My Encoder Comparison Results

[reboot]

Because MC is so fast, I usually do single pass, and I'm quite satisfied. As you mention, there is the odd file where tmpgenc would seem to do a better job. In these cases, I just do a 2 pass in MC, and it solves the jitters nicely.

[trock]

Funny thing was I tried 2 pass in MC and it didn't solve it whereas both 1 pass and 2 pass in TMPGEnc were fine. Mostly MC does a good job though, as you say.

[tigerman8u]

@dipstick
Thanx for the comparison test. I use Mainconcept almost exclusively. Don't let one person's arrogance/ignorance or whatever discourage you from contributing more constructive posts etc. to this site. Keep up the good work.

[proxyx99]

....I like the Procoder output best (brighter color and higher contrast it seems).
/Mats

... until you see a good material ruined by these brighter colors and higher contrast. This is not what encoder is supposed to do.

i just finished a film conversion project for the producer of a movie that was shown at the sundance film festival ... for mass duplication ..

i did color conversion for film to video and authored both a pal and ntsc master onto DLT , they also had extras shot on video and those were included .. The source was from a scan.

the project was encoded using procoder -- and the producer, director and cinematographer all checked the proof disks and found the results "stunning" ..

procoder will not ruin the source - if you know what you are doing is all i can say ..

BJ_M, if you look at the frame above there's a lot of clipping, so were the images in the other comparative test (glass dome, remember...?).
You try to sound like using Procoder is a Voodoo science but the truth about it much more trivial. As to the stunning results I'd have to see it... but they are possible with any good encoder if the material is prepared properly). Quoting unnamed sources is ludicrous (like saying my neighbor said it was good, is this an argument to be taken seriously? ). Final product especially for big screen NEVER resembles the original so it is up to the individual to judge the outcome.

We're talking encoding with default settings, out of the box, and Procoder output is modifying the source in an unacceptable way. Can you adjust the settings for individual taste? Of course, but that is what color and exposure correction is for... and that goes beyond encoding fidelity. CCE, TMPEG, MC stay true to the source how come Procoder is not? Is everyone that uses it an idiot except for just 1, you? I did a lot in Procoder and I know how to use it to get good results but that requires tweaking. Don't talk about footage nobody has seen but speak to the posted results.

Long ago a contour feature was implemented in consumer audio to make listening more pleasing for an average Joe. You won't find it in pro grade amps. Same here, color and contrast boost is implemented to fool the eye of a viewere into thinking that it is better. It is not... and if you had no access to the source clipping illustrated here is unrecoverable, permanently damaging the source image. You don't want to say the the outcome presented here is good or "stunning"? If it is as "stunning" to you as the work you mentiond then it is a time to rethink your profession (not personal, but if you support you argument with your "pro expertise" then yuo have inadvertantly put it here on the line). We are talking basics here...

[BJ_M]

proxyx99 , I do work with film and compression technologies for large screen , as well as for dvd ... it is what I do.. i'm sorry , but thats a fact ..



now lets look at some FACTS ..

Here is a smpte chart ..

the first image on the top is the source display on the video scopes and the image on the bottom is the encoded result .. the encoder is procoder 2, completely standard preset for dvd, with no filters of any kind added or any settings changed.

please show me any changes that you see - all levels are exactly the same ..





please show me any changes that you see - all levels are exactly the same ..

[BJ_M]

here are the source and target images -- you can duplicate these results ...


top is the source smpte chart

bottom is a screen shot of the mpeg file clip after being encoded in procoder --

[BJ_M]

load both images into photoshop and check the RGB values even ..

[BJ_M]

now we come to the issue why the first pic has an problem --

it is a simple mater and anyone who works with video on more than a hobby level should recognize it ..

two issues -- out of bounds color and (way) out of legal values video ...

see pic below - i JUST checked the area marked huffyuv and cropped the rest out - i also made sure the letters were under 235 - so they are not a factor ..


we have a high value of 260 and a low value of -10 .. clearly this is wrong ....

the color values are also out of bounds ..


this shows why proper source files are one of the major keys to good encoding ...

(it also shows I am right)

[proxyx99]

OK, just spent few min. (oh, how I hated that...) putting it all together and I see in the meantime you have been busy as well...

What you indicate in short is that Procoder has made all values legal, am I right? So it did modify the source exposure values and look, right? Why? Because it didn't like them. There is an algorythm in PC that makes the pictures fit its presets, that is why what happens happens. I was under the impression that you were saying that it does not modyfy anything (test board above). You have just proven yourself wrong.

Below is my messy compilation of few captures.
Left PC column, middle is source right is reference image (tested). There is one more pic of how MC handles the source (on the dot!).
Images were borrowed from other thread:
https://www.videohelp.com/forum/viewtopic.php?t=255200&highlight=





PS. so now you say the source is at fault? Procoder good, sources suck badly. Now, keep in mind this is videohelp.com and we are not working on the next blockbuster big screen release (they don't use procoder anyway). All other tested images/encoders (CCE, MC, TMPEG, QUENC) fall into sources histogram range (like MC shown here) but not Procoder. So how would you explain that. They can but Procoder has its special way of dealing with poor stuff : FIX IT its way!

[proxyx99]

BJ_M pls fix you images layout, makes page unbearable. Hope that was accidental, or was it?

I wonder what would it take for you to step aside look at the material at hand and admit there is something wrong with how PC handles video.
I don't feel like spending more of my time on gathering info that would be visible even for a man with advanced glaucoma.

[BJ_M]

OK, just spent few min. (oh, how I hated that...) putting it all together and I see in the meantime you have been busy as well...

What you indicate in short is that Procoder has made all values legal, am I right? So it did modify the source exposure values and look, right? Why? Because it didn't like them. There is an algorythm in PC that makes the pictures fit its presets, that is why what happens happens. I was under the impression that you were saying that it does not modyfy anything (test board above). You have just proven yourself wrong.

Below is my messy compilation of few captures.
Left PC column, middle is source right is reference image (tested). There is one more pic of how MC handles the source (on the dot!).
Images were borrowed from other thread:


PS. so now you say the source is at fault? Procoder good, sources suck badly. Now, keep in mind this is videohelp.com and we are not working on the next blockbuster big screen release (they don't use procoder anyway). All other tested images/encoders (CCE, MC, TMPEG, QUENC) fall into sources histogram range (like MC shown here) but not Procoder. So how would you explain that. They can but Procoder has its special way of dealing with poor stuff : FIX IT its way!

what i said is that procoder does not alter a image -- which is contrary to your statement that it always darkens or adds more contrast etc ...

the scopes do not lie ..

and by the way --- yes , several commercial releases (i did some of them in fact) were done with procoder ..

what you fail to understand is the other encoders clipped the images - or they just cored it .. in procoder - you have a wide range of selections to handle such cases and it handles it properly .. otherwise PC will just clip at 0-255 (which is what it should do)

furthermore - i took the "source image" from the first post and encoded it with no settings changes (no filters) and it looks (the target file) the same as the source - i then added filter for 601 and it was back in spec ...

i could not duplicate the image shown in the top post .... I would contact canopus and make sure you are running the latest release and the service pack .. i do know that "warez" versions of procoder so not work properly either (im not saying this is the case - but it is a known fact) ..

its not my fault you dont know how to use the filters properly .. procoder comes with two great books - one of which is the manual and the other is "video compression" , i suggest you read them both ..

[BJ_M]

BJ_M pls fix you images layout, makes page unbearable. Hope that was accidental, or was it?

I wonder what would it take for you to step aside look at the material at hand and admit there is something wrong with how PC handles video.
I don't feel like spending more of my time on gathering info that would be visible even for a man with advanced glaucoma.

the images look fine on my systems -- perhaps you need a larger monitor ?

[monzie]

Look, I KNOW a lot of the REGULARS here think I'm an arrogant (spell?) bast@rd, but ALL i'm trying to say is use an encoder (or encoders) that your happy with...find its (the encoder/s) good points/ bad points and work within its parameters......how many people here REALLY nead to go to BJM's lengths when all they (well most people) need to do is see MOVING PICTURES on the TV that look good via a DVD disc (plus good audio if available)?

Most NORMAL (average income) people do NOT employ bricky's, cleaners, electricians, plumbers, window cleaners, gardeners, etc but DO IT THEMSELVES..because its cheap and GOOD ENOUGH for their requirements..end of story.

Moral: Use a tool your happy with, but DONT piss (dis? for NA readers?) on others for using OTHER tools....a hammers a hammer REGARDLESS of the blow it strikes.

[BJ_M]

sorry - didnt mean to piss on the other encoders - in fact i like all of them for different things and i still think tmpgenc is one of the best bargins in an encoder for a long time ..



and true -- whichever one works for you ...


its like arguing over which is best , ford or gm

[monzie]

Which was my arguement all along (though I agree from re-reading my posts that IT can well be misenterpreted as arrogant twaddle)....hell i'm an amatuer not a pro (unlike yourself)....I just find encoder 'comparisons' rather pointless....or car v car, truck v truck, plane v plane, ship v ship etc (they ALL do their main job..but how they get there DOES varey).

[proxyx99]

OK, that's what I thought. We have 1 Procoder (yours) that works properly and all the rest that doesn't. I did not use any of my encodings so your remarks are improperly addressed .
By reusing other's stuff I just illustrated (for everyone to see) that those who use it with default settings and no filters should expect the results as posted. Histograms do not lie and no matter what mumbo-jumbo you try to sell here it is not going to fly. I think that you underestimate the intelligence of some if not all the members here. BS is BS and I just try to help other see it for what it really is.
If you call black white and the opposite regardless of facts it's your own prerogative but don't mislead the public on this forum. It is well known that procoder makes images more punchy and I just provided (one of many) proof of that. You can say whatever you want but I just posted histograms from someone else's work (in case you didn't notice ) and these do tell what happens. For the sake of objectivity I have not and I will not post my encodings knowing full well what your answers will be (not latest version, not latest patch, wrong settings, improper source, illegal colors, illegal version and God knows what else... let's not forget spitting over left shoulder or lack thereof could have affected the results).

[Edmund Blackadder]

OK, before most of you trash ProCoder, here's the thing. The color crushing happens most of the time when you import DV AVI's into After Effects and then save them into some lossless codec or uncompressed. Depending on your DV codec it may happen and it may not. From my experience Canopus DV codec converted to Huffyuv retains proper color values, while MainConcept DV codec converted to Huffyuv does not. After Effects doesn't seem to have a solution for this, neither Premiere (in any form of dedicated filter).

So said all this, if you do have such clipping in ProCoder this is all you have to do (screenshot of ProCoder 1.5, but 2.0 has the same filter):



ProCoder is a great encoder, so please learn using it before coming to trashing conclusions. Try encoding your footage again with this filter and this will solve everything. Case closed.

[BJ_M]

OK, that's what I thought. We have 1 Procoder (yours) that works properly and all the rest that doesn't. I did not use any of my encodings so your remarks are improperly addressed .
By reusing other's stuff I just illustrated (for everyone to see) that those who use it with default settings and no filters should expect the results as posted. Histograms do not lie and no matter what mumbo-jumbo you try to sell here it is not going to fly. I think that you underestimate the intelligence of some if not all the members here. BS is BS and I just try to help other see it for what it really is.
If you call black white and the opposite regardless of facts it's your own prerogative but don't mislead the public on this forum. It is well known that procoder makes images more punchy and I just provided (one of many) proof of that. You can say whatever you want but I just posted histograms from someone else's work (in case you didn't notice ) and these do tell what happens. For the sake of objectivity I have not and I will not post my encodings knowing full well what your answers will be (not latest version, not latest patch, wrong settings, improper source, illegal colors, illegal version and God knows what else... let's not forget spitting over left shoulder or lack thereof could have affected the results).

i'm sorry - you just don't want to see the facts -- look at the smpte chart encoding, no change ... end of story ....

the chart and the scopes are industry accepted - repeatable methods of analysis ..

[dipstick]

Sorry guy's, but I'm with proxyx99 on this one. It doesn't matter what source I send Procoder (even Canopus DV, or untouched dv for that matter) gets changed after encoding with standard DVD templates.
--------------------------------------------------------------------

I do have a question for all you Procoder Gurus. How can you manually set the Block scanning order in Procoder? For instance, I want the finnished m2v to have either Alternate scan order or Zig Zag scan order that I choose. It's easy to do with CCE or MainConcept, but I can't seem to find any setting that will do it in Procoder. I can set the DCT to Field or Frame, but it doesn't have any effect on the Scan order used. Even if I feed it interlaced dv and choose Field based picture encoding, it will usually encode it with Zig Zag scanning order. Sometimes it will encode it with Alternate scanning order. It basically does what it wants with no regard to the source. I scan all my encoded videos in Bitrate Viewer for verification.

I'm sure one of you smart guys have figured this out.

Thanks.

[proxyx99]

So how would you explain clipping experienced by other members?
Why is the before and after Procoder histogram different?
Why are both testers facing the same challenge and have obtained the same results (clipping)?

BJ_M this IS problematic and I would welcome a constuctive discussion rather then "you all don't know what you're doing" attitude.

Obviously histograms are not fabricated (you can derive your own) nor are the pictures so intelligent mind has to post a question why?

If you have something to say that would explain this phenomena say it.
But don't tell that it is an illusion. We all can see it and the tester has applied all the same default settings in all scenarios. In both quoted encoder tests Procoder outcome is unlike all others. Why?

In my opinion it is adjusting the image color spectrum to fit its presets. That is clearly visible.

Let's not talk about you and me but why other guys are getting consistently the same results which are different the other encoders?
Or is it fraud on their part?

Would you say that their results are here to mislead? If so say it.
But you can't pretend that there is no problem here.

PS. encoding smpte chart is not the same as encoding random matererial that consists of variety of differently lit scenes.

Anyone with Procoder may post be4 and after pictures of whatever material and let's see. How about that? BJ_M you don't need to do that. We know you are partial to Procoder, your objectivity and fairness is severly compromised.

I will always admire PPL like Dipstick who seek real answers rather then allow beeing fed bullshit.

[BJ_M]

send me a smal clip where this shows up (at least 120 frames) and i will be glad to take a look at it and see why there is

A) a problem

b) what can be done to fix it



is this fair enough ?

[proxyx99]

let's talk about what is at hand rather then enter a new area with this or another clip. Tests were done and why not have a discussion what has happened and why?

One of your first reactions was
"Procoder doesnt blow out whites -- so either an error was made in conversion in AE or you didnt select the correct levels in PC".

Results from 2 guys are in. Be constructive and abjective (I hope you still can).

PS. I would really appreciate you repositiong the images above. I do have 19 inch and see no reason to buy another one or 21 just to see this thread. I just don't like microscopic resolution. Hope you understand.
Btw. I'm once on my laptop and the other time on a desktop and the way you posted your stuff is a real pain on both.

[BJ_M]

reposition what image ?

and results from who ? i am the only who posted results from industry accepted tests .. the rest are meaningless

[proxyx99]

That is exactly what I expected. You have nothing to say. Facts and pictures speak volumes, thank God. Don't know what is bigger: your ignorance or your ego?
Don't pose as an expert, cause you're not. Experts address issues in a concrete and respectful way unlike you just did.

[vhelp]

Maybe I can throw in a couple of cents worth here.

I think both sides have good points. But, I do believe that the test
was done accurate on the demonstrated pics above. However, I also
beleive that from another perfective, the test was invalid.., though
they weren't.

Let me try and explain.

BJ_M was correct. But, from an already processed image point-of-view.

However, instead of arguing about the tests that he did, which again,
were accurate (IMO) ..forget about what was said, and at least consider
giving his suggestion a try.

Post a clip of the 'acutal' source, BEFORE you send it off to Procoder

I believe, that the problem lies *BEFORE* procoder gets to it, and processes
it. Heck. Just U/L two frames worth in AVI form. Don't save it to another
format. for instance, if your finished AVI is ready for Procoder, stop there,
and follow these directions, (please) ...

* open avi inside vdub
* highlight just a couple of frame in the time-line (use HOME and END to mark)
* then, proceed to select Video/Direct Stream Copy/..
* then, proceed to select File/Save AVI..
* and save those two (ore more) frames

And let us (BJ_M) and others process them in Procoder (and TMPG) with
there prefered method to see if they match the color space. If all works
right, both frames (original vs. encoded) should match in a fluid manor.

-vhelp 3182

[proxyx99]

I'm sorry but BJ_M's objectivity and expertise connot be trusted anymore. I called for others to post an original test file (Dipstick, can you help?). Let others do the test and post before and after together with histograms. BJ_M has brought nothing here so you may sit on a sideline... and wait for the results to come.

What I would like to see is how consistent the encoding is across different users (not BJ_M who's results we all know by heart, always excellent).

You say the problem is be4 Procoder? ... but not be4 CCE and MC and other encoders. Why is the "Problem" only showing on Procoder?
How come other encoders output is just like the source (with the "problem" embedded) but it extends like a stick in a pile of shit (in Procoder).

Ican't speak for everyone but I'm personally not interested in BJ_M's test results (no offence). They are meaningless to me (I never encode test boards, it's kind of boring to watch them later on, don't you think?).

And stop (to all) this BS about illegal colors, warez and patches and other irrelevant stuff. It either works or it doesn't.

[adam]

For those of you who have levels/luminence problems with Procoder, are you really trying all sources or are you maybe talking about DV sources only? The Canopus DV codec is known to keep the source at 16-235 after converting to RGB, which is something that almost no other DV codecs do. If you don't account for this then you could easily get these suspected "changes." Its the codec that is doing this, albeit Canopus's codec, not the encoder. More specifically, it is user error.

I use Procoder (1.5) for all kinds of different sources and I can confirm BJ_M's results. There is nothing wrong with Procoder in this regard. Like ANY other mpeg encoder, you have to take into consideration colorspace conversions.

BJ_M posted histogram and scope captures. Exactly how much more objective can he be? I do have to fault him for one thing though...this post's tables are broken now. I've got to scroll right and left to read a post...very annoying.

[Edmund Blackadder]

Did anyone even read my previous post about fixing the color space problem in ProCoder? Why are you people still arguing? Just apply the stupid 601 Correction (shrink) filter and your troubles will go away. That's one simple answer to this topic.

Or is it that some people simply want to hear what they want to hear (i.e.: ProCoder is crap)?

[proxyx99]

Did anyone even read my previous post about fixing the color space problem in ProCoder? Why are you people still arguing? Just apply the stupid 601 Correction (shrink) filter and your troubles will go away. That's one simple answer to this topic.

Or is it that some people simply want to hear what they want to hear (i.e.: ProCoder is crap)?

I think you got it all wrong. The purpose of the thread is to find a common platform for all encoders, determine for the sake of all forum users what is their typical performance and use this knowledge for future projects. No one has ever said that Procoder is crap. The intention is to explore up- and downsides of each and every one and find ways to address common issues.

If the topic of a specific encoder's performace vs. another one sufaces from time to time it is an indication that not all the questions are answered and all the issues cleared. Everyone has a right to share their experiences with others in search for explanation. It also shows that this subject is still controversial.

I don't believe that we can bring some definitive answers. This is not a test lab and no one expects perfect, undisputable, scientific tests.
Consumers Report is not testing a device to establish industry or scientific standards but rather to assist a consumer in making a purchase decision and providing a good background information on its use. This in turn does raise the bar and helps users get the most of a specific product.

It doesn't help anyone to flip the problem, make it sound cryptic, ignore or scold. Good advice and information is always in short supply.

I agree with Adam that problems with Procoder do originate with Canopus codec implementation (I used to own some Canopus stuff be4). The issue is to develop a way where whatever material you take you will get a predictable result. The fact that tested clips behave almost identical in all other encoders (but Procoder) forces a question why and how to make Procoder output original-like product. Default settings don't cut it.
And don't tell anyone that there is no problem. It is clearly visible and confirmed by a histogram.

Again, this is not to bash but rather find a way to address the issue. Keep this in mind.




Source is on the left, Procoder on the right. Notice loss of detail in the middle (white clipping).
I wonder if Procoder Express does that as well (consumer version of the product) and whether Canopus has educated users in the manual in both Procoder and the Express ver. how to avaid that. This is a major defect.
You may argue that experts should know (... no they shouldn't in my opinion) but how about average Joe, has it been fixed in PC Express?

[Ron Cahill]

I'd like to add my "two bob's worth" here regarding Procoder 2.0 and MainConcept.
I've used both encoders extensively and found colour-saturation is quite good with Procoder, while MainConcept is about 5% underdone. But I suppose it all depends on peoples priorities.