View Full Version : RealVideo 9 Information Thread Discussion
Pages :
1
2
3
4
5
6
7
8
9
[
10]
11
12
r0cket
7th August 2003, 05:59
sorry, didn't find it in the forum or in documentation to producer, but is there any way to force constant framerate in real media files? vdubmod only handles this types of streams but i always get some dropped frames (original - 25 fps, result - 23.1 fps).
Sirber
7th August 2003, 12:51
RV9 is Variable FrameRate. At encoding set Max Framerate to 30.
kilg0r3
7th August 2003, 15:01
yep, and videoquality to normal or below
Sirber
21st August 2003, 00:27
RealAnime v1.12
*** New version with CPU Fix ***
http://forum.doom9.org/showthread.php?s=&postid=336534
midiguy
24th August 2003, 02:22
lol.
Sirber, you claim that it is 30% faster, but really, the bug was making it 30% slower. M$ style marketing, I like. :D
iwod
1st September 2003, 11:52
long time no post, ever since the release og EHQ and Dup......
The new features in Producer 9.2 allow user to customize packet size, however the maximum size allowed is only 16K, due to a stupid bug in Decoder. I.e if we want to use anything larger than 16k then we will break the bitstream. And if real were to release a new decoder that will coz too much hassel........... :angry:
Anyway i am wondering will there be any quality improve if the packet size was unlimied or no packetasation.
And i am wondering if there is any chance that the decoder for Rv9 could be optimise further........ seeing Divx 5.1 has done a job in these area to allow people with slower computer to view it.........
karl_lillevold
1st September 2003, 17:06
there is no improvement from larger packet sizes than 16K, so there is no reason to force decoder upgrades for that reason. BTW, there is little hassle for RealNetworks to release a decoder upgrade, it's a hassle for the average end user :) The bug has been fixed though, and will be released along with the next major release.
However, all that does not really matter, because the encoder currently at least splits a frame in two packets, to enable decoder multi-threading, where the two threads decode the packets in parallel. In many cases, this makes the packets less than 16K in size anyway.
With regards to decoder optimizations:
1) I had an interesting experience on my PIII 700 MHz laptop. I found I could not play back 640x352 XviD using ffdshow. It was playing 18 fps at 100% CPU load. So I installed DivX 5.05, changed to 4CC to DIVX and now it plays at 50% CPU load at 24 fps.
2) On this same 700 MHz laptop, RV9 plays at around the same CPU load for the same resolution. I found that an old PII 450 MHz laptop was just the limit for playing 640x272 RV9.
3) Every CPU intensive function in RV9 has been heavily MMX and SSE optimized, and lately we have spent a lot of time to optimize for ARM processors on WinCE / PocketPC devices. Further PC optimizations would be possible, but the impact would be very small, and would take away time from future video quality improvements, since we are only 2 engineers in the video codec team. Which would you prefer? To enable slower than 500 MHz PCs to play RV9, or improved video quality for everyone else? :eek:
TFC
1st September 2003, 17:22
For sure i prefer u improve image quality!!!!! :D
SeeMoreDigital
1st September 2003, 17:24
Hi Karl
With regards to decoder optimizations:
1) I had an interesting experience on my PIII 700 MHz laptop. I found I could not play back 640x352 XviD using ffdshow. It was playing 18 fps at 100% CPU load. So I installed DivX 5.05, changed to 4CC to DIVX and now it plays at 50% CPU load at 24 fps.
Are you still using ffdshow for playback or did you use the DivXPlayer v2.1?
I only ask because some time ago I noticed that WMV9.wmv files use up more CPU power than WMV9.avi files. Infact WMV9.avi files appear to use less CPU power than DivX and RV9 (non EHQ) files!
I've not been able to do CPU load tests on very many PC's. So could you try could you have a go with your laptop please?
Cheers
iwod
1st September 2003, 17:34
Of coz improve quality......
But i am curious......... apart from better VBR, what else could be improve??
This may be a silly questions....... what will happen if we don't have packet at all in RV9?? ( apart from not streamable )
karl_lillevold
1st September 2003, 17:35
Originally posted by SeeMoreDigital
Are you still using ffdshow for playback or did you use the DivXPlayer v2.1?
I only ask because some time ago I noticed that WMV9.wmv files use up more CPU power than WMV9.avi files. Infact WMV9.avi files appear to use less CPU power than DivX and RV9 (non EHQ) files!
I've not been able to do CPU load tests on very many PC's. So could you try could you have a go with your laptop please?
I did not install DivXPlayer.. Why would I want to..? I just installed the DivX codec (and removed Gain), and still play in my favorite media player : Media Player Classic.
I don't have wmv9.avi and wmv files that are the same content and bitrate so I can not run that test, but if there is a difference I would suspect it to be caused by the player choosing different renderer mechanisms, possibly with different levels of postfiltering.
karl_lillevold
1st September 2003, 17:39
Originally posted by iwod
But i am curious......... apart from better VBR, what else could be improve??
Well, EHQ helped, right..? Nothing specific to report yet, but my colleague came up with a cool idea that looks promising.
Originally posted by iwod
This may be a silly questions....... what will happen if we don't have packet at all in RV9?? ( apart from not streamable )
Nothing. (maybe 0.1% improved compression efficiency, but you would lose the advantage of dual threaded decoding).
SeeMoreDigital
1st September 2003, 17:58
Originally posted by karl_lillevold
I did not install DivXPlayer.. Why would I want to..? I just installed the DivX codec (and removed Gain), and still play in my favorite media player : Media Player Classic.
I don't have wmv9.avi and wmv files that are the same content and bitrate so I can not run that test, but if there is a difference I would suspect it to be caused by the player choosing different renderer mechanisms, possibly with different levels of postfiltering. Yes, I like to use MediaPlayer Classic too. I was using this player when I noticed the WMV9.avi and WMV9.wmv CPU issue. I then tested using the WinMediaPlayer9 but still the .avi file ran more smoothly that the equivalent .wmv file!
Never mind. It will have to remain a mystery!
Cheers
daiyam
2nd September 2003, 21:32
use graphedit for show the different renderer mechanisms used by the player.
DKDIB
3rd September 2003, 08:59
Or GSpot (http://www.headbands.com/gspot/download.html): even if it doesen' t support a container, it show the graph correctly.
alx
3rd September 2003, 10:35
Really thanks for all your support Karl, i think RV9 has a very good present and an excellent future!!!
As soon as tools become available to work with RV9 files, no more Di...what??? Xv...who??? ;) (just a joke, please donīt kill me)
One question, is there any way to extract the audio, or the video to another file from an .rmvb???
Keep the good work.
Alx
Sirber
3rd September 2003, 17:01
yes
RMVB --> Matroska works using mkvmerge or graphedit
alx
4th September 2003, 06:21
Sirber, thanks for your reply, but maybe u donīt understand me or viceversa.......I meaning extract a .rm audio file from a .rmvb which has 3 .rm inside.
Any ideas?
Alx
karl_lillevold
4th September 2003, 06:39
hi alx, there is no tool to directly demux an RM file with multiple tracks into several RM files. Until now, this has rarely been requested, because there was no way to then re-mux these files. However, if you trans-mux the RM file into Matroska, maybe there is a way to then de-mux this Matroska file into multiple files, for instance one MKV and several MKA files. I just tried 'mkvextract' with a Matroska file with RealAudio + RealVideo, but was not able to get it to work.
L:\>mkvextract tracks atest.mkv 1:atesta.mka 2:atestv.mkv
Warning: Unsupported CodecID 'A_REAL/COOK' for ID 1. Skipping track.
Warning: Track ID 2 is missing some critical information. Skipping track.
Nothing to do. Exiting.
so this may not be possible yet. I don't know Matroska or its tools well enough to answer for sure though. Maybe you could try to ask on the 'mkvtoolnix: new version' thread.
I will ask internally as well. Maybe there is some developer/Helix tool that can help.
alx
4th September 2003, 07:55
As usual you came into my help karl.......i chose a very bad example!!! :(
I made a test with AutoRV9 which gave me a .rmvb file with a video stream and audio stream inside, and i want to test same video with an .mp3 file to compare sizes and quality.......but, then....i realized that i donīt how how to extract the video stream (of course i deleted the temporary video stream ;)
It is possible to extract one stream from a .rmvb file??? any stream? .rm audio or video?
Thanks
Alx
PS. Hope this "real case" does not bring any new confusion :)
karl_lillevold
4th September 2003, 20:59
@alx and everyone using RM files: There is a very powerful tool available via the Helix Community. As far as I know you currently have to check out source code and build it yourself, but I will see what I can do...
The tool is called dtdriver, and here are its options:
Version 6.0.9.2-Release
usage:
dtdrive [+|-<Options>] [<Commands>] {[[<FSName>,]<Protocol>://]<File>}
Options: (+ = add, - = remove)
p = dump packet content
h = dump headers/processing stages
v = verbose (show all buffer contents)
u = uncompress (decode) content
c = decrypt content into unprotected (clear) form
f = when decoding filter-out any streams unable to decode
s = when decoding maximize speed on detriment of standards
i = when decoding, set source properties in outgoing stream header
= = when decoding, have decoders simply pass through data
q = when decoding, block any audio streams
y = when decoding, block any video streams
e = dump event counters
w = dump phase execution times
d = if writing a file, recompute duration
r = if writing a file, recompute preroll
o = if writing a file, overwrite the existing file
t = if writing a file, use relative timestamps
Commands:
-L <Path> = load plugins from specified path
* overrides HELIX_LIBS environment variable setting
-F <File> = place output into specifed file
-R <Rate> = read only packets for the specified transfer rate (bps)
-J <Time> = skip to specified media time (ms)
-S <Mode> = run in server emulating mode
0 = run as client (local playback)
1 = run as server (streamed playback)
-W <File.Ext>= write out file in format associated with extension
-SL <MBytes> = set the file size limit for written out files
-D <DirPath> = use the specified directory for intermediate files
-T <Time> = fetch packets for the specified amount of time (ms)
So, in order to de-mux an A/V RM file, all I had to do was:
dtdrive +u=q l:\atest.rmvb -W video.rm
dtdrive +u=y l:\atest.rmvb -W audio.rm
It works great. Pretty cool!
alx
5th September 2003, 01:35
Karl, thanks (again) to enlight us in this arena...As you probably know, i am not a programer, just a fan an enthusiastic person about movies, so i can NOT compile the dtdriver source....if you or anyone here can, please let me/us know where we can get it.
Any news about milestone 9.3 with "video.stats generator ;)"??? I recently bought "Titus" with Anthony Hopkins and its an 1:33, interlaced, 160 minutes movie, which left my machine unusable for about 30 hours :( .......sniffff......i think Titanic or Red Rose have to wait to milestone 9.3 ;)
Thanks again Karl
Alx
RadicalEd
5th September 2003, 04:22
Pardon my ignorance Karl but I can't find dtdriver on helix community for the life of me. Where's the source code at?
karl_lillevold
5th September 2003, 05:09
Once you have set up the Helix build environment, and run the 'build' command, with the public BIF branch 'helix', you can select Target 'dtdrive', and check out source and/or build this target. The problem is I am not sure this builds the RM writer module, which is required for the above functionality, and other functionality related to writing out RM files. I will investigate further when I have time, to see if this can be built from public binary libraries, since this tool is very useful indeed.
EDIT: use profile 'helix-client-all-defines'
[For any Helix build environment questions {what is a BIF branch}, cvs, python, compiler problems etc, this is not the right forum. Please read the relevant FAQs on the Helix site. Most questions are indeed answered there. Then try the Helix IRC channel. Most of the time there is someone there that knows. Or you could try a Helix mailing list]
karl_lillevold
5th September 2003, 06:26
ah, the missing RM writer module, rmwrtr.dll, is available in the plugins directory for Producer. So until a binary release of dtdriver can be made available, it can be built from public source code, target dtdrive. Then download the latest Producer Milestone in the Binaries section, and copy its rmwrtr.dll into the directory where dtdrive.exe and its DLLs reside (when building 'release', this is the 'release' folder). I just built this from scratch on my home system, added rmwrtr.dll, and now I can de-mux an A/V RM file into separate A+V RM files.
Sirber
7th October 2003, 22:53
Hi
I wonder what gain in speed I would get if I was moving from my actual machine (see my signature) and this computer:
http://sirber.no-ip.com/~sirber/webernic.xls
for RV9 encoding...
deXtoRious
8th October 2003, 19:40
@sirber
A bit offtopic, but do you think that the 266 Mhz RAM (DDR-2100) is actually better (for the particular configuration of course) than 333 Mhz (DDR-2700)?
Sirber
8th October 2003, 21:33
Main PC: AMD Athlon XP 2000+, GeForce 4 TI4200 128MB AGP8X, 512MB DDR@266 Low Latency, GA-7VAX VIA KT400, 80GB HDD, Windows XP SP1a
running at 266 already :D
bilu
15th October 2003, 18:20
Are there any guides available that gather all the info needed to start doing some kickass RV9 encodes? :D
(including installation procedures and packages,please.)
Bilu
bilu
17th October 2003, 15:50
Is it possible to play RV9 with Vobsub subtitles using Gabest's Realmedia DShow Splitter or something?
Bilu
Sirber
17th October 2003, 16:35
You need to enable "Special Renderer in MPC"
karl_lillevold
17th October 2003, 16:38
VSFilter works very well for .srt and .ssa subtitles. Great visual quality. MPC even comes with RealMediaSplitter built in. Just make sure you set the playback to "DirectShow" in the pull-down in the Formats menu, instead of "RealMedia".
karl_lillevold
17th October 2003, 16:41
try to avoid the Special Renderer. It adds a lot of CPU overhead and introduces tearing. The DirectShow method is strongly recommended. This corresponds to RealMediaSplitter.
bilu
17th October 2003, 17:14
I'm now running RV9 under Zoom Player with raw Vobsub subtitles :cool:
Just needed to "C:\WINDOWS\system32>regsvr32.exe RealMediaSplitter.ax" :)
Bilu
bilu
17th October 2003, 18:05
@karl
Does Anamorphic resize work with the DShow filter on other players?
Bilu
karl_lillevold
17th October 2003, 18:16
It does not look like anamorphic RV9 works in mplayer2, even though it works great in MPC/RealMediaSplitter. I don't know DirectShow well enough to guess why. You should check with the Zoom player developers and/or the thread for RealMediaSplitter. You can use
http://www.lillevold.com/files/atest.rmvb
http://www.lillevold.com/files/atest.mkv
as test files, if you like.
bilu
17th October 2003, 18:31
I had already downloaded atest.rmvb before, but hadn't noticed it was anamorphically encoded.
It doesn't work with Zoom Player, probably because of the same reason that stops FFMPEG MPEG-4 encoded movies aspect ratio flag to work on Windows players.
On my PIII-500 laptop this test video doesn't run as near as smooth than under Real Player. But t01_ehq.rmvb from your site works fine, so it's probably related to bitrate :confused:
EDIT: Any chance to use DShow filters with Real Video on Real Player (optional of course)? ;)
Bilu
gabest
17th October 2003, 19:25
Originally posted by karl_lillevold
It does not look like anamorphic RV9 works in mplayer2Try it in fullscreen, wmp64 cheats there.
bilu
17th October 2003, 22:21
@Gabest
I could only work with raw Vobsub on Zoom Player.
tried your MPC and WMP9 without success.
Also the movie kept stuttering everytime it buffered a subtitle.
I was never able to go to the Filters section during RV9 playback, even after choosing DirectShow on MPC.
And your special rendering method for subs under RV9 made the movie look like both a bad resize and a slideshow, but played the subs :)
Enough of bad news :D Any hints?
Bilu
gabest
17th October 2003, 22:32
Is the "framestep" button enabled? It should be in dshow mode then. And when it is set to dshow (options/formats/realmedia/2nd column), it will never try to use real's api, it would report some error instead of opening the file.
fyo
17th October 2003, 22:36
Was any solution found to the frame-dropping problem RealOne "Rakaz" had a few months ago? (this thread: http://forum.doom9.org/showthread.php?s=&postid=333689&highlight=problem+playing#post333689 )
I am having a very similar problem when playing back rv9 content:
The video (and audio) will freeze and then continue some frames later. The frames are dropped (and listed as such in the statistics) and CPU usage DROPS while the video is frozen. CPU usage is normally at around the 35-45% mark for the files I've tested with (1.1Mbps), very rarely over 50%.
The odd thing is that the playback is absolutely perfect when I use the Helix DNA Client instead (however, as a "media player" I find it fairly useless). Using Media Player Classic reproduces the effects of the RealOne player.
System: WinXP, 700MHz Athlon, 384MB RAM.
TIA,
fyo
bilu
17th October 2003, 22:50
Originally posted by gabest
Is the "framestep" button enabled? It should be in dshow mode then. And when it is set to dshow (options/formats/realmedia/2nd column), it will never try to use real's api, it would report some error instead of opening the file.
You were right :) I hadn't realised how to change from Realmedia to DirectShow, just changed in the "rstp://..." option :D
Now it shows up the same problem as Zoom Player: it delays the picture while buffering the subtitle and then accelerates the picture to compensate. :confused:
Bilu
bilu
17th October 2003, 22:55
@fyo
Have you tried on MPC to:
- change between "DirectShow" and "RealMedia" on RealVideo (2nd column) in the Formats menu;
- disable special renderer under the RM/QT tab?
Bilu
gabest
17th October 2003, 22:57
Which subtitling mode are you using, mpc's or dvobsub? Is the buffering set to use 0 textures? (options/subtitles) This is the same as dvobsub's "prebuffer subpics" options, just can be given in exact numbers, dvobsub always uses 10 at most.
edit: The next release of mpc will have one dedicated page to choose the renderers... I'm already working on it :P
bilu
17th October 2003, 23:07
I was using DVobsub's.
The MPC method seems to work only when I select Realmedia? :confused:
It doesn't show up in Filters when using DShow.
EDIT: and I also got stuttering everytime I use DirectShow instead of Realmedia, even now without DirectVobSub running. Only Filters show are the default DShow audio and video filters.
Bilu
fyo
17th October 2003, 23:59
Bilu:
Have you tried on MPC [...] to change between "DirectShow" and "RealMedia" on RealVideo (2nd column) in the Formats menu;
Made no difference. The freezing still occurs, although enabled "View Statistics" now works, but the number of frames dropped seems unrealistic - on average 50% of ALL frames are said to be dropped, which doesn't LOOK like it's happening (clarification: stats claim frames are being dropped continously, in addition to the freezes). Using RealOne, no frames were reported as being dropped, apart from the freezes.
"Special renderer" was never enabled. Post-processing in RealOne is disabled.
I tried to verify my DNA Client tests, but am now unable to get it to play back any video file at all. Exact same install, nothing changed. Same command line parameters. Same everything. Really odd. Error code is "80040016", which is apparantly a "missing filesystem error" (missing some dlls or the like - which doesn't appear to be the case, btw).
One other thing I've noted by testing a bit more: The freezing occurs in the exact same positions each time (although the number of dropped frames varies significantly, which was what prompted me to attempt a verification of my DNA Client experience).
TIA,
fyo
karl_lillevold
18th October 2003, 00:11
@fyo: someone in the netherlands actually had an error in the h/w timer on his Athlon / nForce2 based system causing random freezes in a similar manner, but this would also affect the Helix DNA client, I think.
See this thread (http://forum.doom9.org/showthread.php?s=&threadid=57343&highlight=nvidia), just in case.
bilu
18th October 2003, 00:26
I have a PIII-850 with a Geforce 2 MX400 32MB and have a similar problem.
I've sent you the tick.txt file now.
Bilu
fyo
18th October 2003, 01:24
karl:
I don't have any negative jumps, but I do have some rather odd "noise". Jump is almost always 10 ticks, but all of a sudden this happens:
5811506 10 5814176 10 5814176 10 5815061 10
5811516 10 5814186 10 5814186 10 5815071 10
5811536 20 5814204 18 5814204 18 5815089 18
5811606 70 5814281 77 5814281 77 5815166 77
5811776 170 5814447 166 5814447 166 5815333 167
5811786 10 5814458 11 5814458 11 5815343 10
5811796 10 5814468 10 5814468 10 5815354 11
5811816 20 5814481 13 5814481 13 5815366 12
5811826 10 5814493 12 5814493 12 5815378 12
5811846 20 5814520 27 5814520 27 5815405 27
5811917 71 5814590 70 5814590 70 5815476 71
5811937 20 5814606 16 5814606 16 5815492 16
5811997 60 5814661 55 5814661 55 5815547 55
5812017 20 5814687 26 5814687 26 5815573 26
5812047 30 5814717 30 5814717 30 5815603 30
I do have an NVIDIA graphics card... a GeForce256DDR (aka a GeForce1).
TIA again,
fyo
bilu
18th October 2003, 01:32
Just like my problem! A significant piece, including a big glitch and other more frequent minor glitches:
3205459 10 3206460 10 3206460 10 3206949 10
3205469 10 3206470 10 3206470 10 3206959 10
3205479 10 3206481 11 3206481 11 3206969 10
3205489 10 3206490 9 3206490 9 3206979 10
3205499 10 3206500 10 3206500 10 3206989 10
3205509 10 3206514 14 3206514 14 3207002 13
3205519 10 3206524 10 3206524 10 3207013 11
3205529 10 3206532 8 3206532 8 3207020 7
3205549 20 3206557 25 3206557 25 3207045 25
3205589 40 3206591 34 3206591 34 3207079 34
3205609 20 3206612 21 3206612 21 3207101 22
3205669 60 3206670 58 3206670 58 3207159 58
3205679 10 3206680 10 3206680 10 3207169 10
3205759 80 3206761 81 3206761 81 3207249 80
3205769 10 3206776 15 3206776 15 3207264 15
3205779 10 3206786 10 3206786 10 3207275 11
3205789 10 3206795 9 3206795 9 3207284 9
3205799 10 3206805 10 3206805 10 3207294 10
3205809 10 3206815 10 3206815 10 3207304 10
3205819 10 3206825 10 3206825 10 3207314 10
3205829 10 3206835 10 3206835 10 3207324 10
3205839 10 3206845 10 3206845 10 3207334 10
3205849 10 3206855 10 3206855 10 3207344 10
3205859 10 3206865 10 3206865 10 3207354 10
3205869 10 3206875 10 3206875 10 3207364 10
3205879 10 3206885 10 3206885 10 3207374 10
3205889 10 3206895 10 3206895 10 3207384 10
3205899 10 3206905 10 3206905 10 3207394 10
3205909 10 3206916 11 3206916 11 3207404 10
3205919 10 3206925 9 3206925 9 3207414 10
3205929 10 3206935 10 3206935 10 3207424 10
3205939 10 3206945 10 3206945 10 3207434 10
3205949 10 3206955 10 3206955 10 3207444 10
3205959 10 3206965 10 3206965 10 3207454 10
3205969 10 3206975 10 3206975 10 3207464 10
3205979 10 3206986 11 3206986 11 3207474 10
3205989 10 3206996 10 3206996 10 3207484 10
3205999 10 3207005 9 3207005 9 3207494 10
3206009 10 3207015 10 3207015 10 3207504 10
3206020 11 3207025 10 3207025 10 3207514 10
3206030 10 3207035 10 3207035 10 3207524 10
Bilu
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.