VCD to VCD gives fuzzy images

[pdv1956]

I have 3 sources for TV capturing: EyeTV on te MAC and 2 PCI cards for windows/linux. I used to put everything thru TMPGEnc on windows to be sure that all my movies have the same characteristics, so i can watch them by MediaMVP on my TV.

When i use FFmpegX to do the same, VCD to VCD and Audio to MP2/192/48, the images gets very fuzzy. subtitling which is used a lot in our country is almost unreadable. I can see no difference between ffmpeg and mpeg2enc, nor by changing Qmin.

I f I use passthru i get a very large avi file, about 13GB/hour, strange ???

I took some avi's from internet, and they give nice sharp VCD output, so probably I'm doing something wrong, but I cannot find what.

my parameters: PAL, VCD, 352x288, bitrate 1150 (default 1138, why ?), audio 192/48/stereo/cbr/normalize, filters nothing changed, options profile standard vcd, qmatrix standard, interlace not interlaced, thread normal, no split, no author, altivec, bicubic, letterbox, constant bitrate, quicktime.

it seems worse after registration

[major]

If the avi sources give sharp VCD, and your VCD source gives fuzzy VCD, then probably the problem is in the source. Is it perhaps interlaced?

[pdv1956]

If the avi sources give sharp VCD, and your VCD source gives fuzzy VCD, then probably the problem is in the source. Is it perhaps interlaced?

yes, the VCD is MPEG1 from EyeTV interlaced. but when i started using FFmpegX i was handling exactly the same kind of TV recordings and they seemed sharper than now, as i said, before registration, which canot have anything to do with it.

[major]

I wonder, if your source is a VCD, why do you need to encode it again to VCD? You could just author the .mpg file as a VCD disk image by using the Author Tool.

[pdv1956]

I wonder, if your source is a VCD, why do you need to encode it again to VCD? You could just author the .mpg file as a VCD disk image by using the Author Tool.

i dont need a VCD on disk, i need VCD (or DVD) formatted MPEGs to watch them as disk files thru MediaMVP to my TV.
the 'need' to do encoding lies mostly in audio, i cannot define mp2/192/48 in some of TV recording hard/software, so demux, convert audio, mux would we the way to go, but takes a lot more key presses and mouse clicks than saying: do this vct-to-vcd.
the other reason to do it, is that but always converting with the same encoder i am sure that quality will be the same.

the trangest thing is that my initial test with this vcd-to-vcd encoding gave much shaper pictures, so i am searching to find out what settings i may have corrupted.

[Cornucopia]

There are a number of fallacies in your statement:

#1 - You don't say WHAT formats your "TV recording" software uses, but you do know already that it is NOT mp2/192/48 (I'm assuming that means Mpeg1Layer2, 192kbps, 48kHz--didn't say mono or stereo, etc). What you've already got is VCD disc. If it was done according to spec (I'll assume), the audio must be Mpeg1Layer2, 224kbps, 44.1kHz, stereo. Of course, you don't say WHY you might want it any other way. Editing? Use the right tools (cuttermaran, etc).

#2 - You and another poster mentioned "interlacing". VCD's ARE NOT interlaced. At <288 vertical lines, they can't be, by definition. Doing any further "deinterlacing" or IVTC on such material is very detrimental to quality.

#3 - Using the same converter has NOTHING to do with it. If you recompress to a lossy format, you are ALWAYS subject to further degradation.

#4 - Looking further into EyeTV, I see you can set the original disc recording format to a number of quality/size/format options. Then you're exporting to VCD via the Toast plugin. This is the 2nd conversion (the "recording" being the 1st). It should be VCD legal--PAL=352x288, 25fps, MPEG1, 1150kbps video, 224kbps audio, Mpeg1Layer2, 44.1kHz, stereo, CBR. What about this isn't subsequently workable on your system?

#5 - This "sharper" older version could be that your recording format originally had a better setting, NOTHING to do with the "conversion" settings of ffmpegX.

#6 - I checked your "Media MVP" software, there shouldn't be ANYTHING wrong with using a direct VCD rip (do it the right way of course--VCDGear,etc). It specifically says that it supports PAL VCD (MPEG1 video, Mpeg1Layer2 audio, etc).

HTH,

Scott

[pdv1956]

I'm sorry if I offended you in any way.

There are a number of fallacies in your statement:

#1 - You don't say WHAT formats your "TV recording" software uses, but you do know already that it is NOT mp2/192/48 (I'm assuming that means Mpeg1Layer2, 192kbps, 48kHz--didn't say mono or stereo, etc). What you've already got is VCD disc. If it was done according to spec (I'll assume), the audio must be Mpeg1Layer2, 224kbps, 44.1kHz, stereo. Of course, you don't say WHY you might want it any other way. Editing? Use the right tools (cuttermaran, etc).

I do not have and I do not want VCD disk. I have many Terabytes of TV recordings, all stored on hard disk, for watching (the tast of wife children and me differs a lot, and we record more than we can view ).
I want disk files with low quality DVD, to be able to make a DVD out of it, without converting again to avoid quality loss, if I decide to keep them. VCD is OK for video but not for audio, so some conversion is needed.

#2 - You and another poster mentioned "interlacing". VCD's ARE NOT interlaced. At <288 vertical lines, they can't be, by definition. Doing any further "deinterlacing" or IVTC on such material is very detrimental to quality.

It was Major himself who asked about interlacing, so I answered. Maybe THIS is the setting that differs from my previous experiments, but for now I select Interlace: Not Interlaced.

#3 - Using the same converter has NOTHING to do with it. If you recompress to a lossy format, you are ALWAYS subject to further degradation.

Sure. But I NEVER watch live TV, always converted to the above spec. I used to use TMPGEnc on Windows for that, and it was fine, but I want to get rid of MS.

#4 - Looking further into EyeTV, I see you can set the original disc recording format to a number of quality/size/format options. Then you're exporting to VCD via the Toast plugin. This is the 2nd conversion (the "recording" being the 1st). It should be VCD legal--PAL=352x288, 25fps, MPEG1, 1150kbps video, 224kbps audio, Mpeg1Layer2, 44.1kHz, stereo, CBR. What about this isn't subsequently workable on your system?

EyeTV has all my needs. But I have other TV tuners too: ATI All in Wonder (windows) and Hauppauge PVR350 (linux).

I know 1150 is the standard, but why does FFmpegX always offer me 1138 ?

#5 - This "sharper" older version could be that your recording format originally had a better setting, NOTHING to do with the "conversion" settings of ffmpegX.

It seems I cannot state my problem clearly enough.

Recordings made by either of the above 3 tools I have always run thru TMPGEnc with good quality.
Then I found FFmpegX and it had the same or even better quality, plus my dual G5 2.7 is a lotter faster than my pcs.
But somehow I screwed up and now FFmpegX is not giving me the same clarity and sharpness as before, and I'm still trying to find where it went wrong.

#6 - I checked your "Media MVP" software, there shouldn't be ANYTHING wrong with using a direct VCD rip (do it the right way of course--VCDGear,etc). It specifically says that it supports PAL VCD (MPEG1 video, Mpeg1Layer2 audio, etc).

MediaMVP does support every MPEG you can think of, but the problem is that it needs a windows server. So I'm trying to het MythTV working on Linux. This far I can either record with PVR350 or playback with MVP but not both at the same time. So I'm still working on that. I want to keep Mac and Linux and get rid of MS totally.

HTH,

Scott

Thanks for all your trouble looking up things and answering me.

[Cornucopia]

You haven't offended me in any way.

I'm just trying to help you, and have to make sense of your situation. I have to say, it seems kind of convoluted...

Yes, I would prefer to have less M$ in my life, but it seems to be one of those crosses you have to bear.
Nonetheless, you still should be able to do what you want without resorting to Windows.

Troubleshooting:
If you have a process that goes like...

• A --> B --> C --> D --> E

and something is going wrong, you look at the whole chain to see if there's a better way, and you look at each piece in the chain. Bad links, bad interconnection between links, etc.

Looking at a whole chain, we see that you've got TV capture cards, and you want to end up with a LongPlay (i.e. lowerquality/bitrate/rez) DVD.
If you're capturing with EyeTV, you can either use the Toast plugin (and export directly to VCD, SVCD, or DVD--but you may not be given a choice of bitrate, etc), or you can export to QT (and any of those codecs) or you can export to one a number of presets. Note: You can probably "figure out" how those presets are made (XML/TEXT?) and create your own custom one and save it to EyeTV's preferences folder.
Basically, though, you just try to cut out the middleman. ffmpegX can do the job, but there's no reason to re-encode more times than absolutely necessary. It's quite likely that EyeTV can do the job directly.

So, let's say you want a 1/4D1 PAL DVD. That's--Video=352x288@25fps, ~1--4Mbps, MPEG1 or MPEG2, Audio=mp2, ~192kbps, 48kHz, stereo/joint (or 1/2 the bitrate if mono source), CBR. These can either be muxed into a MPEG2 Program stream, or left as unmuxed/elementary streams. Depends on what you're using to author/burn to DVD...DVDStudioPro would be best, prefers unmuxed, iDVD or Toast prob. requires muxed Program streams.

...or are you undecided on how to view--DVD settop or HTPC?? If you go htpc, your EyeTV output format should be determined by what the app (that you use to view) supports.

BTW, don't know as I'd say ffmpegX was better that TMPGEnc--faster, probably. It most likely is better than EyeTV at encoding/exporting--so maybe you want to export via QT uncompressed/losslessly/DV/etc and then encode with ffmpegX. Yeah, there'll be big interim files, but once you've got a good final encode, you can throw away the bigguns. AFA 1138, go figure. It's just one more thing to have to adjust...

Assuming you're not going to be using the ATI tuner (since it's Windows), What app do you use in Linux for that? That may determine (comon?) output options.


more food for thought. Anybody else...?

Scott

>>>>>>
edit: BTW, look at orig. post...13GB/Hour is exactly right for DV rates + quality (25Mbps).

Are these cards analog capture (where you set the rez/codec) or digital stream downloading? (I don't use TV cards very often).

[pdv1956]

You haven't offended me in any way.

I'm just trying to help you, and have to make sense of your situation. I have to say, it seems kind of convoluted...

Yes, I would prefer to have less M$ in my life, but it seems to be one of those crosses you have to bear.
Nonetheless, you still should be able to do what you want without resorting to Windows.

Troubleshooting:
If you have a process that goes like...

• A --> B --> C --> D --> E

and something is going wrong, you look at the whole chain to see if there's a better way, and you look at each piece in the chain. Bad links, bad interconnection between links, etc.

Looking at a whole chain, we see that you've got TV capture cards, and you want to end up with a LongPlay (i.e. lowerquality/bitrate/rez) DVD.
If you're capturing with EyeTV, you can either use the Toast plugin (and export directly to VCD, SVCD, or DVD--but you may not be given a choice of bitrate, etc), or you can export to QT (and any of those codecs) or you can export to one a number of presets. Note: You can probably "figure out" how those presets are made (XML/TEXT?) and create your own custom one and save it to EyeTV's preferences folder.
Basically, though, you just try to cut out the middleman. ffmpegX can do the job, but there's no reason to re-encode more times than absolutely necessary. It's quite likely that EyeTV can do the job directly.

So, let's say you want a 1/4D1 PAL DVD. That's--Video=352x288@25fps, ~1--4Mbps, MPEG1 or MPEG2, Audio=mp2, ~192kbps, 48kHz, stereo/joint (or 1/2 the bitrate if mono source), CBR. These can either be muxed into a MPEG2 Program stream, or left as unmuxed/elementary streams. Depends on what you're using to author/burn to DVD...DVDStudioPro would be best, prefers unmuxed, iDVD or Toast prob. requires muxed Program streams.

...or are you undecided on how to view--DVD settop or HTPC?? If you go htpc, your EyeTV output format should be determined by what the app (that you use to view) supports.

BTW, don't know as I'd say ffmpegX was better that TMPGEnc--faster, probably. It most likely is better than EyeTV at encoding/exporting--so maybe you want to export via QT uncompressed/losslessly/DV/etc and then encode with ffmpegX. Yeah, there'll be big interim files, but once you've got a good final encode, you can throw away the bigguns. AFA 1138, go figure. It's just one more thing to have to adjust...

I totally agree that all unnecessary steps should be avoided. I can tell EyTV to export MPEG1 352x288 1150 + MP2 192/48, so that needs no afterwork done. I can view it on Mac or pc, with MediaMVP on TV, and author without reconversion to DVD, so that is all fine.

Assuming you're not going to be using the ATI tuner (since it's Windows), What app do you use in Linux for that? That may determine (comon?) output options.

ATI on windows is the second source. here i can specify MPEG1 352x288, but sound is MP2/224/44.1 instead of MP2/192/48. so, here only audio converting is needed. ideally i would FFmpegX with video passthru and only audio conversion, but then i get the DV.avi not the MPEG. i thought that passthru was what it sounded like: copying the stream without any modification.

probably the nice way to do it is demux, convert audio, mux. but as a programmer i should be lazy they told me 25 years ago, so i just let ffmpegx handle both video and audio, and the result is terrible.

what i really would like to make in the future is a shell script launched on folder modification, that handles everything in batch.

PVR350 Linux is still not working so i start windows when i need a third synchronous recording.

more food for thought. Anybody else...?

Scott

>>>>>>
edit: BTW, look at orig. post...13GB/Hour is exactly right for DV rates + quality (25Mbps).

Are these cards analog capture (where you set the rez/codec) or digital stream downloading? (I don't use TV cards very often).

EyeTV200, ATI all-in-wonder, PVR350 all accept analog cable and have built in MPEG encoder/decoder. EyeTV is firewire, the others are PCI cards.