/jp/ - Otaku Culture

vlc is superior

>>5022537
sketchbook AND mizore?
I like you, sir.

So what OP? Mine's just a bit older, but it works very well.

>>5022537
>vlc
>superior
pick one.

get out MPC devs

>>5022537
fuck i want your monitor

>>5022537
>SOUNDMAN.EXE
>vlc
>eXperience

>>5022558
His display appears to be 4x3, which is a bit unusual these days. How can you be sure he doesn't have some old CRT at way too high resolutions? If he had it at 1920x1200 it might have been more believable.

>>5022537
>Halo
Oh god, what?

>>5022567
I don't care about black bars or aspect ratio. More pixels = more space to be used.

>>5022567
that picture was originally posted by a namefag called "isac", his monitor is an old 19" CRT. For some reason he runs it at 1920x1440 and 60Hz (I dont want to imagine the headache he gets from that, CRT monitors should ALWAYS be 85Hz+. If I was him, Id use some smaller resolution like 1600x1200 to achieve that)

Hey guys.
Is there any way to tell haali renderer not to downscale my 1080p?
I am on a 1680x1050 1920x1080 dual monitor setup.
MPC+HaaliRenderer automatically downscale my 1080p to 590p. I have no such problem if I use overlay mixer.

>>5022581
That sounds pretty terrible. Too bad >>5022558 didn't know what to wish for.

>>5022580
Do you even know what you said?
Do you realize what bad AR means? Try looking at a circle or watching a movie with long or fat faces/bodies. Disgusting.

>MSVC

>>5022582

Are you sure its actually downscaling it? That's just the default resolution when you start the player, if you set it to fullscreen you'll see the full resolution. It should be easy to tell 540p from 1080p.

>>5022582
MPC's reported resolution is not always the real one. If you're in full screen, Haali's renderer won't downscale. If you want to check the effective resolution, do this:

Play->Filters->Haali video renderer
Click on "Show OSD", Press Ok. Go fullscreen and look at "Scaler" line.

>>5022591
Superior on Windows. Don't be making exes with gcc on Win32 please. GCC adds so much needless bloat, unless you take a lot of effort to avoid it. MSVC generates much cleaner and leaner exes. It may not be portable to anything besides Win32, but it's actually good on this platform, along with ICC.

>>5022593
>>5022592
Hmm.
You might be right? Scaler is showing 1920x1080.

>>5022589
I meant of the monitor. I do not care if the monitor is 16:9, 16:10, 4:3, 5:4, etc. Of course I care whether or not what's ON the monitor is the correct aspect ratio.

These threads seem to get guarranteed replies, for reasons I cannot fucking fathom.

How about a REAL player?

>>5022655
People take their playback system seriously.

>>5022664
I've been trying to understand for a while what exactly makes one player better than another.

I see it now: practice.

GNOME MPlayer here.

>>5022664
Needs less frontend.

>>5022675
First, you have to consider how splitting/rendering/decoding (+subs,etc) is done. Then you consider individual components. When it comes to Windows players like MPC and such, they offload most of the work to DirectShow which lets you plug any filter you want. Common DShow filters would be ffdshow(decodes H264(AVC1)/ASP(XviD/DivX)/WMV and many other kinds of video and audio, based on ffmpeg's decoders), haali's splitter(splits mkv/mp4/avi/ts/... into streams), vsfilter (provides subtitle rendering). Renderers control on what device the drawing is done, as well as maintaining proper sync and dealing with jitter, converting colorspaces if needed(decoders can do this too). A player like MPC uses DShow, but also has internal filters, and it offers many ways to control and tune the filter graph as well as individual per-filter settings. Other Windows players are like that too. As for KMPlayer, it's a DShow player like that, with some interesting features, but it could break some things like proper subtitle rendering in subtle ways (some think this is an advantage). In relity, there's nothing I can't do in MPC(with tweaking the filter graph as well as playing with each filter's setting), that you can do in KMPLAYER, although some things may be more user-friendly in it.

Now if you step out of Windows players, you have VLC and mplayer, which do share code with ffmpeg(ffdshow shares a lot with that too). Things are a lot more monolithic there, and everything comes standalone, there's little to no plugging new filters or managing the filter chain in detail, but that doesn't mean these players are bad. Mplayer is actually pretty good. VLC is just sloppy, but it has some nice qualities when it comes to streaming crap.

>>5022714

WHY THE FUCK IS THIS NOT IN /a/?

Go take your animu players there.

>>5022714
lol you care about this shit

Shit keeps lagging on my laptop D:

>>5022769
mpchc + coreavc, no haali mediasplitter

enjoy

>>5022675
it actually working
VLC doesnt play even nearly everything I've tried, MPC-HC+CCCP has never had any problems with any non-corrupted files. I do have VLC installed for the sake of live streaming anime, though.

>>5023075
I have MPC + CCCP + CoreAVC and since episode 43, FMA brotherhood (1280x720 h264) is full of artifacts.
All other videos seem to work normally.
I have tried MPCHC and different settings in "options", but nothing changed.

>>5022537
could you post your wallpaper image here please?

>>5023351
Im not isac and neither was the guy who posted that screenshot, but I think the wallpaper picture is this one
and now listening to Kugimiya Rie - Snowstorm

sup
I can't play 1080p videos at all, and 720p anime lags during OPs/EDs.
Tried both vlc and Mpc+cccp, no difference.

I have a 2ghz dualcore and 512mb 8800GTS... Time to upgrade or am I doing it wrong?

>>5023384
go ask /a/ or /g/, they might post a guide for you.
Im pretty sure you should be able to play 720p without problems, 1080p I dunno

>>5023379
Thank you sir! I'm also listening to that.

>>5023312
Update to coreavc2.
Those encoders are being an elitist prick and they are breaking the old coreavc on purpose.

For a few seconds I thought I somehow missclicked on /a/.

tl;dr

Use *DXVA* and MPC-HC for best quality and performance.

>>5023312
They're using some slightly superior settings which compress much better. This is a good idea since Eclipse's releases are way too overcompressed and tiny (330mb for a show full of action? I'd go for 500mb easy!), so I'm glad they're using some better settings at least.
>>5023410
Not their fault CoreAVC had that decoding bug, although the problem has been known for ages, and CoreAVC2 has been around for some time too. While, I wouldn't break compatibility in my encodes, my encodes are also huge, because I like preserving quality and don't care about huge filesizes.
>>5023384
> 2ghz dualcore and 512mb 8800GTS
720p sure, 1080p maybe. Try and see if CUDA decoding works, if CUDA fails, try DXVA which is a bit more limited in the profiles it supports. If those aren't enough, make sure you're using ffdshow-mt (I believe CCCP includes it now, but don't hold my word for it, I've installed it before), and have it use 2 threads - see if that's enough to allow realtime decoding. If I remember right 8800GT should be capable of doing CUDA decoding of 1080p stuff a few times faster than realtime, but 8800GTS has an older pixel shader version, and is overall worse, so CUDA decoding might not work right... So to list your options:
1) CUDA
2) DXVA
3) ffdshow-mt (2 threads)
4) CoreAVC2 (try different settings)
>>5023426
There is no "best" quality, there's just correct and incorrect decoding. IDXVA is fast decoding, but fast does not mean incorrect, however that doesn't mean the decoding is perfect either - I'd actually like to see some tests which verify if there are any decoding bugs in GPU accelerated decoders. Early CoreAVC had some deblocking bugs. Since I have a fast CPU, I just do full software decoding, even for 1080p, because that's known to be the correct. The downside is slightly increased overall CPU usage.

>>5023536
Isn't CUDA supported only by CoreAVC which known as an inferior quality codec?

>>5023384
I'm on 2.93GHz single core with 64mb integrated graphics and play 720p perfectly fine. Fix your shit.

I have CoreAVC 2.0 and the latest CCCP beta. Make sure Haali media splitter is unticked for everything, as well as directvobsub.

>>5023536
>>5023410
Thank you, it works.

I wish I could use madVR as my renderer, too bad my 9500GT can't render 1080p stuff at realtime. Stupid GPU based thing.

>>5023561
CUDA is supposed by NVIDIA's driver. Lots of applications can use it regardless. CoreAVC can decode using it, but it's not the only decoder. There's some more "professional" quality decoders out there which use it too.

CoreAVC isn't an "inferior quality codec". What the heck does that even mean?
Video is encoded using some encoder(like x264) to yield a H264 stream. A decoder takes that H264 stream and generates a string of images from that compressed H264 stream. That's what decoding is. It's an uncompression process. Just like a JPEG should decode to the same picture in all compliant readers, the same way H264 streams should decode the same in all compliant decoders. Now, old versions of CoreAVC(1.3-?) had a decoding bug which was noticeable in the presence of high/low deblocking values. A bug which was fixed a long time ago.

Or in short: There's only correct and incorrect decoding.

CoreAVC will most likely have lower CPU usage than pure software decoders like ffdshow's libavcodec, but most people will trust libavcodec to be always accurate as opposed to CoreAVC which is known to have had major decoding bugs (libavcodec had some too, to be fair).

My other reason for prefering ffdshow is that it had really nice filters that you can use if you need to, and they're applied right after decoding.

>>5023590
>I wish I could use madVR as my renderer, too bad my 9500GT can't render 1080p stuff at realtime. Stupid GPU based thing.

what is madevr
if its just an EVR - then it doesn't support gpu acceleration
use EVR custom pres.

>>5023590
Haali's renderer is enough for me. madVR is a good idea, but chances are it won't make a difference on current LCD displays with current video drivers. If you're actually experiencing banding problems, I suggest you use some slight gradfun2db/gradfunkmirror, or just ffdshow's deband filter (at 1.1-1.2).

>>5023614
A GPU assisted renderer which has a better output than EVR's. And yeah, not being able to use it I have to go with EVR custom.

http://forum.doom9.org/showthread.php?t=146228

>>5023626
Yeah, I'm using GradFunkMirror() which is pretty good.

By the way Nvidia still didn't fix their drivers for correct .avi usage. Proof here: http://codecs.freeforums.org/latest-nvidia-graphic-driver-t2745.html

And these fanboys call drivers by ATI "shitty"?

>>5023536
How does this DXVA thing works?
Coreavc2 with CUDA acceleration break my geforce8600gt. Even though it turns green, my 1080p is still rendered at 3frames/sec. I have to turn off to get lag free video but at the expense of 40% cpu load.
Using mpc-hc + haali renderer + coreavc decoder at the moment.

Anyone else having problem with sound not being in sync playing some stuff(in my case some anime raws) using MPCHC?

>>5023636
I've had plenty of instances of NVIDIA fucking up their drivers. A few rare BSoDs, one time it fully broke the font subsystem, and some other crap. The solution was to always revert to a good driver. They should test their drivers more.

As for ATI, certain things are plain unsupported and undocumented territory. Even if the hardware is capable of doing the things, the drivers don't have full support for everything, and when it comes to registry fixes for certain issues, the situation is even more dire than NVIDIA's.

In the end, both companies suck at it, just NVIDIA's drivers suck less when it comes to video support and CUDA hw acceleration.

Here's a common example: VMR9 colorspaces are notoriosly broken on both ATI and NVIDIA drivers. If you want to tweak them to achieve correct color spaces, you can change some registry settings for NVIDIA's driver, and the problem is fixed. For ATI, they had one single setting which didn't offer full control over details, and in a latter version, they've removed all support for this, so it's unfixable now. If you want it fixed, you'll have to harass their engineers.

>>5023676
Is video decoding fast enough? Is there some "lag"? Or is it just desynced audio/video because of encoding issues? You could try loading the video in Aegisub and checking the sync yourself.

>>5023686
I'll elaborate a bit more on the colorspace issue: There is no right solution. NVIDIA's driver lets you tweak it to fit your needs, which may change over time, while ATI removed the option, so you can't tweak it. So what do do you do with ATI? You harass their engineers, so they'll change the colorspace for you once, and it'll be included in the next driver, but this change will break things for another person which needs a different color range, and they'll bitch at ATI as well, which will cause a revert in these settings, and so on, and so on.

>>5023634 here.
Disable CoreAVC2 and used DXVA.
15%cpu usage and constant framerate., definitely much better performance.
Problem is I have to use EVR custom pre and it disabled my aero effect for windows7.
Any workaround for this?

>>5023693
No lag in neither sound nor video the sound is just slightly delayed. Video plays at double speed the first few seconds.

>They still use VLC

>>5023792
this is why I hate using vlc

Watching H.264 videos using DirectX Video Acceleration (DXVA): http://imouto.my/watching-h264-videos-using-dxva/

>>5023410

Actually, x264 changed the default settings to break old CoreAVC by default.

>>5023716
Right-click in MPC-HC, Renderer Settings, Presentation. Make Sure Disable Desktop Composition (Aero) is NOT CHECKED.

>>5023817
x264 is known for adding new features to defaults all the time. It's the encoder's fault for not learning about it, but I believe a lot of groups are fine with mbtree being enabled by default anyway.

>>5023814
Unfortunately, DXVA doesn't work with some h264 profiles. Still, it may be a good solution for people without a powerful enough CPU and who can't use CUDA. I myself just go with ffdshow mt since I never had that problem.

>>5023840
DXVA works with all profiles, even L5.1 16 reference frames on Nvidia.

>>5023840
>Unfortunately, DXVA doesn't work with some h264 profiles

Very rare.
Most of HD content I watch (and I watch only HD nowadays, always 1080p with blu-ray quality if it possible) supports it.

>>5023856
What? Maybe I'm late, but when did they fix that? Not like I'm going to use it, but if it really is true I'm going to tell some guy I know who has quite the weak CPU but some good ATI graphic card.

>>5023873
You're such a slowpoke. Nvidia had L5.1 16 reference frames support since 2008.

>>5023878
Way to be slow, me ヽ(´ー`)ノ

By the way about "slow" CPUs.
I can't watch 1080p blu-rays with my q9550 (@2800, didn't try after clock to 3600) without a DXVA support.

>>5023897
Something's wrong there.
I have no problem watching 1080p BDs, using either software (ffdshow-mt) or hardware(CUDA). Q6600 here.

>ATI DXVA 2.0 and h.264 with profile 5.1 and 16ref frames decoding should be possible in ATI Catalyst 10.5 and ATI 4xxx/5xxx and Vista/Win7

Oh, this is quite sweet. Definitely a good alternative to software decoding now that it's fully supported by both ATI and nVidia.

I've been using jetAudio ever since I saw this comparison.

>>5023975
The important thing is if it can decode most things you usually throw at it correctly, with good performance. I wouldn't make a big deal with something using more RAM or taking 1-2 seconds longer to load, as long as it offered proper playback for all the formats I usually view.

>>5023384
You need outside help. I'm running a laptop with a 8600M GT and Intel Core 2 Duo 2.00 GHz and I have no problems with 720p, but a little stuttering sometimes with 1080p. You must have some seriously screwed up stuff going on with your computer.

>>5023838

Of course, and it isn't as if MB-tree affects compatibility with anything anyways.

>>5024108
Didn't it break old CoreAVCs?

>>5024124

No bro, that's weightp 2. MB-tree is just an encoder-side decision algorithm for distributing bits. weightp 2 is (one use of) a particular feature in the spec.