any panasonic MPEG1 encoder junkies there...?

[deltaboy]

i understand that most people think that tmpgenc is the best encoder out there, but i am one to differ. i know there are far more tweaks and setings within tmpgenc, but why not just use a program like virtual dud to fine tune your AVI before inputting it to an encoder. the one thing that keeps me clinged to panasonic is the encoding time. WOW. it is way faster than tmpgenc by far! i can encode a half hour af video in about 1 hour with panasonic, whereas with tmpgenc, it takes me an astounding 2 and a half hours!

i am still in the process of fine tuning the advanced settings within panasonic. i would like to hear peoples feedback on what i have to proclaim:

under Image Filter, i am still not completely fluent in what setting will produce a better, cleaner, running video (for high action scenes, of course).

i use a weak Noise Reduction, a medium fixed Video Filter, with the Color Tone on PC. does this sound correct?

what about the Quantizer Matrix? should i keep it at natural? and i have no idea what to do with the GOP Sequence.

anyway, thanks for the time in reading my post and good luck to all the quality freak encoders. keep up the good work...

[aldus4]

If I do encode VCD's I too use the Panasonic MPEG Encoder, but more regarding quality not time. It was interesting to hear you say that the PWI was faster than TMPGEnc as so far all test results published by other websites said the opposite. My question to you is what processor and what OS are you using?

[deltaboy]

REALLY!!! on many comparisons i have read that tmpgenc is slower but inturn creates a better output. but since i mainly create snowboard and skateboard VCDs, i am dealing with very fast action scenes - constantly. i do believe that PWI give me a much smoother looking video, thus degrating quality only a little, but this is indiscernable to me.

oh ya! sorry man, i am using a dual PIII 866 system with 512 RAM and dual ATA100 RAID controlled 40G HDs run under a Windows 2000 box.

but why would my proccessor and OS have any controll over which software would run faster? arent they all in retrospect to whatever speed you are running?

[aldus4]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-07-18 17:22:11, deltaboy wrote:
arent they all in retrospect to whatever speed you are running?

</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

I thought that too. But as you know sometimes software is specially tuned to a certain processor type.

[disturbed1]

I have a Duron 850 w/256mb, and PWI is slightly faster than TMPGenc. On default settings, TMPGenc(beta A, & H) will encode an hour in ~3.5 hours. PWI with the same clip does it in just under 3 hours (like 2:53x).

So it isn't just Intel chips. Also to my knowledge PWI doesn't take advantage of multi cpu's. But TMPG does.

<font size=-1>[ This Message was edited by: disturbed1 on 2001-07-18 18:35:39 ]</font>

[vitualis]

That's interesting... I'm sure that something else is at play here. I thought that the newer versions of TMPGEnc (that is those > 12a) were SIGNIFICANTLY faster than PWI.

For TMPGEnc, are you setting the "motion search precision" setting to "high" or "highest"? The "highest" setting doesn't yield much better video than the "high" setting but is much slower.

TMPGEnc has optimisations for both the PIII (and P4?) and Athlon. PWI only has optimisations for MMX.

I too use PWI because to my eyes, it is better. This is, however, subjective, and to some people, the sharper video output of TMPGEnc looks better.

As for the filters with PWI, if you are doing a DVD rip, turn them ALL OFF. If you are doing a video capture, I usually don't use anything more than:

Noise reduction: weak
Video filter: adaptive --> weakest or weak

This can work wonders. However, I find that setting it anything higher leads to too much loss of detail. As I stated before, for DVD rips, turn these off!! PWI probably softens the video too much by default already.

Quantizer Matrix -- I've never tried changing it from the default, and according the help file, it is the best setting for everything (including CGI).

GOP sequence -- for standard VCDs, just leave it on the default setting.

Regards.

[disturbed1]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-07-18 17:03:14, deltaboy wrote:
under Image Filter, i am still not completely fluent in what setting will produce a better, cleaner, running video (for high action scenes, of course).

i use a weak Noise Reduction, a medium fixed Video Filter, with the Color Tone on PC. does this sound correct?

what about the Quantizer Matrix? should i keep it at natural? and i have no idea what to do with the GOP Sequence.

</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

As far as the difference in filters, only your eyes can be the real judge to quality. The soften block noise does just that, it adds a soften image filter that tends to remove unsightly block noise that's appearent with the MPEG1 technology.

The video filters will add a slight noise reduction that may help to remove *fuzzies?*. The fixed setting is fixed, the same amount through out the clip. The adaptive filter is said to do a type of prerendering that may look at the clip and decide what need filtered and what doesn't. Both of these are similar to LSX's median filters.

The quantitizer matrix is how the encoder decides to encode each macro block with reference the said DC coefficient. Check out BBmpeg's help files, it offers indepth explainations for this.

The GOP is a Group Of Pictures. This includes I (intra frames, key frame, the most data is here, least compressed, this is the framed that is referenced when searching and such), P (can be considered Predicted frame, has more compression, less data than an I frame, this frame too can be referenced), and B (maybe Bidirectional, these frames have the most compression, and are not referenced).

According to VCD spec's the GOP should contain a set sequence following the IBBPBBPBBPBB order with the distance between I frames being ~.5 seconds. That's A GOP size of 15 frames(IBBPBBPBBPBBPBB) for NTSC 30fps, and a GOP size of 12 frames (IBBPBBPBBPBB) for PAL 25fps, and FILM 24fps. The GOP has a direct reflection on quality and overall size. With more I frames placed closer together one could expect better quality, but a larger size due to less B and P frame compression.

[disturbed1]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-07-18 19:06:27, vitualis wrote:
That's interesting... I'm sure that something else is at play here. I thought that the newer versions of TMPGEnc (that is those > 12a) were SIGNIFICANTLY faster than PWI.

For TMPGEnc, are you setting the "motion search precision" setting to "high" or "highest"? The "highest" setting doesn't yield much better video than the "high" setting but is much slower.


--- snip snip
Regards.
</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>
I just set the quality to High, not highest. The files I encoded were 352x240 29.976 video only avi's compressed with Huffyuv forced YUY2. This way it doesn't test the encoder's resizing capabilities/speed, or audio engine, only their ability to encode.

Of course doing DVD rips VIA DVD2AVI yeild the same results, PWI is always faster than TMPGenc, BUT, I use AVIsynth w/MPEG2dec.dll for PWI, and the unaltered .d2v file for TMPG. The speed increase is expected there, as TMPG needs to resize, but AVIsynth's bicubic resize is pretty speedy.

[KoalaBear]

Panasonic MPEG-1 Encoder: Budding Quality Freaks Take Note

Special Processing:

Panasonic is an MPEG encoder, not an image processor, so virtually every option on this screen should be considered "for emergency use only."

If your input file requires cropping, letterboxing, deinterlacing or resizing, use a program like VirtualDub or MediaCleaner to process the image beforehand, and let Panasonic concentrate on what it's good at -- encoding.

Advanced Settings:

• Quantizer Matrix -- leave at "Natural Image".
• GOP Sequence -- There are only two legitimate reasons to change this from the default 15/15/3: (a) you need to create an MPEG file using I-frames only, or (b) you need to align the GOP for inverse telecine [see below].
• Noise Reduction -- If the quality of your input material is high (such as a capture from DVD, satellite, digital cable, etc.) you do not need to use this filter at all. Again, if the image requires cleanup, do it in VirtualDub.
• Video Filter -- same as above.
• Color Tone Correction -- If your MPEG file is intended for computer display ONLY, select PC. If the file is intended for playback on a DVD/VCD player, select TV. Despite it's name, this setting controls the range of brightness values a pixel can take, not its color. TV requires you to reserve space for "blacker than black" and "whiter than white" signal components, so this switch tells the encoder that "black" = 16 and "white" = 235 (rather than "black" = 0 and "white" = 255, which is correct for a PC).
• Color Control. Allows you to fine-tune the color balance. Another function that should be performed by Panasonic only in an emergency, and by VirtualDub all other times.

Inverse Telecine:

Film is recorded at 24 frames per second, video is recorded at 30. When a film is transferred to videotape, a device called a telecine interpolates an additional six frames per second by slicing and dicing fields from adjacent frames in a 3:2 pattern to make up the difference. Using VirtualDub, it is possible to undo this process so that the film is restored to 24 fps. Encoding a file at this frame rate effectively boosts the bitrate by 20%, and your DVD/VCD player performs the frame rate conversion on-the-fly for playback if necessary. However, it doesn't make sense to use a 15-fps GOP for a 24-fps video, so this is one case where lowering the GOP to 12/12/3 makes good sense.

The moral of the story is that Panasonic is a fabulous MPEG encoder, but a lousy image processor. Leave the extra settings alone unless you absolutely have to change them, and you'll thank yourself every time you play your homebrew VCDs.

[vitualis]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-07-18 19:43:16, KoalaBear wrote:
Panasonic MPEG-1 Encoder: Budding Quality Freaks Take Note</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

Thanks for the info. However, I disagree strongly with this:

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>[*] Color Tone Correction -- If your MPEG file is intended for computer display ONLY, select PC. If the file is intended for playback on a DVD/VCD player, select TV. Despite it's name, this setting controls the range of brightness values a pixel can take, not its color. TV requires you to reserve space for "blacker than black" and "whiter than white" signal components, so this switch tells the encoder that "black" = 16 and "white" = 235 (rather than "black" = 0 and "white" = 255, which is correct for a PC).</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

This may be true, but on the TV setting, the resulting MPEG looks like absolutely crap (faded out colours). Just stick to the PC setting.

Regards.

[KoalaBear]

Actually, the problem you'd like to avoid is one of overload. If the luma value of a pixel exceeds the whitest value the TV can display, it results in an ugly quantization artifact that looks like this:


See Digital Video, MPEG and Associated Artifacts for more details

Actually this is a grossly exaggerated example. In reality the overload tends to follow highlight contours, particularly around the edges of objects.

The saturation problem you've referred to is a horse of a different color. TV phosphors are like klieg lamps compared to the tiny precision phosphors of a computer monitor. It is very difficult to create a picture that looks good on both regardless of which "Color Tone Correction" setting you choose.

[Anonymous65]

There should be no difference between a picture tube for a PC monitor and a TV set. The only difference is that the shadowmask has smaller holes on a PC monitor than a TV monitor, otherwise both monitor looks and works the same. They have the same reproduction of colors. I do not understand why Panasonic has this options, they should know better. Anyway, do not use the TV setting on the Panasonic encoder, the picture will look wrong on a TV set and a PC monitor!

[vitualis]

I don't pretend to know the techinical details behind it, but using the "TV" setting with the Panasonic MPEG Encoder leads to horrible washed out colours which is grossly different from the source (e.g., comparing original DVD to VCD on the TV).

Using the PC setting leads to very good colour fidelity for an MPEG-1 encoder (again, compared to the source).

The problem shown in the above picture has never happened to me in the hundreds of MPEGs I've encoded.

BTW, that was a pretty good article you linked to KoalaBear. Thanks!

Regards.

[KoalaBear]

The important point is that no amount of fiddling with Panasonic's advanced features can turn a sow's ear into a silk purse. The quality of the output is directly proportional to the quality of the input, and no amount of parameter tweaking within the encoder will compensate for an AVI source that is less than optimal to begin with. In fact, it could make the end result much worse.

Sounds self-evident, I know, but it was one of those lessons that I needed to learn for myself the hard way.

I'm familiar with the saturation problem, but I believe the culprit is the nature of MPEG compression itself and not the PC/TV setting, though toggling this value may provide an apparent improvement. At MPEG-1 bitrates, the higher the color fidelity, the higher the number of compression artifacts, because one must rob Peter to pay Paul. It's a zero-sum proposition.

According to MPEG compression theory, chroma is the first thing to be sacrificed in lowering the overall bitrate because the eye is more susceptible to changes in brightness than color. A great number of bits can be saved by scaling the chroma precision alone, leaving more bits available to compress motion between frames, where loss of precision is much more evident when it occurs.

But unless the luma is scaled proportionately along with the chroma, more brightness + less color = bleachier picture.

On my system (your mileage may vary of course) AVI captures appeared too dark and oversaturated to be "correct," so I would adjust the brightness and contrast levels in the capture driver to whatever looked "right" for the material I was capturing at the time.

Running these files through Panasonic usually resulted in faded, washed-out looking MPEGS, so I tried boosting the chroma levels in the Advanced Settings menu to compensate. They looked good on the PC, but when I would burn them to disk and watch them on televison, the image would again appear to be faded and bleached, so I'd reach for the TV color controls to compensate for *that*, and voila, I discovered the vicious cycle I was trapped in.

These days I leave the capture driver settings at completely neutral levels and capture the video as it comes in, using the brightness and contrast controls on my computer monitor to adjust the picture gain to taste. The captured video still looks darker on the monitor than I'd prefer, but the MPEG output from Panasonic is well-saturated with excellent contrast. And when I burn these MPEGS to CD, the brightness and color look great on my TV with no further picture adjustment necessary.

Monitors and televisions are just too different in their frequency responses for a single MPEG encoding parameter to fix. Calibrating the monitor certainly helps, but it cannot be relied upon to give you a true idea of what the TV will display any more than the TV will give you a true idea of what the PC monitor should look like.

In other words, undersaturation may be a symptom and not the disease in itself.

[Stinky]

I'd just like to add one little bit to this thread.

Hard to convert DivX in TMPGEnc = Error, conversion fails

Hard to convert DivX in Panasonic = VCD!

I love my Panasonic encoder!

[vitualis]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-07-19 09:52:57, KoalaBear wrote:
On my system (your mileage may vary of course) AVI captures ppeared too dark and oversaturated to be "correct," so I would adjust the brightness and contrast levels in the capture driver to whatever looked "right" for the material I was capturing at the time....[edit] </BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

As you've found from experience, the BEST setting for your capture card is on NEUTRAL.

The AVI captures look too dark and colours oversaturated because on most PC monitors, the gamma is incorrectly set and Windows doesn't natively have an app to correct it

If you install Adobe Photoshop, it also installs an app that will allow you to correct the gamma (+ other things) of your monitor, so that the colours displayed on screen is a mostly accurate depiction of the REAL colours.

With your monitor correctly setup, the video on your PC and TV will look pretty much the same.

Regards.

[Aquarius]

I've been using Panasonic MPEG-1 encoder for some time and there is one thing I can't figure out. I'm unable to fast forward SOME clips on Philips DVD/VCD players. It's usually the fist one (I gather closest to the disc's centre) that fails to do so. I heard that Panasonic-made MPEG-1s lack some header info that are a must for some Philips DVD/VCD players to fast forward VCD clips, but please correct me if I'm wrong. I used to burn my disc with Nero, but recently I have switched to WinOnCD 3.8 PE.
Regards,

[vitualis]

Philips VCD/DVD players can't FFW or REW without regular sequence headers before each GOP. This can be fixed like so: http://www.geocities.com/medinotes/vcd/chapters.html (it's under "Panasonic MPEG Encoder&quot.

There's actually a page on this very issue somewhere in the VCDHelp guides (look to your left) but I can't seem to find it.

Regards.

[Aquarius]

Thanks a lot!
Regards,
Peter

[twilie]

Can anyone tell me where I can get panasonic mpeg encoder?
I'm having trouble with Tmpgenc and would love to give this a try.
I can only find the trial version and it only allows 30 seconds of encoding..while
Tmpgenc doesn't crash on me till almost 2 hours..30 seconds won't help me
figure out if it too will crash ..I'm wondering if there is a trial (or something)
for 30 days or so ..that way I can see if it will encode a whole movie for me..
thanks in advance

[abularkin]

twilie, twilie, twilie...
Check your private messages.
BTW, this isn't about encoders OR warez. It's just about the private message sex.