TMPGEnc Plus - CCE 2.67 SP - MC 1.4.2

[LSchafroth]

I tried all three encoders last night. I wanted to see which encoder I want to stay with. I've read so many opinions and wanted to see for myself.

Captured 8mm analog home movies at Half D1, using PicVideo 19. Use AVIsynth with convolution3d to feed encoders. Crop noise from bottom of screen and pad top and bottom to center it.

All were 2 pass VBR on a 1.40s clip. 1000min, 2200avg, 3500max

TMPGEnc Plus (latest version)
5:03 encoding time
24,472Kb

MainConcept 1.4.2
1:40 encoding time ( I don't think it is doing 2 pass even though it is set to)
52,365Kb

CCE 2.67.027?? (trial)
4:07 encoding time
26,931Kb

3-Pass VBR
5:20 encoding time
26,920Kb

I see no difference in the video quality between the four at a quick glance last night. I will burn them to DVD and see if I see any difference on the tube.

Does anyone know how to make MC do a 2 Pass VBR? Also, why does CCE create three files? One MPV, one MPG and VAF. I only need the M2V file for DVD-Lab. I will use BeSweet for the AC3 on all of these.

Any suggestions welcome.

LS

PS If there are any good tweaks or guides for setting up CCE or MC, I would appreciate a link to them. I did a quick search last night, but did nto have time to do a good in depth search.

[adam]

Just so you know, you are not doing 2 pass VBR in CCE, you are doing 3 passes. If you set it to 2 passes, it does one pass to create the .vaf file. Then it performs an additional 2 passes. Its creating three files because you have it set to output to file after each pass. So its outputting to a video stream (mpv) after the first pass, then giving you your mpg program stream after the entire conversion is done. The .vaf file is of course necessary for any multipass conversion in CCE, since this is where all of the information is stored. Just read the manual, it explains all of this.

One major thing I have to point out is that your Mainconcept file more than twice as big as the other encodes, which means you used twice as much bitrate. If you want to make a fair comparison of quality you need to get your outputs to roughly the same size.

Also why such odd min and max settings? If you are making a DVD take better advantage of VBR. Use a lower min like 300 or 500 or 0, and raise your max some. Let it peak up to 9mbits if it needs to. If you are worried about the encoder not using enough bitrate on simple scenes then that is something you should set in the encoder (ie: image quality priority, bias, picture spoilage etc...). Don't just arbitrarily limit your VBR range, it only lowers quality.

As for MainConcept, sorry I've never used it. It seems like a great encoder but I'm more than happy with CCE so I haven't bothered to make the switch.

[LSchafroth]

Just so you know, you are not doing 2 pass VBR in CCE, you are doing 3 passes. If you set it to 2 passes, it does one pass to create the .vaf file. Then it performs an additional 2 passes. Its creating three files because you have it set to output to file after each pass. So its outputting to a video stream (mpv) after the first pass, then giving you your mpg program stream after the entire conversion is done. The .vaf file is of course necessary for any multipass conversion in CCE, since this is where all of the information is stored. Just read the manual, it explains all of this.

One major thing I have to point out is that your Mainconcept file more than twice as big as the other encodes, which means you used twice as much bitrate. If you want to make a fair comparison of quality you need to get your outputs to roughly the same size.

I have all of them set to the same bitrates. It appears to refuse to do 2 Pass VBR, hince the size difference. the settings are identical.

Also why such odd min and max settings? If you are making a DVD take better advantage of VBR. Use a lower min like 300 or 500 or 0, and raise your max some. Let it peak up to 9mbits if it needs to. If you are worried about the encoder not using enough bitrate on simple scenes then that is something you should set in the encoder (ie: image quality priority, bias, picture spoilage etc...). Don't just arbitrarily limit your VBR range, it only lowers quality.

I've seen those values recommended in a lot of posts so that is what I went with. I've seen people state that 9000 is a waste of bitrate when using Half D1 resoluton. I read that anything above 3500-4000 is useless.

I'm new so anything that I'm doing is from what I've gathered from other's points of view. I'm looking for any critics to correct me!

LS

[adam]

I have all of them set to the same bitrates. It appears to refuse to do 2 Pass VBR, hince the size difference. the settings are identical.

Yes I understand. I just meant that with these current settings MainConcept will have a huge advantage when comparing quality. However you go about equalizing the size, you need to do so in order to compare quality. Someone else will have to help you there since I have never used MainConcept.

I've seen people state that 9000 is a waste of bitrate when using Half D1 resoluton. I read that anything above 3500-4000 is useless.

And I would generally agree with that, but if that is true for all of the given scenes in your source then the encoder will never use that much bitrate anyway. The point is not to arbitrarily limit it when it would, in fact need to raise the bitrate above 4mbits, or lower it below 1mbit, for example.

The general rule when encoding in VBR is to provide the encoder with the greatest interval from which to select an applicable bitrate. So 0 all the way up to the DVD max is the best way to go. Let the encoder make its own decisions, pursuant to your guidelines (ie: avg bitrate plus quality controls like bias etc...)

[LSchafroth]

I ran MC again and told it to do 2 pass VBR with 500min, 2000avg and 6000max and it finally encoded it as a 2 pass.

with a 1000min, 2000avg & 3500max it will do CQ every time.

Strange. I put all three encoded files to DVD and they all look identical. No difference in edges, blurriness, nothing.

Is it the source material's quality that makes it difficult to see any difference at all between the encoders?

It's very good quality 8mm analog home video. Looks very clean when watching on TV.

LS

[LSchafroth]

Its creating three files because you have it set to output to file after each pass.

I downloaded the PDF manual for 2.66e but I am using 2.67 so the screenshots in the manual do not match. There is nowhere I can find to tell it to NOT create the multiple files. I gave each step the same file name and it only created one along with the VAF.

I see no speed advantage with CCE. It's moderatly faster then TMPGEnc and slower then MC. I think I will stick with TMPGEnc for now since all the guides and templates are geared towards it. I've found a few guides for MC and CCE but none go into detail. I like MC the best out of the three in ease of use and speed and quality.

LS

[adam]

Are you sure you are comparing apples to apples? Again if you set TMPGenc to 2 pass and CCE to 2 pass then that is definitely not a fair comparison, because CCE is actually doing 3 passes instead of 2. CCE's multipass set to 1 pass = TMPGEnc's two pass. Just in case you haven't retested these two encoders yet.

Also CCE has a huge speed advantage with YUV sources. It can encode them directly whereas TMPGEnc, and maybe MainConcept has to convert it to RGB first. Encoding with a YUV source is about 30% faster for me in CCE then with an RGB one.

No offense, but I think with further testing you'd be able to achieve more accurate speed results in your encoders. MainConcept is indeed very fast, probably faster then CCE. But both encoders are leaps and bounds faster then TMPGenc. For my sources, which again are YUV, CCE encodes just under 4 times faster then TMPGenc.

But if you don't see any quality difference between the encodes, and you aren't willing to spend money just for faster encodes, then TMPGenc is the ideal encoder for you.

[racer-x]

I just want to put my 2 cents in. I can't tell the difference in quality between 1 pass and 2 pass on Main Concept encoder, therefore I only use 1 pass.

Im my tests with CCE vs Main Concept, CCE needs at least 3 pass to equell the quality of 1 pass on Main Concept. At that rate, it's about 4 times slower. These test were done with DV footage shot in Frame Mode and encoded @ 30p. The settings were Identical @ 3500 min - 6500 avg - 8500 max. The Main Concept file was slightly smaller then the CCE file (1.35 gig - 1.36 gig).

From my conclusions, Main Concept is the better encoder.

[LSchafroth]

Also CCE has a huge speed advantage with YUV sources. It can encode them directly whereas TMPGEnc, and maybe MainConcept has to convert it to RGB first. Encoding with a YUV source is about 30% faster for me in CCE then with an RGB one.

I'm trying to remember off the top of my head here at work. can you force PicVideo to encode in YUV12? I do not recall. I will most certainly try it if I can get that much speed out of it.

I did notice that while MC is encoding a file I can browse the web, check my email and do other applications with no problems. when I fire up CCE, I don't dare do anything else as the system is already swapping to virtual memory!

I want to try a 720x480 capture with reduction to 352x480 again and see if there is any difference. My first trials showed no difference, but I was also using the wrong settings in PicVideo at the time. I'm using a BT chipset and heard soemthing about resizing works better with 704 instead of 720. anyone heard of that?

LS

[LSchafroth]

I had decided on CCE for my encoder of choice. Then last night, all encodes with CCE came out blocky as hell.

I've reverted back to the old trusty TMPGEnc. It is so painfully slow, but it works every time. CCE & MC do soemthing different every time I sit down to start the long project. I will have to settle with TMPGEnc and take the extra time necessary.

As for AVIsynth, I know nothing about it's filters. I've been using convolution3D because that is what I was show how to do.

Could anyone list their favorite AVIsynth filters for analog video? I would like to see a few different ones and try each one for myself.

THanks!

LS