mode?

File: how2fix2.jpg (13KB, 658x36px) Image search: [Google]

13KB, 658x36px

How to fix the error in the picture? My code:

profile-desc=cond:get('width') <= 720

>>61287975
Never mind me. I've just put math huge and it werks fine now.

>>61287849
mpv is open source and nothing happens there except attempts to steal from madshi.

File: 1478418950500.jpg (399KB, 1093x1099px) Image search: [Google]

399KB, 1093x1099px

>>61288124

Where does shinchiro expect you to put your input.conf file?
You can just put them in the same folder as the exe with the lachs0r builds.

>>61287499
He's worried someone will take his "incomplete" script, and complete it / use the code for something without attribution.
Point is that exploitation would never be a real concern, and he knows this as well:
>Open-sourcing something in-progress that someone "could" grab, improve/complete, claim as theirs/profit from? Nope, although the fear of exploitation is near irrational.

That's a script I haven't seen, but on the other hand it only does -c:v copy. Fine for grabbing 1:1 clips for later usage, but not really one for muh animu webams.
(Also I wonder if mpv could do -c copy to rid the ffmpeg + script. Didn't see it in --ovc=help, but it's ffmpeg inside so..?)

>>61287394
Kagami also has one https://github.com/Kagami/mpv_slicing

He has a lot of other great software related to mpv and video in general. Truly a great contributor unlike this whiny faggot.

>>61288185
Create a folder named mpv

>>61288141
madshi's well ahead of mpv and always has been. Seems mpv is the only one who'd get any benefit out of madVR being open source.

>>61288185
You can put it in the same folder as .exe or create mpv folder and put it there. Priority in first case is bigger.

File: 1467599462732.jpg (111KB, 1280x720px) Image search: [Google]

111KB, 1280x720px

>>61288200
See >>61287748
Stay mad madshi.

Does nnedi3(luma) work with krigbilateral?

>>61288197
>>61288221
Neither seems to be working for me
I even added
input-file="~~/mpv/input.conf"

>>61288225
If open source is so good, why haven't you invented this yet?
>>61277809

>>61288244
You just create input.conf and put it in the root folder. No need to do anything else. Guess you fucked up such a simple thing. Works for me.

>>61288244
Apparently it's "input-conf" not "input-file".
That doesn't work either though.

File: 1464197465911.png (84KB, 250x256px) Image search: [Google]

84KB, 250x256px

>>61288247
If proprietary software is so good why is this >>61287748 happening?
Also
>Defending proprietary software
The current state of /g/ folks.

>>61288239
Sure

>>61288267
If you had read my first post, you would have seen that I had no problems what so ever with the lachs0r builds.

>>61288283
mpv isn't even discussed on Doom9.

>>61288291
What build you use doesn't matter. Both builds operate as designed by mpv devs. WORKS FOR ME and literally everyone else except you.

>>61287499
Yeah it's less powerful than the other scripts but I figured I'd mention it for completeness' sake.

File: 2423424234.png (23KB, 885x257px) Image search: [Google]

23KB, 885x257px

>>61288319
Are you getting butthurt aren't you?
>mpv isn't even discussed on Doom9.
And?
Maybe isn't discuss on doom9 because it can't run on Windows XP.

>>61287228
anything more than spline36 is a placebo meme

File: 4242423423.png (20KB, 1921x1040px) Image search: [Google]

20KB, 1921x1040px

>>61288291
Create a folder inside the mpv folder and name it mpv like pic related, drop the conf files there.

>>61288357
What the fuck is that screenshot supposed to mean? mpv isn't mentioned in there.

File: smh - Copy.jpg (32KB, 896x174px) Image search: [Google]

32KB, 896x174px

>>61288291
Is this really so hard?

>>61288196
Neat! (although raw clips, not directly usable files for sharing)
But I still don't think that's the right way to go about this, pushing him away. You don't have to suck his dick but goddammit if you drive him completely away, there's zero gain for me, for us. Who cares if he goes against the idea of OSS, I'll rather get that script than show how oh so wrong he is! Trying to bully him into releasing the script is a fools errand, and chasing him off is a disservice for the others.

>>61288411
>>61288411
>2017
>Still running XP
End yourself kid.

>>61288414
Somehow, you're not getting that this was the first thing I did.
I have another working build with the exact same file/folder structure. It does not ignore the input.conf file.

>>61288483
And? I thought mpv was supposed to be cross-platform. What's the matter, don't know how to program for it?

>>61288499
It is cross-platform, it runs on linux, Windows Vista and up, OSX and Android(unlike madvr that is winblows only).

>>61288553
Not very cross platform if it leaves out an OS with many times the market share of Linux.

Why is video-sync=display-resample so heavy?

>>61288566
Are you seriously defending the use of windows xp in 2017?
Can you get more retarded than that?

How do I configure to use NNEDI3?

>>61287394
>but judging by these responses nobody is inclined to actually help develop this thing or try and convince him that hey, it's fine, nobody will exploit you.
I don't give a shit about his script, I just think his attitude is pathetic

>>61288597
Is not heavy, the only problem is that you need to keep your frame timings according to your monitor refresh rate for a correct interpolation.
Example:
1000ms/60hz=16.6ms
1000ms/144hz=6.94ms

>>61287129
Stats.lua doesn't measure anything, values are from mpv and only displayed by the script

>>61288644
>Finally, --video-sync=display-* currently comes with one important drawback: Due to OpenGL's rather severe limitations when it comes to timing, the only way to reliably figure out when vsyncs happen is to actually draw a frame on every vsync. The consequence of this is that, even for 24 Hz video, you need to draw frames at 60 Hz even if they are the same frame over and over again - thus increasing power usage by a factor of 2x-3x in such a case.
Wouldn't this mean with vulkan you could use better quality scalers/shaders because you have a lot more time to render a frame watching 24fps video?

>>61288609
Create a folder inside the mpv folder and name it shaders, drop the .hooks there.
Open your mpv.conf file and write something like:
opengl-shader="opengl-shader="C:\Program Files\mpv\shaders\nnedi3-nns32-win8x6.hook"

That will run nnedi3 with 32 neurons and a window size of 8*6 for luma only.

>>61288797
If vulkan resolves the opengl limitations when it comes to timing, then yes.

>>61288832
Well, what about all this chroma, yuv, 8x4, 8x6 and neurons? Which do I chose? How are they different?

>>61288832
I fuck it up:
opengl-shader="C:\Program Files\mpv\shaders\nnedi3-nns32-win8x6.hook"

Change the file path with yours.

>>61288869
I don't want to buy a stupid nextgen GPU to use Vulkan.
Furthermore Vulkan is even more complex and bloated than OpenGL for developers.

>>61288885
If you want to upscale chroma use:
nnedi3-nns32-win8x6-chroma.hook

To upscale yuv use:
nnedi3-nns32-win8x6-yuv.hook

More neurons will provide better quality(in theory) as well as a bigger window size.

>>61288890
Wait for Q1 2018, when nvidia releases their new gpus.

>>61288928
Why not win8x4? Should be faster. Is win8x6 really better than win8x4?

>>61288928
No shit. Do I want to upscale chroma or yuv? Or do I want to upscale all?

>>61288955
>Why not win8x4? Should be faster.
Yes,win8x4 is faster than win8x6.

>Is win8x6 really better than win8x4?
In theory provides better quality.

>>61288955
Id say 8x4 is safer choice. I tested both with anime and test patterns. 8x4 outputs more stable and consistent results.

>>61288890
>I don't want to buy a stupid nextgen GPU to use Vulkan.
vulkan works on ten year old gpus. how third world are you?

>>61288987
YUV is a color space.
yuv.hook will upscale both luma and chroma in the yuv color space.
all.hook will upscale both luma and chroma in all color spaces(including yuv).

>>61289005
Is it working with hd3000 and mesa?

>>61288987
all is if you want to upscale bmp/png images.
What is XYZ though?

>>61289048
yes
https://en.wikipedia.org/wiki/Vulkan_(API)

>>61288987
>>61289033 anon here, I recommend you to read this two wikipedia articles to learn the difference between luma and chroma:
https://en.wikipedia.org/wiki/Luma_(video)
https://en.wikipedia.org/wiki/Chrominance

MOUSE_BTN3 and 4 no longer work after updating.
Did they change the labels for the scroll wheel directions?

>>61288630
You're just pissing in the well for the rest of us, so thanks. I'm certain he got the message already.

>>61289114
Yup, it's AXIS_UP and AXIS_DOWN now.

>>61289127
Perfect, thanks.

>>61287639
That's just peak, that means one single frame was like that in the past 20 seconds. That's not unusual. The first activation of hard ass drawings also make you peak. For fluent playback peak doesn't mean much.

>>61289188
>For fluent playback peak doesn't mean much
But it peaks way higher above 16ms! Wont it cause dropped frames or other shit? What "peak us" even means?

>>61289236
A lot can create peak frames. That's just the highest value in the last half minute. It depends on the exact cause if this results in dropped frames. Fullscreening can also peak timings, or changing certain properties.

>>61289325
Can you answer >>61287928 ? Thank you!

New thread here: >>61288989

Real new thread:
>>61289536