What's a good, free MPEG2 encoder?

[-Habanero-]

TMPGEnc seems to be shareware as I can only use the MPEG1 feature now after having that thing installed for years. Is there another one that can utilize multiple cores like TMPGEnc?

[turk690]

HCencoder:

http://hank315.nl/

[-Habanero-]

Thank you. Does it support CQ encoding as well or just 2-pass?

[El Heggunte]

^ Yes, it can do CQ encodes as well.
Warning: accepts only .AVS and .D2V files

[-Habanero-]

Wow, this piece of shit produced way worse quality than TMPGEnc did for MPEG1. Guess I'm stuck with MPEG1.

[davexnet]

Tmpgenc 2.5 was always known as a very good mpeg 1 encoder. But HCenc excels @ mpeg-2

[El Heggunte]

Apparently -Habanero- doesn't know how to configure HCenc properly.
Selecting a *good* (or 'adequate' at least) quantization matrix should be the first step.

[turk690]

Wow, this piece of shit produced way worse quality than TMPGEnc did for MPEG1. Guess I'm stuck with MPEG1.

Sorry I thought you were looking for an MPEG2 encoder. My bad. Then you may well be stuck, not really with MPEG1, but with the undead TMPGenc 2.5...

[-Habanero-]

Wow, this piece of shit produced way worse quality than TMPGEnc did for MPEG1. Guess I'm stuck with MPEG1.

Sorry I thought you were looking for an MPEG2 encoder. My bad. Then you may well be stuck, not really with MPEG1, but with the undead TMPGenc 2.5...

I was and you gave me an MPEG0 encoder instead. What's its excuse for doing worse than its predecessor?

[turk690]

I was and you gave me an MPEG0 encoder instead. What's its excuse for doing worse than its predecessor?

HCenc is about the best freeware MPEG2 encoder around where output quality is concerned, but it can have a steep learning curve. It threw me off at the beginning too, but a crash course in Avisynth and debugmode, for one, helped a lot. It has lots of options for all sorts of MPEG2 encoding tasks that can be confusing and may make you choose settings that are not optimum for the source material and the kind of output quality you like, but hey lots of help on the www, no? For me, about the only current practical use of HCenc is for creating *.m2v elementary streams for DVD, which I rarely do now. I also used HCenc to create streams for blu-ray in the past, but x264/265 has made that moot.

[racer-x]

Try ffmpeg. That's all I ever use for mpeg anymore, although I rarely use mpeg1 or mpeg2 anymore. It's very fast and quality is good enough for what I use it for.

[_Al_]

Wow, this piece of shit produced way worse quality than TMPGEnc did for MPEG1. Guess I'm stuck with MPEG1.

Sorry I thought you were looking for an MPEG2 encoder. My bad. Then you may well be stuck, not really with MPEG1, but with the undead TMPGenc 2.5...

I was and you gave me an MPEG0 encoder instead. What's its excuse for doing worse than its predecessor?

You especially need to write down what you are actually doing, so it is not like in x265 case (70kbps, 16 reference frames and elaborating over result), reveal your settings and not just cranking all sliders to the right , and all chokes.

[-Habanero-]

Apparently -Habanero- doesn't know how to configure HCenc properly.
Selecting a *good* (or 'adequate' at least) quantization matrix should be the first step.

I'm using the same matrix as the one on TMPGEnc, not sure what's so significantly different. Either way, I got MPEG2 to work on TMPGEnc again and it's still performing worse than MPEG1 so I give up. This is not what I remember back in 2009.

You especially need to write down what you are actually doing, so it is not like in x265 case (70kbps, 16 reference frames and elaborating over result), reveal your settings and not just cranking all sliders to the right , and all chokes.

I'm revisiting an old CG video I had back in 2002 which was widely shared on the net. It had a very low bitrate and as expected, bad quality. I was curious if it could've been decent if encoders back then used MPEG-2 rather than MPEG-1 and with slower settings.

About the 67 kb/s video, that's actually a good idea. It would be interesting to know what it takes for MPEG-1/2 to produce the same quality as the state-of-the-art at dialup bitrates.

[El Heggunte]

Apparently -Habanero- doesn't know how to configure HCenc properly.
Selecting a *good* (or 'adequate' at least) quantization matrix should be the first step.

I'm using the same matrix as the one on TMPGEnc, not sure what's so significantly different. Either way, I got MPEG2 to work on TMPGEnc again and it's still performing worse than MPEG1 so I give up. This is not what I remember back in 2009.

I said ''the first step'', NOT ''the only step''

You especially need to write down what you are actually doing, so it is not like in x265 case (70kbps, 16 reference frames and elaborating over result), reveal your settings and not just cranking all sliders to the right , and all chokes.

I'm revisiting an old CG video I had back in 2002 which was widely shared on the net. It had a very low bitrate and as expected, bad quality. I was curious if it could've been decent if encoders back then used MPEG-2 rather than MPEG-1 and with slower settings.

About the 67 kb/s video, that's actually a good idea. It would be interesting to know what it takes for MPEG-1/2 to produce the same quality as the state-of-the-art at dialup bitrates.

Did you know that MPEG-1 actually supports resolutions up to 4095x4095

Did you know that TMPGenc Plus is (apparently) the only MPEG-1 encoder that shines when creating non-VCD_stuff?

Did you know that MPEG-1 is extremely inefficient (bitrate-wise)?

Did you know that MPEG-2 sucks at low and very-low (average) bitrates

[johnmeyer]

Either way, I got MPEG2 to work on TMPGEnc again and it's still performing worse than MPEG1 so I give up. This is not what I remember back in 2009.

Then you are almost certainly doing something wrong.

I was going to tell you that there are ways to get TMPGEnc to re-enable MPEG-2 encoding, but it sounds like you figured that out yourself.

As for poor quality, I have several "professional" MPEG-2 encoders, and TMPGEnc can produce results that, for most video, is indistinguishable from what the $$$ encoders produce.

What bitrate are you using for encoding? As you may know, to get exceptional results, you want to use an average bitrate of 8,000,000 bps or more. You can still get very good results -- almost as good as using the max bitrate -- even down at 6,000,000. As you go below that, things start to fall apart. However, for home videos and other things that don't have much motion, I have been able to get reasonably decent results down to 4,000,000 bps.

You most definitely should NOT be fiddling around with oddball settings. This is especially true of creating your own matrix, something that is almost guaranteed to give you weird results. There are thousands of ill-advised posts in various forums with "magic" matrix settings for TMPGEnc. I've tried many of them and finally realized that these people didn't know what they were talking about.

So, if you use standard settings, TMPGEnc should give you results that look pretty much as good as the original material, at least when viewed in motion. Of course if you look carefully, frame-by-frame, you can see artifacts with pretty much any encoding scheme.

[El Heggunte]

.........

You most definitely should NOT be fiddling around with oddball settings. This is especially true of creating your own matrix, something that is almost guaranteed to give you weird results. There are thousands of ill-advised posts in various forums with "magic" matrix settings for TMPGEnc. I've tried many of them and finally realized that these people didn't know what they were talking about.

You know that the 'default' QM used by TMPGenc already is a "hack", don't you?
I mean, the 'standard' inter-frame compression matrix is an "all-16" square, whereas in TMPGenc, the 'default' starts with a '16' and ends with a '33'

As for the people who didn't know what they were talking about... I hope manono is not one of them

All Indian DVDs use either that MPEG Standard quantisation matrix or the Standard matrix. That's one of the (many) things that make them the worst DVDs in the world. Hollywood studios quit using those matrices years ago. I might question just how surprisingly good a movie with a bitrate of 1900 looks. Sounds like a boot to me.

http://forum.doom9.org/showthread.php?t=128618

http://forum.doom9.org/showthread.php?t=126446

http://forum.doom9.org/showthread.php?t=104679

Speaking in general: if you use a high average bitrate or low quantizers with a high-compression matrix, you'll be wasting bitrate (the encoder might pad the stream with zeroes, in order to give you some kind of placebo effect), and if you use a low average bitrate or high quantizers with a low-compression matrix, you'll get a festival of artifacts or something

[-Habanero-]

Did you know that MPEG-1 actually supports resolutions up to 4095x4095
Did you know that TMPGenc Plus is (apparently) the only MPEG-1 encoder that shines when creating non-VCD_stuff?
Did you know that MPEG-1 is extremely inefficient (bitrate-wise)?
Did you know that MPEG-2 sucks at low and very-low (average) bitrates

Shit man why you ask me so many questions!

1. Yes.
2. I suspected so, all the others were garbage.
3. I don't remember if MPEG-1 supports VBR, I know MPEG-2 does.
4. Nope. I thought MPEG-2 was better in all situations than its predecessor. Admittedly, this is probably my first time testing it at a low bitrate.

Then you are almost certainly doing something wrong.

I was going to tell you that there are ways to get TMPGEnc to re-enable MPEG-2 encoding, but it sounds like you figured that out yourself.

No, tell me. I wanna know if it's exactly what I did.

As for poor quality, I have several "professional" MPEG-2 encoders, and TMPGEnc can produce results that, for most video, is indistinguishable from what the $$$ encoders produce.

What I'm distinguishing is the fact that it looked worse than MPEG-1 from the same program with the same settings.

What bitrate are you using for encoding?

The same one as the original, 360 kb/s for 352x240. On the plus side, it still does look better than the original.

You most definitely should NOT be fiddling around with oddball settings. This is especially true of creating your own matrix,

I didn't, the default matrix is the same for both TMPGenc and HCenc.

[-Habanero-]

The color is also being changed, I don't remember this happening before either.

[johnmeyer]

Admittedly, this is probably my first time testing it at a low bitrate ... What I'm distinguishing is the fact that it looked worse than MPEG-1 from the same program with the same settings ... The same one as the original, 360 kb/s for 352x240.

OK, I didn't realize that you were trying to encode at really, really low bitrates. Different encoders do better at low bitrates, and MPEG-2 is not a good choice for low bitrate (dial-up, as you called it) encodes. That is why, fifteen years ago, we saw the emergence of several different codecs for low bitrate encoding, like Divx/Xvid.

While I'm still not sure what you are trying to do, and whether you are re-encoding something that is already 360 kb/s 352x240 or whether you have a high-quality original. Either way, 360 kb/s is going to produce some pretty lousy results, even at 352x240.

Also, a lot depends on whether interlacing or pulldown is involved anywhere in the workflow. When I first got started doing this, back when we were all encoding to put things on CD (VCD, SVCD, and eventually XVCD), I worked with a lot of music videos. Many of them were shot on film and telecined to broadcast at 29.97. When encoding at really low bitrates, if you didn't remove the pulldown, the TMPGEnc encoder totally broke down at low bitrates, even using MPEG-1 (VCD). It got even worse with MPEG-2 (SVCD & XVCD).

As for the people who didn't know what they were talking about... I hope manono is not one of them

Obviously he knows his stuff, but that quote really doesn't have much to do with what I said. I wasn't trying to compare one matrix with another and say which is better, but instead simply noting that I'd seen a lot of matrices that people had developed on their own (i.e., neither the "stock" TMPGEnc matrix or the matrices used in India), and many of those, which were touted as making a huge difference at low bitrates, provided no visible advantage and often made the encode look horrible.

[Tom Saurus]

You could try the Video To Video Converter.

[ezcapper]

Maybe try Handbrake? It's easy to use and can utilize up to 9 cores. But might not have all the features you want. Encode at a very high bitrate. I got good results on a very low quality video when I set the qp to 1 or 2 and changed the container to mkv. I left everything else the same.

[-Habanero-]

OK, I didn't realize that you were trying to encode at really, really low bitrates. Different encoders do better at low bitrates, and MPEG-2 is not a good choice for low bitrate (dial-up, as you called it) encodes.

I'm doing both. And MPEG 1/2 can't go anywhere near dialup bitrates for 480p.

While I'm still not sure what you are trying to do,

Doing quality tests, hence it irked me that MPEG-1 did better than its successor.

and whether you are re-encoding something that is already 360 kb/s 352x240 or whether you have a high-quality original. Either way, 360 kb/s is going to produce some pretty lousy results, even at 352x240.

No, I have a higher quality source. The version that was being widely shared in 2002 was garbage. I'm trying to see the very best that could've been done with it back then. But it doesn't matter now because I've done the comparisons and I'm satisfied with the data. The people in ~2002 obviously used settings that would get the job done sometime this century.

Also, a lot depends on whether interlacing or pulldown is involved anywhere in the workflow.

Nope, I unchecked those. The video is progressive.

[pandy]

For MPEG-1 or new ffmpeg (search for magic mambo jumbo with lambda settings i.e. new way to control ffmpeg quality) or i use personally very old quenc - it is way worse than HCEnc in MPEG-2 but MPEG-1 is from my perspective OK.

Beware that MPEG-1 was designed to provide 352x240@30fps with 1152kbps.
You may consider x262 also as interesting MPEG-2 (1?) experience.

--
hint
in case of quenc check first comment...

[El Heggunte]

..............
Beware that MPEG-1 was designed to provide 352x240@30fps with 1152kbps.

IMNSHO, that was just a lame excuse from QuEnc's author for not doing his homework.
BTW... sometimes 1152kbps is not sufficient for VCD-video, in the same way 9.8Mbps sometimes is not sufficient for DVD-video...

[-Habanero-]

For MPEG-1 or new ffmpeg (search for magic mambo jumbo with lambda settings i.e. new way to control ffmpeg quality) or i use personally very old quenc - it is way worse than HCEnc in MPEG-2 but MPEG-1 is from my perspective OK.

Beware that MPEG-1 was designed to provide 352x240@30fps with 1152kbps.
You may consider x262 also as interesting MPEG-2 (1?) experience.

--
hint
in case of quenc check first comment...

I will try to use ffmpeg to do the job and see if it provides better quality than TMPGenc. I disagree with you about what MPEG-1 was designed for, tho. Its application is diverse. From the HD 4K resolution and 100 Mb/s to the 320x240@64 kb/s designed for videoconferencing over an ISDN connection.

I do recall from previous discussions that some people preferred MPEG-2 at high bitrates than x264 because they said the quality is higher. But just now I've discovered otherwise. At the very maximum (highest CQ), both MPEG 1 and 2 failed to deliver perfect quality while x264 did exactly that. Seriously, don't use mpeg-2 anymore, folks. It's 2016.

[El Heggunte]

^ Yes$S, the year is 2016, and x264 is 12 years old already.

So let's forget x264 as well, because HEVC is the future, and the future is now

https://www.google.com/search?tbm=isch&hl=en&source=hp&biw=&bih=&q=snk+%22the+future+i...h+Images&gbv=1

[-Habanero-]

So let's forget x264 as well, because HEVC is the future, and the future is now

Good idea.

[usually_quiet]

AV1 is the future.

I expect to mostly live in the past for quite a while yet.

[_Al_]

Yes, just act rationally
testing 20 year old codec using bitrate nobody even remotely thought it was going to be used for or latest codec, that is not developed quite yet. H.264 is in ripe stage right now, x264 is good with low birates (no pumping etc., not sure about extremely low bitrates like 70kbps). In future there can be a great flood or whatever ,I'd stay on the ground and use what has been proven.

Certainly I'd would not want to x265 developers stop right now and evaluate Habanero's posts about worse results for 70kbps and SD resolution. They increased speed, everyone was crying about encoding speed. So there it is. Now just wait what or if they can do more with it, but not a 70kbps and 720x480, with bitrate starving to the point of death almost. That is certainly last stage in development, bitrate starving tune up, where 10 year of development shows up (understand ability to camouflage, to deceive, what one can see, what can be taken away and yet so it does not look relatively that bad) .

[-Habanero-]

_Al_, I did new tests, see the thread. x265 did a lot better than x264 at higher bitrates and I'm not sure about the speed, 2.0 seems slower than 1.2 and slightly worse quality.