CCE Basic

[DonS]

I've been using Shrink with good result on most, mediocre results on a few. I'd like to do better. Rebuilder looks like it'd fit my experience level well. I've spent much of the past two evening reading these discussions on Rebuilder.

Throughout these forums there is repeated discussion of "CCE". To which CCE does this refer? There is no possible way that I am going to buy CCE SP, but I might be able to swing CCE Basic. What do I lose with the less-expensive package? Will CCE Basic permit me to do what I need with Rebuilder? Surely there is an added value with the expensive package. What is it and will that limit the quality of output?

Another newbie question. I was going to download a CCE Trial version to check out the capabilities. I read that EclCCE can be used to interface Rebuilder with a trial version of CCE. However, I understand that CCE Trial versions place a logo in the output video. If it is true that outut video is marred in this fashion, why is the EclCCE/Rebuilder combination so popular? Obviously, I'm missing something.

[wmansir]

1. CCE Basic will work fine with DVD-RB.

2. There are several limitations with Basic, compared to SP, but overall they are minor. The only one that will directly effect DVD-RB's process is the limitation of 2 VBR passes, SP can do up to 9. But there are diminished returns by increasing the number of passes, so it's not too big a deal and 2 passes is the DVD-RB default setting anyway. More passes might be nice in a low bitrate situation, but it's not a huge difference in quality.

Also, Basic doesn't have built-in noise filters. On it's own DVD-RB doesn't use them, but some add-on programs will let you turn them on. However, the built-in filters aren't too good anyway and if you really want to do some filtering you would get much better results using Avisynth filters.

There are other very minor restrictions (like you cannot specify I-frame insertion), but they have no impact on DVD-RB's process, or output quality, at all.

3. In addition to allowing the trial versions to open .ecl files, EclCCE adds functionality to the full versions of CCE too. It is needed with DVD-RB and CCE 2.5 because that version lacks commandline support. It also has useful features like: hide the Cancel button (to avoid accidentally canceling a segment encode, which is all to easy in 2.5), run minimized to systray, set CCE's CPU priority, exit when finished, shutdown computer when finished, open multiple ECL files and bitrate tweaking options. Some of these aren't used with DVD-RB, but some make it much easier to work on you computer while CCE is running.

If you do want to try DVD-RB with a trial version of CCE, be sure to download the SP trial because EclCCE doesn't work with Basic. Also, don't forget about the QueEnc option.

[luphy]

One blind mouse leading another:

From what I've read, eclcce is mainly used in two situations: to test out trial versions and be able to save the result, and to use it with CCE 2.50 - to add the ability for commandline arguments. CCE 2.6 and above had that ability built in is what I understand.

And CCE Basic is limited to 2 passes, and does not support OPV (I think).

Addendum: Dang, I only typed a few lines, and wmansir beat me to it with a huge response. Sure....take all the glory from the noob :p

[jdobbs]

A lot of the way DVD-RB works is based upon CCE Basic. In fact CCE Basic is one of the reasons (not the only one) it breaks out by cell. While I sometimes complain about Cinema Craft -- like when they change the meaning of parameters, their encoding engine is just truly phenomenal. Luckily they've included it in CCE Basic.

I suggest you download one of the demo versions that is available right here in doom9s download page before purchasing Basic. It's a lot more comforting to see it work before forking over cash... and the results will be virtually identical (well, except for the demo's watermark)

[archaeo]

And CCE Basic... does not support OPV (I think).

yes, this is true. And it's too bad because it's a feature I'd like to be able to use.

[jdobbs]

I guess they have to give you something extra for the additional $1,900 they charge for CCE SP... :eek:

[archaeo]

Also, I'm not sure about this but would like to know if CCE basic will allow you to adjust 'VBR Bias' and 'Quality precision' settings found in RB's advanced CCE options?

[quantum]

Originally posted by wmansir
Also, Basic doesn't have built-in noise filters. On it's own DVD-RB doesn't use them, but some add-on programs will let you turn them on. CCE Basic does have filters. The difference is you can't specify the parameters as you can with SP. But you do get the 3 presets, and you can adjust the slider from 5 strength levels. Overall I think this is more than good enough. I always stick to the presets and the slider anyway.
However, the built-in filters aren't too good anyway and if you really want to do some filtering you would get much better results using Avisynth filters.I see some people saying this but my own tests don't support it. I'd like to see some screen captures to compare to my own. I hope these comments are a result of comparitive tests and not something that was "read somewhere" and repeated. It reminds me of the old "don't use Nero to burn DVD's" mantra you see around here from time to time.

http://forum.doom9.org/showthread.php?s=&threadid=76662&highlight=tweaker

[archaeo]

Also, I'm not sure about this but would like to know if CCE basic will allow you to adjust 'VBR_Bias' and 'Quality_Precision' settings found in RB's advanced CCE options?


Found out that these settings are FIXED at the default settings when using CCE Basic.

[wmansir]

Originally posted by quantum
CCE Basic does have filters. The difference is you can't specify the parameters as you can with SP. But you do get the 3 presets, and you can adjust the slider from 5 strength levels. Overall I think this is more than good enough. I always stick to the presets and the slider anyway.
I see some people saying this but my own tests don't support it. I'd like to see some screen captures to compare to my own. I hope these comments are a result of comparitive tests and not something that was "read somewhere" and repeated. It reminds me of the old "don't use Nero to burn DVD's" mantra you see around here from time to time.

http://forum.doom9.org/showthread.php?s=&threadid=76662&highlight=tweaker

I stand corrected about Basic's inclusion of filters. But I still disagree about the quality CCE's internal fitlers. I ran several tests when CCE 2.66 was released and found the amount of detail loss was unacceptable, except on the lightest settings. For the most part I found it to be a little better than a basic blur fitler. I'm running some tests now and CCE still isn't looking good.

Unfortunatly the amount of different fitler settings and combinations is pretty overwhelming, so I'm just going to compare CCE's Natural Picture medium (slider in center) preset, to FluxSmooth and Undot.Deen. I'll post my results in a little bit.

[wmansir]

After getting started I realized how much work it would take to do a truly thorough test and present convincing evidence. The problem is visual quality is very subjective. I don't have the webspace for sample encodes and I'm too lazy to take multiple screenshots, host and present them. So I ran a few tests just to see for myself. So the following is just my experience, doing one test.

I encoded a good "clean" source (Eurotrip). It wasn't reference quality, but was a good transfer. I picked a scene with pronounced grain, yet contained some detailed parts (The shot of London bridge just after the London title sequence). It was a fairly typical scene. I picked it because it was bright and had several good elements to judge noise reduction and image distortion. It was about 15 seconds long and contained a bright blue skyline (great for spotting film grain), water, some detailed stonework, leafy trees and human faces.

I encoded the original, CCE's Natural Picture Medium (middle slider) filter, Fluxsmooth(), and Undot().Deen(). All had the exact same settings (even the Quant Prio, even though CCE's filter settings change that, I set it to the Nat-med setting for the other encodes).

I encoded all of them at OPV Q:30 to see how the filters affected file size. The files size results were as follows:

Original:
7885 KB
100 %
CCE Nat-med:
7047 KB
89.4%
FluxSmooth():
7094 KB
90.0%
Undot().Deen():
6893 KB
87.4%

Visual Quality:

I'll divide this up into two sections. 1. Noise reduction and 2. Image distortion. In order to be fair to the CCE output I had to judge the visual quality based on the encoded samples, not the filtered original. However, I did compare the input and output in several instances to separate left over noise from induced artifacts. I made a Q1 encode of CCE's filter for this purpose. (I also did a Q1 encode of the original, untouched source and as expected it was identical to the original visually).

Noise reduction:

The noise we are looking to get rid of is very light, just the film grain really and slight encoding artifacts. This was very visible in several areas, notably the clear blue sky. 1= original 10 = perfectly clean.

CCE: 5
The grain was reduced, but still very noticeable.

FluxSmooth: 7
Majority of grain is gone, still detectable, but much better than the original in this department.

UnDeen: 9.5
Undot.Deen really surprised me here. The sky looks perfect. There is some grain around edges of smooth areas, but comparing this to the input I see it is mostly CCE induced artifacts and only partially unremoved grain.

Image Distortion:
Given the light filtering, and this particular source, the most noticeable effect was lose of detail (or blurring). I didn't notice any banding or ghosting, but those are usually more apparent in dark scenes. 10 = original, 1= do I have my contacts in?

CCE: 7
There were places where CCE's output was the sharpest, like skin tone and some detailed stonework/cement.

However, the problem with CCE's distortion is that it is not nearly as selective as Flux or UnDeen's. There are several areas where Flux and UnDeen's output is virtually identical to the original, however CCE slightly blurred look is still present in these areas. It doesn't look bad per se, but just not as sharp as the original.

Flux: 8
Like I said above, many areas were untouched and looked identical to the original. The areas that were filtered had a slight blur, just a little stronger than the CCE filter, but detectable when comparing stills.

It looked to me like Fluxsmooth had the lightest touch overall, but in the obvious areas (low contrast, gradients) it had the strongest effect (but very close to UnDeen).

UnDeen: 8
Looking at the encoded output Undot.Deen appeared to affect slightly more areas than Fluxsmooth(), but didn't distort other areas as much. In this category it was very very close to Flux's output overall. In fact, comparing the original input they are visually identical in many areas. But whatever differences are present are slightly more evident in the CCE encoded output.

Conclusion:

In this test I would declare Undot.Deen the winner. It did by far the best job in removing film grain and artifacts, while leaving large portions of the video visually untouched, and turned in the best file size reduction. Some areas where distorted more than the CCE fitler, but the overall picture was sharper looking to me.

Runnerup would be Fluxsmooth. Very similar to Undot.Deen in image distortion, but not quite as good at grain removal.

CCE's Natural Picture preset did an OK job. It did acheive roughly the same amount of file size compression as FluxSmooth, however it was clearly the worst in removing film grain and had the largest overall image distortion.

Notes:

Normally I don't use FluxSmooth with it's default settings, which is (8,8), as I have seen that introduce ghosting in some cases. Instead I use (6,6) or (4,4), however I decided to use the default here because it was simpler and since the output size of my sample was very close to CCE's natural preset output size I thought it would make a good comparison.

I have a feeling CCE's filter would be less prone to ghosting and banding then either UnDeen or Flux because of the nature of the fitler. Unfortunately I don't have time to run another set of tests on material which would exibit that better than the one I used today.

I also started playing around with AviSynth's subtract method to help visually compare the impact each fitler has on the image, but that too will have to wait until I have more time.