CasparCG Server 2.1 status

#1
Questions has been popping up here and there about CasparCG Server 2.1 and the current status, when it will be released and so on.

I will use this thread to shed light on the progress. What I'm currently doing is this:
  • Fixing bugs introduced by me in the FFMpeg module when I have implemented support for:
    • Using audio from multiple audio streams. Audio is often arranged like that in MXF files.
    • Framerate conversion.
    • Changing playback speed of clips.
  • Writing automated test cases in JUnit for the server to make the manual tedious testing before a release to become less needed. I will put this up on github when it is polished enough.
  • Working on instability issues with Quad 2 and Duo 2.
  • Taking Armin up to speed on the CasparCG Project.
This is what still needs to be done (that I can think of):
  • Merge ffmpeg_consumer and streaming_consumer into one and use ffmpeg_pipeline there. Implementing file_sink in ffmpeg_pipeline is needed first.
  • The reverse of "Using audio from multiple audio streams" -- Possibility to record in that way.
  • Remove frame_muxer and instead use ffmpeg_pipeline in decklink_producer.
  • Update the changelog and readme.
What I could use help with is:
  • Determining what points I have forgotten above.
  • Comparing 2.0.7 and 2.1.0 feature-wise and behavior-wise, and letting me know if there is something that I have forgotten to merge from 2.0.7 and 2.1 or if I have introduced some breaking change.
I will not say anything about release dates, it will only create disappointment if they are not fulfilled.

Re: CasparCG Server 2.1 status

#2
Good to hear Helge! Glad you and Armin take some time to give us updates in the forum and at Github.

I intend to write a full test suite for AMCP commands through our node.js casparcg-connection library. This will run against an Amazon windows instance to test the 2.1 AMCP protocol: all commands with various combinations of valid and invalid parameters.

Having Travis-CI in the pipeline for casparcg-connection could in the future mean doing automated tests agains the latest builds of CasparCG Server, both Linux and Windows.

I guess the AMCP implementation isn't your biggest concern at this stage. But working my way through it over the last few months has already revealed some strange and/or undocumented behaviour, as previously discussed directly or through Github issues.
Jesper Stærkær
Independent Consultant at SuperFly.tv

Re: CasparCG Server 2.1 status

#6
Hi,

It is nice to hear that people are working on CasparCG Server 2.1. We are eagerly waiting for it.

I would like to add my observations. I downloaded

Code: Select all

CasparCG Server 70235ae09df45c874e133fd4d3fcc1e6e34e993a_windows.7z
updated on 9/9/2016.

It is working alright with Decklink Duo2. But it is not seeking to particular frame. The code (.NET) with which I wrote in VB and working with 2.0.7/Duo is not working in 2.1/Duo2. Especially, the seeking the frame and OSC as well.

kindly look into these.

Thanks,

Re: CasparCG Server 2.1 status

#7
For those that aren't aware; current beta builds of CasparCG Server 2.1 can be downloaded at http://builds.casparcg.com/

Please post bug reports and feature requests as GitHub issues, so the developers can keep track of them:
Jonas Hummelstrand
Independent Consultant at SuperFly.tv
Former Community Manager for CasparCG

Problems? Guide to posting Bug reports & Feature requests

Re: CasparCG Server 2.1 status

#8
Helge Norberg wrote: This is what still needs to be done (that I can think of):
  • Merge ffmpeg_consumer and streaming_consumer into one and use ffmpeg_pipeline there. Implementing file_sink in ffmpeg_pipeline is needed first.
  • The reverse of "Using audio from multiple audio streams" -- Possibility to record in that way.
  • Remove frame_muxer and instead use ffmpeg_pipeline in decklink_producer.
Update: After some discussions with pros/cons we have now started to lean towards using the 2.0.7 ffmpeg producer in 2.1.0 because it is much more stable. I will still try to keep framerate_producer in the mix (and reimplement "Using audio from multiple audio streams".

Question: The architecture in 2.1.0 depends on producers always deinterlacing their frames before they are sent to the mixer. This means that if you are playing an 1080i file on a 1080i channel without MIXER commands requiring deinterlacing, it will be deinterlaced and reinterlaced anyway. This simplifies the code a lot but will force a performance and possibly a quality penalty on interlaced material being played. Both geometric mixer transforms and framerate conversion/speed change mechanism depends on incoming material being progressive. What are your thoughts on this?

A possible solution would be for the user to be required to know before-hand if they intend to do geometric transforms or speed changes and therefore pass FILTER DEINTERLACE_BOB. In the case of framerate conversion Caspar can detect the need automatically though.

Re: CasparCG Server 2.1 status

#12
Helge Norberg wrote:
Helge Norberg wrote: Question: The architecture in 2.1.0 depends on producers always deinterlacing their frames before they are sent to the mixer. This means that if you are playing an 1080i file on a 1080i channel without MIXER commands requiring deinterlacing, it will be deinterlaced and reinterlaced anyway. This simplifies the code a lot but will force a performance and possibly a quality penalty on interlaced material being played. Both geometric mixer transforms and framerate conversion/speed change mechanism depends on incoming material being progressive. What are your thoughts on this?
People using 1080i instead of 1080p are not aiming for maximum quality anyway, if they do they would be using 1080p, so quality penalty of deinterlacing/reinterlacing should not be critical. Performance penalty on the other hand could be an issue if it is significant, how bad is it in practice?

Re: CasparCG Server 2.1 status

#14
100% agree with Jesper, the complete Broadcast World is interlaced, 1080i50/59.94. At least on the production side. See the television production at the Olympics in Rio, over 60 OB Vans with over 800 broadcast cameras from all over the world did their productions in 1080i59.94 or 1080i/50.
All P productions are mostly in the Event World or from the "filmlook" companies.
So the question is: is Caspar CG mainly a broadcast or event tool.

Re: CasparCG Server 2.1 status

#15
peteraellig wrote:So the question is: is Caspar CG mainly a broadcast or event tool.
It's main sponsor is a national broadcaster (SVT) so the question is answered.

I also think we should mainly optimize for interlace. But it would be good, if we could manage to get it to look good in either mode. Because the streaming world is mostly also work in progressive...
Didi Kunz
CasparCG Client-Programmer, Template Maker & Live CG-Operator
Media Support, CH-5722 Gränichen, Switzerland http://mediasupport.ch/
Problems? Guide to posting Bug reports & Feature requests

Re: CasparCG Server 2.1 status

#16
well i'm from event's world. but with one leg in broadcast world.

Off coarse progressive is the best way to go in the end!

(where i must say that for motion portrayal 1080p25 looks worse then 1080i25 )

But the real world is different broadcast is interlaced on the moment!
and will be for a while, as a broadcast tool CCG should do interlaced as best as possible !

and as a event's man.
When it comes to the bigger event's you will find your self handling and giving
interlaced feeds to broadcasters so please let it be the best possible

i guess 1080i will be the last interlaced standard but we have to live with it for some years to come.

LS

Re: CasparCG Server 2.1 status

#17
peteraellig wrote:All P productions are mostly in the Event World or from the "filmlook" companies.
So the question is: is Caspar CG mainly a broadcast or event tool.
You have to ask the question: Why the "filmlook" companies are working exclusively in progressive mode? - The quality of the picture of course. I'm not saying CCG should not work well with interlaced, I work with interlaced mode most of the time, just saying that the quality penalty of deinterlacing/reinterlacing should not be a big issue in broadcast where many things are "acceptable", and that performance penalty of that process is more problematic since you don't want your setup working without some reserves.

Re: CasparCG Server 2.1 status

#18
Jesper Stærkær wrote:iTod: i50/59.94 is the defacto standard for broadcasting. We must treat this as the default and optimise for it. I can not agree with your reasoning on this.
Idem... ;)
itod wrote:....just saying that the quality penalty of deinterlacing/reinterlacing should not be a big issue in broadcast where many things are "acceptable"....
Many things are "acceptable", and that's precisely one of the big problems in Broadcasting today...
The broadcast quality criteria are often neglected.... :(
Any codec, any format, any convertion process is good as long as the thing is visible and audible...This is so common today...

Friendly ;)

Re: CasparCG Server 2.1 status

#19
Great work on the 2.1 release, it sounds fantastic.

+1 for CEF upgrade.

I don't know about this sort of thing so please excuse my ignorance, but how hard would it be to make it so future updates to CEF could be a "drop-in" replacement (at the risk of the end-user obviously - it would be untested by SVT)

Re: CasparCG Server 2.1 status

#20
A lot of stuff seems to work allright with the latest 2.1 build (october 2016)
- MXF files now play with transparent audio output
- audio is not always 100% in sync, needs some more testing
- there's colorloss with file recorder with any codec (colorspace seems not right for HD)
Regards,
Theo Kooijmans
UniversalDV

Re: CasparCG Server 2.1 status

#21
Theo wrote:- there's colorloss with file recorder with any codec (colorspace seems not right for HD)
Maybe only some parameters needs to be added to the FFMPEG consumer to indicate BT.709 operation and using "not fullrange" values in the written files.

For x264 encoder this is done by

Code: Select all

-x264opts colorprim=bt709:transfer=bt709:colormatrix=bt709:fullrange=off
for example. Maybe other encoders have similar options.

Re: CasparCG Server 2.1 status

#24
I get a error when I send this
ADD 1 FILE Testcap.mp4 -x264opts colorprim=bt601:transfer=bt601:colormatrix=bt601:fullrange=off

If I send
ADD 1 FILE Testcap.mp4 -x264opts colorprim=bt709:transfer=bt709:colormatrix=bt709:fullrange=off
I get a file with color loss

IF I send
ADD 1 FILE "Testcap.mp4" -x264opts colorprim=bt709:transfer=bt709:colormatrix=bt709:fullrange=on
color is good but blacklevel/contrast is effected
Regards,
Theo Kooijmans
UniversalDV

Re: CasparCG Server 2.1 status

#26
Update with latest 2.1 build
We now can record with bt709 colorspace but...
- all h264 files always record in yuv444 format and bitrate is ignored
- all MXF files always record in yuv420, so HDcam yuv422 setting is ignored.

So in general -pix_fmt yuv422p and bitrate is ignored
Regards,
Theo Kooijmans
UniversalDV

Re: CasparCG Server 2.1 status

#29
Yes the "channel freeze" is currently the major showstopper. It even made me stop testing the nightly builds.

Personally I would prefer if compatibility with last or at least a more recent Flash version could be validated.
I am still hoping that instant replay/slow motion/play while record will be included in final release - preferable using a well known media file container format.
And at least there seem to be some issues with the PSD producer where text layers are repositioned on CasparCG output which would be nice to be fixed before final release if possible.


But so many thanks for your great work until now!

Re: CasparCG Server 2.1 status

#32
Hi, im looking for the new release of caspar cg server. Actually i have a machine working whit 2.0.6 server becouse 2.0.7 have problems whit noise and clips in the audio output. This week i need to assembly a new machine for a 4 channels playout server for a live broadcast production, but i have a question. In owr production environment we need to send to the audio console the stereo output per any video channel output. But in the server
I can not send audio from each video channel separately to the audio interface outputs. In the new server this is possible? i have many audio interfaces for doing this, usb and firewire. In the other side, in the new machine im going to have a new intel corei7 7700 processor, but this processor cant support windows 7. Caspar CG can run properly in windows 8 or 10?? Thanks for your help.

Re: CasparCG Server 2.1 status

#35
most people have positive results with Windows 10. windows 8 is out of the question. stability is not guaranteed at all though. it just happens to be mostly stable.
Last edited by balte on 10 Mar 2017, 22:00, edited 1 time in total.
CasparCG enthusiast and broadcast geek - Sometimes, I do get payed for this stuff.

Re: CasparCG Server 2.1 status

#37
I already do long multiple clip playouts and this seems to work fine with latest 2.1.0.3394 5422850 Beta 1.
(Running two HD Decklink channels on one i5 Intel system for several weeks...)
Mayor show stopper is however the MIX command and the File recorder. (please see reports Github)
The MIX command causes bad offsync between video and audio, (PLAY 1-1 "my clip" MIX 50)
The File recorder ignores all options.
Regards,
Theo Kooijmans
UniversalDV

Who is online

Users browsing this forum: No registered users and 1 guest

cron