What is the BEST Mpeg 2 Encoder Today?

[chowmein]

LOL who's been taking their grumpy pills today

[deadrats]

And yes suggesting a forum where people DEVELOP encoders isn't ridiculous.

i just have to say this, claiming that the doom9 forums is superior to these is stupid beyond belief. that forum, and the copycat doom10 forum, is dominated by the 2 x264 developers and their blind followers, who take anything those two clowns say as gospel. hell, the guys that run the respected MSU comparison tests have been attacked and abused over their, because their tests didn't "prove" that x264 was the clear choice under each and every circumstance.

to make matters worse an intel engineer, the guy that was responsible for a couple of SSE4 instructions, showed up to offer his help in getting a modified x264 to work with intel's Quick Sync and he was mocked, ridiculed and ultimately chased away; same thing has happened to anyone that showed up to help the x264 developers get their encoder to run on a gpu.

those guys, for the most part, are complete assh*les, pure and simple.

now are their some professional caliber forums where people more knowledgeable here may hang out? sure. but doom9/10?

f*ck most of them.

[souper]

I don't think were talking about x264 here. I was talking more specifically hcenc and quenc. I'm not saying doom9 is superior, I'm just saying there are smarter people there which is true.

[deadrats]

"smarter" is a very generous description to use; you get the hcenc developer that posts there but he also posts here as well. quenc, while i may have liked it, hasn't been updated since 2007, quite old by software standards.

about the only "advantage" doom9 offers over this forum is that there is some guy over there that's trying to use the x264 code base to create an mpeg-2 encoder, which is an interesting project and some other guy that has slapped together an h264 encoder written in pascal. i think the avidemux devs may also hang out there, but as i said almost all those guys come off like such douches, if you don't worship at the alter of x264 and one or two other apps, you're automatically the bad guy.

the atmosphere is like a x264 circle jerk and the best part is that dark shakari was once banned from that board for disagreeing with the site admin who is also the dev of a cuda powered decoder and banned temporarily, so he set up doom10 and now does the exact same thing if you have the balls to in any way criticize his software.

smarter? i would take the advice that the guys in this forum give me over anything the guys over at doom9 say any day of the week.

but hey, it's a free country, if you feel that they are smarter, so be it.

[poisondeathray]

hell, the guys that run the respected MSU comparison tests have been attacked and abused over their, because their tests didn't "prove" that x264 was the clear choice under each and every circumstance.

Did you bother to even look at why ? or what was being said? They raised very valid objections.

MSU was blasted 2 years ago, because they didn't use the encoder properly or the correct settings. Before you say it's a "cop out", this is all well documented. MSU initially stated were testing PSNR, but later changed to SSIM, so they used the wrong encoding parameters.

So what happens when you use the correct settings? Have you looked at the most recent MSU comparison?

It wasn't that long ago when a certain Mr. ----rats claimed "DivX was better than x264:, so and so, and so forth, x264 sucks... blah blah blah... IIRC , Mr.----rats blasted some certain Mr. -----ray for saying "you probably didn't use the encoder correctly." And so some tests were conducted. Transparent testing methodology with freely available sources, repeatable by anyone. Nothing to hide. And indeed Mr. ----rats x264 encodes looked like crap.... But what happened when x264 was used properly ? Now do I have bring up that dirty thread again?

So how is this relevant to the topic at hand ?

It's relevant because when the testing methodology isn't sound eg. has problems, bias, confounding factors etc... the observations and conclusions that you draw from those observations are not necessarily valid. That's not the same thing as "useless." Every test - even poorly conducted ones - provide at least some information.

I mentioned this already, but most the comparisons done are garbage for the various reasons mentioned. These are valid critcisms. Examining the testing methodolody is a good thing. It's the heart of scientific method. If there is a valid criticism you need to re-evaluate your methods or improve your tests. Objective testing, eliminate confounding variables, and transparent, reproducable testing is what it is all about.

to make matters worse an intel engineer, the guy that was responsible for a couple of SSE4 instructions, showed up to offer his help in getting a modified x264 to work with intel's Quick Sync and he was mocked, ridiculed and ultimately chased away; same thing has happened to anyone that showed up to help the x264 developers get their encoder to run on a gpu.

Now that was pure stupid . I was around for that discussion. I've seen Francois on several other hardware forums years before that fiasco, and he is the real deal from Intel.

But that 2nd part is not entirely true. People generally aren't "mocked" (that 1 case with Francois was an exception, I think there was some sort of psych complex kicking in). It's happened a dozen times or so over the past few years - and the same thing happens. Rinse & repeat. Right on cue. These "GPU" developers are actually encouraged to submit work and patches. They begin all enthusiatic and "gung ho" about it until they realize how difficult it is, never to be heard from again after a few days. <crickets> . The next guy is due pretty soon...

[poisondeathray]

I have had very good results with TMPGEnc Plus 2.5 despite its age. Then again when I do home movie transfers I max out the bitrate and only put an hour of video on a DVD tops. Even noisy sources like video shot with a BetaMovie tube camera come out without artifacts. Then again, it does help to clean your source video as much as possible before encoding.

Maybe I'm reading what you're saying incorrectly, but isn't "coming out without artifacts" in that scenario a bad thing?

IMO, the job of an encoder is to reproduce the signal given to it as close as possible. If you give it a noisy signal, then a "perfect" encoder would reproduce that noise exactly. In fact , that's what a lossless encoder would do, like lagarith or huffyuv

[lordsmurf]

@deadrats: Doom9 is great for Avisynth. Many years ago, it also had some great discussions of capture theory (even if I did disagree with a lot of it; for example, too much emphasis on Kell factor). The people having the discussions were often too interested in the subject matter to waste time on attitudes. The only thing that really irked me at doom9 was the saturation of users with obviously cracked copies of software like Scenarist, Maestro, CCE, etc., as it was most of them who had the bad attitudes.

@poison: A "perfect" encoder depends on the situation. I disagree that an ideal encoder reproduces noise. MPEG-2 DVD-Video, for example, has limitations in bitrates that would allow for it, so proper intelligent choices need to be made by the encoder to sacrifice noise without losing detail. H.264 inherently has graceful degrading in it (i.e., softness instead of blocks), for that exact reason. An encoder should only work to reproduce all detail if that's what the situation demands. Panasonic MPEG-2 encoding is a great example of what not to do: attempts to save "all information" lead to blocky near-unviewable crap.

@souper: I don't recall being anybody's employee here. My rate is $100/hour for research, prepaid. And I also have the right to turn down projects. So enough of this demanding I do anything for you BS. Your tax dollars pay for librarians, go use one. If your library sucks, move to a bigger city. Start with Broadcast Engineering magazine, and then branch outwards from there.

This appears to be getting off-topic now.

[poisondeathray]

A "perfect" encoder depends on the situation. I disagree that an ideal encoder reproduces noise. MPEG-2 DVD-Video, for example, has limitations in bitrates that would allow for it, so proper intelligent choices need to be made by the encoder to sacrifice noise without losing detail.

That's a good point, but you have more "intelligent choices" and far more control preparing the encode (e.g. with filters) than "rolling the dice" by sending a noisy image to the encoder and hoping it makes a "proper intelligent choice"

Perhaps that's what NJRoadFan meant , and I interpreted what he wrote incorrectly

One man's "detail", is another man's "noise" . This thread is about VHS, but is film grain "noise" or "detail" ? or both ?

[lordsmurf]

it is film grain "noise" or "detail" ? or both ?

Nikon vs Canon
Kirk vs Picard
Mac vs Windows

There's never a right answer to that one.
I'm very anti-noise, but at the same time I can appreciate grain.

[NJRoadfan]

Maybe I'm reading what you're saying incorrectly, but isn't "coming out without artifacts" in that scenario a bad thing?

I'm referring to artifacts created by compression, specifically macroblocking during high motion scenes. Given the lousy quality of the originals (ie: more challenging for the encoder to compress), I was impressed with the results. I have used a few really lousy and expensive encoders in the past (LSX-MPEG, blah) that struggled with even clean studio shot VHS sources.

I am not overly aggressive with filtering, but I do have my lossless originals so I can experiment with cleanup or re-compress with another (better?) encoder in the future.

[souper]

@lordsmurf ohh geez no proof again not surprising, Maybe I'll go look at my library, I don't think I'm going to find anything. Ya, everyone at doom9 is some scum pirate in alarming numbers. Stick to your ******* vhs restoring bullshit, I wouldn't pay you ******* squat. I'm gonna say it as nice as I can you are a idiot, you lie to make it seem like your right (appearant publications twice a year you see, but can't provide any proof of their existence). STOP GIVING PEOPLE BAD ADVICE

hxxp://broadcastengineering.com/ <-- here's the official website, show me the comparisons. And I don't think I'm gonna see HCenc in there either do you? Answer my questions properly, don't beat around the bush <-- this part is important. Keep digging you might get to the bottom of that hole you dug.

No more x264 talk (even though it is the best), back onto mpeg2 discussion. Best advice is to try them all, use what you think looks the best to you, people can only give their opinions, but when there's a bunch of opinions that lead to same conclusion, its usually the right one. COUGH CCE HCENC COUGH. Now ask yourself how many people use mainconcept, or even recommend that pos?

[lordsmurf]

Most BE content isn't online. Just FYI. There's a great big world out there full of offline information. That's the difference between your average keyboard warrior and the rest of us. When we walk away from the computer, we still know what we're doing. That includes methods to do research and find information. I was doing it long before the web existed, too.

I also didn't say BE was the only pub with testing.

To think nobody is conducting tests is both naive and foolish.

I'm just not going to feed the troll anymore. /done

[souper]

ok thats great, beating around the bush again thats a good one, produce the god damn "offline" publications? Probably not going to are ya? What are the other ones published? keep on digging your almost at the bottom. It's simple just ANSWER the QUESTION, understand?

[johns0]

Time to closed the thread.