0.56.04 madness build - features plus encoding changes

[Kharay]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

I just tested this build now, and I agree with everything Muf had to say.

I actually couldn't disagree more with what was said; if anything, the media player output and Twitch player output look exactly the same to me. I also tried enforcing varying color spaces in the media player and could not get it looking game-like.

I strongly feel full range should be made optional and should be disabled by default.

 

[OG See OH Zee]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

I don't know if it's just me, but it seems the picture output is a lot darker than the last build. So far, I only tested this on my external capture card so I might have to tweak the picture settings on Elgato until further updates on it.

Regardless, it encodes better than the official builds on similar settings.

 

[Boildown]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

Just tested this build with 480p30 (My usual) and the stream was pretty pixelated, in comparison to my usual quality. Would it happen to be because I'm using the strict CBR method + tune=animation while having CRF enabled with this build?

Edit: Yeah, it's a bit too pixelated for my taste. Have to go back to stable until this one is fixed.

CBR and CRF are mutually exclusive. You can't have them both enabled, the encoder will pick one or the other.

 

[Kharay]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

I think it was an honest typo on the part of Unknown Richard. CRF, CFR... the 2 are so alike yet mean something so different.

 

[DeMoN]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

Audio is now encoded at 48khz instead of 44.1khz (again by Extrems)

Bad change.

Why you did this?

More than 44,1 khz is just wasting bitrate in the totally wrong direction :(
I mean a 128 kbit AAC @ 44,1 khz will sound better than using a wider samplerate of 48khz @ 128 kbit.
Why you ever would need 48khz? Especially in a lossy format? 44,1 khz is alread redbook cd audio quality samplerate and with that more than enough. its better to have more quality inside 44,1khz instead of wasting bitrate into wrong direction :/

Please make it optional then.

It would have so many disadvantages for me. Also recording wise - to prevent upsampling I would have to set all sound devices to 48khz. No I dont want to do that.

 

[soulreaver2662]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

I'm not sure if this only applies to me, but i've been using this build and have had instances where OBS would still be running in task manager despite closing all instances of it.

This locks up my Elgato to that phantom instance of OBS and forces me to reboot before Elgato can initialize the device again. Ending process/process tree doesn't work and the phantom instance refuses to close.

I'm also not really sure how to recreate it, but so far it has been when my Elgato stops capturing while switching games and requires a restart of OBS. I stop the stream, close OBS and there'll be an instance of OBS.exe that runs in task manager that i'm unable to close.

 

[Shinblindrezo]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

I just tested this build now, and I agree with everything Muf had to say.

I actually couldn't disagree more with what was said; if anything, the media player output and Twitch player output look exactly the same to me. I also tried enforcing varying color spaces in the media player and could not get it looking game-like.

I strongly feel full range should be made optional and should be disabled by default.

There are several possible reasons as to why the colour appears off to you.

If you have your computer connected (via DVI, not HDMI) to a proper monitor (not a TV), then you will need to check all these settings for your graphics card / media player and make sure that they are all set to RGB.

Wish I could help more, but it would be difficult without being there in person to have a look at it, heheh. :)

 

[Kharay]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

HDMI to an actual TV; already enabled any and all colour spaces/profiles that ffdshow tryouts ( http://ffdshow-tryout.sourceforge.net/ ) would allow (which basically is... all of them) and, the colors still seemed off. And, well, so they did as well in the Twitch player. But not in-game. Had it been a monitor issue the difference would be apparent both pre-recording as well as post-recording (ie -- both in-game as well as in the recording of the game).

 

[DeMoN]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

If MPC (well, more specifically the decoder filter used) decodes it incorrectly, then that means it's not reading the colour space tag properly. I'd suggest trying CoreAVC, which should have support for full range H.264.

MPC-HC 32bit supports all colorspaces! Without any 3rd party filters. (where I would rather recommend latest ffdshow svn 32bit)

Just ensure that you have in options @ output as video renderer the Enhanced Video Renderer (custom presenter) selected. Should be defaulted to this, but I got to hear that with win8 users mpc-hc defaulted this to VMR9 for whatever reason. VMR9 does output the colorspace wrong. EVR does it right. So use EVR instead. EVR is better renderer anyway.

Players which tend to have wrong colorspace output are VLC and WMP. VLC you can correct it with disabling a specific checkmark though.

__

But because of the importance for me I repeat:

Please please make that 48khz for audio optional at least.

 

[Shinblindrezo]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

HDMI to an actual TV; already enabled any and all colour spaces/profiles that ffdshow tryouts ( http://ffdshow-tryout.sourceforge.net/ ) would allow (which basically is... all of them) and, the colors still seemed off. And, well, so they did as well in the Twitch player. But not in-game. Had it been a monitor issue the difference would be apparent both pre-recording as well as post-recording (ie -- both in-game as well as in the recording of the game).

Sounds to me like either your graphics card is overriding the video (media) colour settings, or your media player/decoders are set to RGB, when they should be set to TV colour range.

At least for my Nvidia card, it has it's own option specifically for how to handle the colour for video media. Regardless of what you set the general Desktop colour range to. I had to disable that feature and tell the nvidia drivers to let the media player handle the colour range.

 

[ball2hi]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

Just tested this build with 480p30 (My usual) and the stream was pretty pixelated, in comparison to my usual quality. Would it happen to be because I'm using the strict CBR method + tune=animation while having CRF enabled with this build?

Edit: Yeah, it's a bit too pixelated for my taste. Have to go back to stable until this one is fixed.

CBR and CRF are mutually exclusive. You can't have them both enabled, the encoder will pick one or the other.

Sorry, I meant to say CFR (Constant Frame Rate?) instead of CRF.

I seem to have removed tune=animation and removed the custom buff-size from the strict CBR. The quality seems to have gone back to normal with this build, still having CBR enabled with CFR enabled.

 

[Gol D. Ace]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

CBR (Constant Bit Rate) works with CFR (Constant Frame Rate).

But CBR (Constant Bit Rate) is NOT working with CRF (Constant Rate Factor).

 

[Greendweller]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

Crashlog:

OBS has encountered an unhandled exception and has terminated. If you are able to
reproduce this crash, please submit this crash report on the forums at
http://www.obsproject.com/ - include the contents of this crash log and the
minidump .dmp file (if available) as well as your regular OBS log files and
a description of what you were doing at the time of the crash.

This crash appears to have occured in the 'c:\windows\system32\kernelbase.dll' module.

**** UNHANDLED EXCEPTION: 80000003
Fault address: 000007FEFDC53C72 (c:\windows\system32\kernelbase.dll)
OBS version: Open Broadcaster Software v0.56.01 MADNESS
Windows version: 6.1 (Build 7601) Service Pack 1
CPU: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz

Crashing thread stack trace:
Stack            EIP              Arg0             Arg1             Arg2             Arg3             Address
000000000017EAA8 000007FEFDC53C72 000007FEE9095870 0000000000007530 000000000017EA08 000000000017EC70 kernelbase.dll!0x7fefdc53c72
000000000017EAB0 000007FEE8F0D433 0000000000001389 000007FE00000000 000000000002058A 0000000077A79AA6 obsapi.dll!OSTerminateThread+0x33
000000000017EAE0 000000013FE4D4EA 0000000000001389 000000000017EC70 0000000000000111 0000000000000000 obs.exe!OBS::Stop+0x7a
000000000017EB70 000000013FE6C5DE 0000000000000070 FFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFF 0000000077A79B43 obs.exe!OBS::OBSProc+0xa8e
000000000017ED00 0000000077A79BD1 000000000085B4F0 000000013FE6BB50 00000000FFFFFED1 000000000002058A user32.dll!0x77a79bd1
000000000017EDC0 0000000077A76AA8 0000000000020560 0000000000000111 0000000000001389 000007FEFC580A29 user32.dll!0x77a76aa8
000000000017EE50 0000000077A76BAD 00000000001E79E0 0000000000000001 0000000000000000 0000000000000001 user32.dll!0x77a76bad
000000000017EEA0 000007FEFC580BBF 000000000002058A 0000000000000001 00000000001E79E0 000007FEFF531D18 comctl32.dll!0x7fefc580bbf
000000000017EEE0 000007FEFC5847FE 000000000000FF00 0000000000030000 0000000000000202 0000000000000000 comctl32.dll!0x7fefc5847fe
000000000017EFA0 0000000077A79BD1 000000000017F288 000007FEFC583B20 000000000085B4F0 000000000085EB50 user32.dll!0x77a79bd1
000000000017F060 0000000077A798DA 0000000000000000 0000000000000000 000007FEFC583B20 0000000000000001 user32.dll!0x77a798da
000000000017F0E0 0000000077A767C2 00000000000205CD 00000000000205CD 000007FEFF532164 0000000001C11000 user32.dll!0x77a767c2
000000000017F170 000000013FE418A9 0000000000000000 0000000000000000 000000013FE20000 0000000000000000 obs.exe!WinMain+0x9c9
000000000017F710 000000013FEB5A20 0000000000000000 0000000000000000 0000000000000000 0000000000000000 obs.exe!strstr+0x1ac
000000000017F7C0 0000000077B7652D 0000000000000000 0000000000000000 0000000000000000 0000000000000000 kernel32.dll!0x77b7652d
000000000017F7F0 0000000077CAC541 0000000000000000 0000000000000000 0000000000000000 0000000000000000 ntdll.dll!0x77cac541

A minidump was saved to C:\Users\Gebruiker\AppData\Roaming\OBS\crashDumps\OBSCrashDump2013-09-02_1.dmp.
Please include this file when posting a crash report.

List of loaded modules:
Base Address                      Module

Log:

22:27:35: Video Source Plugin strings not found, dynamically loading 38 strings
22:27:53: Open Broadcaster Software v0.56.01 MADNESS - 64bit (　^ω^)
22:27:53: -------------------------------
22:27:53: CPU Name: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
22:27:53: CPU Speed: 3332MHz
22:27:53: Physical Memory:  8159MB Total, 4904MB Free
22:27:53: stepping id: 7, model 10, family 6, type 0, extmodel 1, extfamily 0, HTT 1, logical cores 4, total cores 4
22:27:53: monitor 1: pos={0, 0}, size={1920, 1080}
22:27:53: monitor 2: pos={-1280, 33}, size={1280, 1024}
22:27:53: Windows Version: 6.1 Build 7601 S
22:27:53: Aero is Enabled
22:27:53: -------------------------------
22:27:53: OBS Modules:
22:27:53: Base Address     Module
22:27:53: 000000003FE20000 OBS.exe
22:27:53: 00000000E8EE0000 OBSApi.dll
22:27:53: 00000000F3040000 DShowPlugin.dll
22:27:53: 00000000E8EB0000 GraphicsCapture.dll
22:27:53: 00000000F4040000 NoiseGate.dll
22:27:53: 00000000E8E90000 PSVPlugin.dll
22:27:53: 00000000E8E70000 scenesw.dll
22:27:53: 00000000F42B0000 VideoSourcePluginWrapper.dll
22:27:53: 00000000E8DF0000 WebSocketAPIPlugin.dll
22:27:53: ------------------------------------------
22:27:53: Adapter 1
22:27:53:   Video Adapter: NVIDIA GeForce GTX 560 Ti  
22:27:53:   Video Adapter Dedicated Video Memory: 1025966080
22:27:53:   Video Adapter Shared System Memory: 3220779008
22:27:53: =====Stream Start: 2013-09-02, 22:27:53===============================================
22:27:53:   Multithreaded optimizations: On
22:27:53:   Base resolution: 1600x900
22:27:53:   Output resolution: 1280x720
22:27:53: ------------------------------------------
22:27:53: Loading up D3D10 on NVIDIA GeForce GTX 560 Ti...
22:27:53: C:/Users/Gebruiker/AppData/Roaming/OBS/shaderCache/shaders/DrawTexture.pShader.blob
22:27:53: C:/Users/Gebruiker/AppData/Roaming/OBS/shaderCache/shaders/DrawSolid.pShader.blob
22:27:53: C:/Users/Gebruiker/AppData/Roaming/OBS/shaderCache/shaders/DownscaleLanczos6tapYUV.pShader.blob
22:27:54: Playback device {0.0.0.00000000}.{ec118582-431d-43b3-93a7-50cf9c8a9acb}
22:27:54: ------------------------------------------
22:27:54: Using desktop audio input: Luidsprekers (Realtek High Definition Audio)
22:27:54: Using 5.1 speaker setup
22:27:54: ------------------------------------------
22:27:54: Using auxilary audio input: Microfoon (Realtek High Definition Audio)
22:27:54: ------------------------------------------
22:27:54: Audio Encoding: AAC
22:27:54:     bitrate: 160
22:27:54: C:/Users/Gebruiker/AppData/Roaming/OBS/shaderCache/shaders/ColorKey_RGB.pShader.blob
22:27:54: C:/Users/Gebruiker/AppData/Roaming/OBS/shaderCache/shaders/AlphaIgnore.pShader.blob
22:27:54: Using Window Capture
22:27:54: ------------------------------------------
22:27:54: Video Encoding: x264
22:27:54:     fps: 60
22:27:54:     width: 1280, height: 720
22:27:54:     preset: veryfast
22:27:54:     CBR: yes
22:27:54:     CFR: no
22:27:54:     max bitrate: 4000
22:27:54:     buffer size: 4000
22:27:54: ------------------------------------------
22:27:54: MMDeviceAudioSource: Frequency for device 'Luidsprekers (Realtek High Definition Audio)' is 4608000, samples per sec is 192000
22:27:54: MMDeviceAudioSource: Frequency for device 'Microfoon (Realtek High Definition Audio)' is 352800, samples per sec is 44100
22:27:55:   Interface: Intel(R) 82579V Gigabit Network Connection (ethernet, 1000 mbps)
22:27:56: SO_SNDBUF was at 8192
22:27:56: SO_SNDBUF is now 65536
22:28:03: ++++++++++++++++++++++++++++++++++++++++++++++++++++++
22:28:03:   New Scene
22:28:03: Using text output
22:28:03: Using graphics capture
Warning -- DShowPlugin: Could not create device filter
22:28:03: Using directshow input
22:28:03: C:/Users/Gebruiker/AppData/Roaming/OBS/shaderCache/shaders/DrawTexture_ColorAdjust.pShader.blob
22:28:03: Using text output
22:28:03: SharedTexCapture hooked
22:28:04: RTMPPublisher::Socketloop: Increasing send buffer to ISB 131072 (buffer: 0 / 532480)
22:28:08: RTMPPublisher::Socketloop: Increasing send buffer to ISB 524288 (buffer: 455 / 532480)
22:29:02: GraphicsCaptureSource::NewCapture:  eliminating old capture
22:29:02: SharedTexCapture hooked
22:29:08: RTMPPublisher::Socketloop: Increasing send buffer to ISB 1048576 (buffer: 0 / 532480)
23:50:29: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:29: SharedTexCapture hooked
23:50:30: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:30: SharedTexCapture hooked
23:50:31: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:31: SharedTexCapture hooked
23:50:36: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:36: SharedTexCapture hooked
23:50:37: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:37: SharedTexCapture hooked
23:50:38: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:38: SharedTexCapture hooked
23:50:39: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:39: SharedTexCapture hooked
23:50:40: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:40: SharedTexCapture hooked
23:50:44: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:44: SharedTexCapture hooked
23:50:45: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:50:45: SharedTexCapture hooked
23:52:13: GraphicsCaptureSource::NewCapture:  eliminating old capture
23:52:13: SharedTexCapture hooked
00:18:21: Capture window 0x01140572 invalid or changing, terminating capture
00:18:28: Total frames encoded: 397994, total frames duplicated 3584 (0.90%)
00:18:58: WARNING: Forcibly terminating a thread after 30000 ms timeout!

and below the dumpfile.

 

[Xphome]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

Just a wild guess: The color problem can't be caused by the monitors color profile in Windows?

 

[AlderaaN]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

Audio is now encoded at 48khz instead of 44.1khz (again by Extrems)

Bad change.

Why you did this?

Why you ever would need 48khz? Especially in a lossy format? 44,1 khz is alread redbook cd audio quality samplerate and with that more than enough.

Please make it optional then.

He's got a point.

Please consider making it optional, rather sooner than later.

Regards,

 

[antome]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

Devs, please read this article Here on sampling rate.
48khz is completely unnessecary at streaming bitrates, and could even cause distortion and make other problems audible with a bad signal.

An audio wave can be represented perfectly fine at 4.1khz. Please make it an option, and the default option. Most streamers won't know whether their setup will have any benefits/problems at the higher sampling rate.

Not only that, but half of recipients won't even be using 48khz to decode, either voluntarily or without knowledge! It would go to waste even if the right setup was used.

TLDR the idea that 48khz is objectively better is a myth.

 

[Krazy]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

I just read that article and nowhere did it say anything negative about 48Khz. It was mostly about why 192Khz is overkill/bad, and that 48 and 44.1 are perfectly fine.

Anyway, for those of you that are reaaaaaaaaaaally peeved about it, it's going to be made an option so no worries.

 

[AlderaaN]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

for those of you that are reaaaaaaaaaaally peeved about it, it's going to be made an option so no worries.

Thank you.

 

[Lain]

Re: 0.56.02 MADNESS build - features plus encoding changes

Updated:

• Fixed issues with quicksync in the new code (palana)
• Added date/time for FLV files (palana)
• Added a "projector" feature that allows you to preview the stream on a separate monitor in full screen while using OBS on a another monitor, can be used by right-clicking the preview window and using the "projector" option from the context menu and selecting the monitor to use. The projector window can be disabled by pressing escape on it, by closing it on the task bar, or by using the "disable" option in the context menu.
• Made both 44.1khz and 48khz optional in encoder settings (no longer hard coded to either)
• Reduced minimum settable FPS to 1fps (*shrugs*)
• Added encoder thread to the profiler
• Fixed a bug where a global source would initialize when switching to a scene with an unchecked global source

 

[Fred_]

Re: 0.56.01 MADNESS build - testing mostly encoding changes

MPC-HC 32bit supports all colorspaces! Without any 3rd party filters. (where I would rather recommend latest ffdshow svn 32bit)

Just ensure that you have in options @ output as video renderer the Enhanced Video Renderer (custom presenter) selected. Should be defaulted to this, but I got to hear that with win8 users mpc-hc defaulted this to VMR9 for whatever reason. VMR9 does output the colorspace wrong. EVR does it right. So use EVR instead. EVR is better renderer anyway.

Interesting you should say that. Before I had read into colorspaces a bit, I had the biggest problem to get colors look just right outside of the game I was capturing (not using obs, but irrelevant). MPC, photoshop, windows photo viewer and paint couldn't agree on a color, and not one of them was exactly like in the game! Eventually I removed some wonky profile in color management that fixed paint and photo viewer.
As for MPC... switching from EVR to VMR9 fixed that, heh.