This forum is for users to exchange information and discuss with other users about a TMPGEnc product.
In case you need official support, please contact TMPG Inc.
Pegasys Products BBS [ Sorted by thread creation date ]
Has anyone run TMPGE in a dual processor system and gained significant performance? I am considering a dedicated box for TMPGE and want to do a lot faster MPEG2 encoding. My current P4 1,5GHZ 512 Meg RDRAM Dell takes about 3 hours to encode an hour of video. How do I get a lot faster speed?
Hey,
I'm running dual-PIII 1.0ghz in Windows 2k. I enabled a few settings in the .ini file that seemed to change the way that this program interfaces with dual-cpus, and I am getting full 100% CPU usage, so I suppose it's working. Help on this topic seems to be a bit slight, to say the least. As far as speed, before I enabled those extensions, it was looking like about 5 hours for Goodfellas, now it's down to 3.5 hours, so I'd say it helps a lot.
My P-4 Class Intel Celeron at 1.7GHZ encodes VCD"s at Real time, and with CCE I can encode AVI files at % times Real time but my old P-# 800mhz would take 8-10 hours to do a 2 hour movie, Pluss P-$"s have SSE-2 which p-3"s don"t so that P-3"s are faster at encodeing that P-4"s is Not True At All...
Minion, Last time I checked TMPG Enc didn't support SSE2, which means you have the floating point processor equivilent to a P-III 1.0 or so, it you are lucky. Check out SIS Sandra and benchmark your CPU without SSE2 and see what I'm talking about.
P-III may be better "IF" we compare at same clock(GHz), however, there is no P-III which is over 1.4GHz, P-4 is now reaching to 2.8GHz. Obviously, P-4 2.8GHz is most easiest way to archive faster encoding speed.
TMPGEnc has supported SSE2 one year ago, and has been updating almost monthly, since then P-4, is faster than P-III if we use TMPGEnc appropriately.
Yeah, the Benchmark on dvdboard.de is the biggest and bestest Benchmark ever done. It test's what realy matter to us: Encoding speed.
BTW: On my P4, VCD is encodet much faster than Realtime with TMPGEnc (0.6 to 0.8).
Half-D1 MPEG2 is between 1.8 and 2.5 (depants on if it is interlaced or not and the settings).
CCE encodes Half-D1 and SVCD faster than Realtime.
>Minion, Last time I checked TMPG Enc didn't support SSE2, which means you have the floating point processor equivilent to a P-III 1.0 or so, it you are lucky.
It must have been a long time since you last checked then because as far as I am aware SSE2 has been supported by TMPG since quite a few versions ago unless that checkbox for SSE2 under the CPU tab is just for show.
As for the P3's are faster than P4's comments, well that's a load of crap too. As minion says the the new versions of TMPG are built to take advantage of the new streaming extensions in the P4 which will boost encoding speed. Take a look in your TMPG folder you will see a file called P4Package.dll. This file is not for show, it does actually do something in P4's.
Also I used to have a P3 500 which would take 10 hrs+ to encode a 1 1/2hr movie. Now with my P4 1.7 it takes less than real time to encode from a .d2v file, usually about 1hr 10 mins and even less from an AVI which by my calculations is nearly 5x faster than the P3 500 even though the P4 1.7 isn't a 5x faster processor, so there must be some sort of optimization within the P4 which is making things encode faster apart from the processor speed.
I've captured a 45 minutes clip from TV using Pinnacle PCTVPro/SBlaster Live! and VirtualDub 1.4, then I've edited the file and saved it as an AVI file.
After I've created a VCD compliant Mpeg clip with TMPGenc 258.
The video is fine, but the audio disappears after about 2 minutes of playtime.
Played with WMediaPlayer, the mpegfile has the same problem, but if I play the AVI clip with VirtualDub, the audio is fine.
In encoding TMPGEnc reports indeed a "division by 0" error in the log, however it finishes the job producing good quality Video but the above problem on the audio. Has anybody an idea on why this can happen?
What format did you capture the Audio to???Allways Capture to WAV, and capture at 44100hz...and if you did this you can allways use something else to encode the audio to Mp2, like "Headac3e"...
thanks, the audio was captured at 44 Khz, 16 bit PCM uncompressed.
Yesterday evening I apparently sorted the problem, extracting the WAV file from VIRTUALDUB and then encoding with TMPGENC using the original video and the extracted audio track rather than the embedded track.
I could get some out-of-sync in this way, but still I do not know cause I have now run the compressor at the highest quality setup and it will take 9 hours before my old K6-2 will output the final file.
Ciao
I have been reading a lot lately about Mpeg encoding and am curious to how Tmpgenc handles Motion Search precesion on the second pass.
It is clear to me the allocation of bits on the second pass becomes more accurate i.e. Bits are taken away from less complex scenes and given to more complex scenes in order to maintain a more contant quality.
But is the motion search precision also improved on the second pass?
For example say I encode a sample with 2500 CBR
Now I encode the same sample again with 300 min 2500 Average and 2500 Max, 2 pass VBR.
Will I gain anything by doing the 2 pass VBR with the average and max the same.
Will the motion search be more accurate? Is it possible I will introduce new distortions?
The Motion Search Settings are used for both passes. You can't change that.
In the First Pass, TMPGEnc looks what Bitrate is necessary for a given Q-Factor. In the second pass it changes the Q-Factor dynamicly to get the given AVG-Bitrate.
This has nothing to do with the Motion Search Precision. This thing says TMPGEnc how to look for Motion in the Video.
There is no reason to do multi-pass VBR unless the average is much less than the max. And this in only useful on long videos with containing a lot of low motion shots lasting many seconds.
2-pass VBR on most music videos, for example, would be senseless.
Thanks for the replys, but the question I am trying to ask is whether the motion search precesion is improved on the second pass or just bitrate allocation?
Hey, i gave you the answer in my first posting.
Motion Search precission has nothing to do with the Bitrate Control Mode.
The second pass is for optimizing the Bitrate used for every GOP, nothing more and nothing left. Every Encoder is acting like this.
If you change the Type of Motion Search after the first pass, all the statistics, the encoder has calculated did not longer match to the new methode.
Motion Search is the process for recognize Motion in the Movie. The methode discripes, how to search motion. The Ammount of found Motion is important for things like Quantisation, so you get other Quantisation-Levels if you use an other Search Level. Theoretical it is posible, to look for Motion Optimisations, but NOT in only 2passes.
Maybe a Encoder can look for Motion in the First pass, then try to verify this in a second pass. A third pass will look for the Quantisation-Levels and a fourth one will do the Encoding-Job.
What I mean is not to change the type of motion search in the second pass, I mean is it optimised in the second pass. ie is the motion detection re calculated/re checked in the second pass.
or is only the bitrate optimised in the second pass.
Your not a pain, it's my terrible bad english that doesn't let me explain things like 2pass VBR realy clear.
OK, here we go: The second pass is for optimisation of the Bitrate for each GOP and each frame. The motion search will be done in the first pass and in the second pass, and there is no chance to change it or optimize it in the second pass. All Quantisation Levels, calculated in the First Pass, are the result of the Pictures and the Motion. Like i said before: if you change the motion search precission, you change every thing.
But, if the encoder is able to use a higher Bitrate for a GOP, he isn't forced to use a high Quantisation Level for motion compensated Macroblocks. This has a very familiar effekt compared to optimized Motion Search.
If you do 2pass with equal AVG and Max, you are doing the same like 1pass, but in the double time. You need a clear difference between AVG, Min and Max do get good results with 2pass.
A Min-Bitrate of 300 kbps will cause problems with some players. To be on the right side, never use Min lower than 1374-Audiobitrate. Some Players are having a small Buffer, an if the Bitrate is to low, thes will stuttering.
It lowers also the average Quality of your Video, if you use to low Min-Bitrates.
If you real want an AVG of 2500kbps and you don't want to left the Standard, use CBR.
But if you need an AVG of, for an Example, 2100 kbps, use Min 1850, AVG 2100, Max 2520 (with 192 kbps for Audio).
And Again: The Bitrate control mode has nothing to do with die Motion Serach accuracy.
OK, there's nothing more i can say about that (and for the German Readers: "Jetzt mag ich auch nicht mehr" ;) )
how do u demux or extract the audio from an asf file. I want a wav file that i can add to tmpegenc so that it doesn't ruin the sound when i convert my asf to vcd
Well you can try this, you can re-name the ASF to WMV, then use "DB-Power AMP" to encode the audio to a WAV file, you just need to make sure you Download the "WMA" Codec...Or you can download "Virtual Dub ASF" and use that to extract the audio.. you can get "DB Power AMP" here : http://dbpoweramp.com and you can get Virtual Dub-ASF here: http://www.apachez.net in the tools section.....
ohhh yes the wma codec...oops. ok woohooo now it works! BUT today tmpgenc can't open my asf file! it did it the other day after i repaired it with asf tools but not today?? Even with high priority for direct show... strange. it opens other asf files. Virtual dub c won't open it either.....says it can't find wmv1 codec. i load wmv8 video codec on my system today and the asf files plays worse! (random blocks now and then)...can igo back to 7? hmmm asf files are tempramental lol ps this asf files is played in mp 6.4
Hi, I have converted a DVD to MPEG1 files. Here are the properties of my MPEG1:
Size - 352x288
bitrate - 1150000bits/sec
framerate - 25.00 frames/sec (or 29.97, an alternative I've created)
Using TMPGEnc 2.57.41 to encode to VCD, the VCD cannot be read or sometimes cannot be read nicely and the image displayed are distorted all over, cannot be view at all. My MPEG1 files are working though.
Can someone help me to solve this problem so I can have my VCDs? Thank you.
If the Mpeg files play fine on your Computer then the problem is not with Tmpgenc, but Probably like Lupy said The way you are burning them, or your Player...I would like to Know how you "Created" a 25 FPS and a 29.9 FPS File from the Same DVD??
Hi there, i am a newbie using this software ... was converting a avi to mpeg file and when complete ..how come the size of the film (was a 702mb size film balloon to 1.02mb) can anyone help me with it and i am looking for the user manual to understand the function but can't find it .. pls help
i was converting an asf file to an mpeg-2 file. everything went smoothly until it froze and stopped. message appeared was "Index of same line is out of range (240)." is there a way to fix this?
I think the header in the ASF file is corrupted, this is very common with ASF and WMV files, what might work is to download "ASF tools" and use the Repair feature to correct the problem with the ASF file...
I use hauppauge wintv to generate a mpeg2 file, now I want to put on VCD. When I use the wizard I get no errors, but there is no sound to the output file. (the video looks great) can any one tell me what I am doing wrong?
Thanks,
james
Tmpgenc does not like to encode Compressed audio formats but you don"t need to encode the audio anyway cuz it ia allready in the Correct format for VCD, so you go to the "Mpeg Tools" to "Simple De-multiplex" and load your mpeg file in and run it, it will make a "M2v" video file and a "Mpa" audio file, you encode just the video the mpeg1, then after go back to the "Mpeg tools" to "Simple Multiplex" and load the Video and audio from the Mpeg2 file, then it will join the audio and video into a Mpeg1 vcd file.....
Sorry, but I don't even know how to look for this.
I am using vdub to cut both audio and video, tmpgenc to encode, vcdeasy to burn. In most cases, all goes without a hitch until I watch in my standalone dvd Sony dvd player. I have vcds that I made before with older versions of tmpgenc and vdub burning with ezcd that play great, with NO problems what so ever. The vcds that I am creating now with the newer versions and vcdeasy are giving me the flux. The first half of the movie plays great. The 2nd half is glitchy, freezes and starts again or just plain freezes. Why does this happen and is there anything I can do to make this not happen. Also, I have a version of tmpgenc 12a..will there be conflicts using the vdub 1.4 and vcdeasy with it?
Thanks again for the help
Well getting a faster Computer is about the only way to speed things up, you can put the "Motion Search Precition " on "Very Fast but you will loose Quality..You must have a pretty Slow CPU if it takes 40 hours, even My old 800mhz only took about 10 hours to do a 2 hour movie on High Quality...
If make sure your input source ratio is set at 16:9 and the output ratio is set at 4:3 then the borders will be added automatically to create the 4:3 display.
As for your PC taking 40 hours, you either have an old dinosaur of a PC or your settings are wrong. I used to have a P3 500 which would take about 10-12 hours to encode a movie. I now have a P4 1.7 O/C at 1.9 and it takes about 2-3 hours for TMPG to encode a movie.