image quality quality, conversion, references, settings

andy

Since some of my D2SRoBa encodings came out with some sort of mosquito noise that is also noticable on my 16:9 tv I was wondering what is recommended to correct this. The CCE manual claims that the quantizer characteristics and bias settings can help. Are there any references for these settings in case of dvd2dvd conversion?

Cheers.

Hi maSp,

I use an IQP of 16 and a bias of 20 (with CCE v2.66.01.07). These settings are ok for CCE v2.50 as well. What were the Q factors of the encodings that you notice the problem on? If they were high Q (gt;30), the quality may benefit by enabling the additional resizing pass in D2SRoBa.

Mosquito noise is a thorn in the side of CCE (probably the only thorn), but I'm very surprised you notice it on many encodes. Lowering the IQP generally reduces the mosquito noise, but may increasing contouring noise (noise on flat surfaces).

I have never adjusted the bias, so I can't comment on that.
Regards,
Bob

bobwillis, a useful and well written answer - like you always do. I appreciate that.

I always force the sizing pass with 0.0 lt;gt; 0.0. However, the settings I fiddled with were IQP 17 and Bias 30 (D2S's default), CCE 2.66.01.07, source 'clean'. What's puzzlling me is the lower the Q the higher the noise. Shouldn't it be the other way round?

Guess I'll do some test samples with lower IQP's and will have a look at CCE 2.50 as well.

I know there's no such thing like 'the best settings' and it depends on the source but my goal is to get the best out of it. Hard work for a newbie like me and without the help of this great forum I would probably get lost.

Regards,
maSp

Originally posted by maSp
What's puzzlling me is the lower the Q the higher the noise. Shouldn't it be the other way round?

This is strange, you're right, the lower the Q the better the encoding should be. Generally, with Qs less than 20, I struggle to see the difference between the source amp; the CCE encode.

I think it's important to find out if you're suffering these problems because of the average bitrate being too low or because of the CCE encoder settings (IQP, bias, filters etc).

What you could do is encode a small clip of your material (say 1 minute) with a Q equal to 1 (this can be done with DVD2DVD without D2SRoBa). Doing this will re-encode with the highest quality possible and will remove any effects of a constrained bitrate. DVD2DVD will encode amp; re-author a 1 minute clip in under 5 minutes.

With Q=1 you can then methodically experiment with each CCE setting until you hopefully eliminate the problem (if possible). Once you've determined the best settings, you can then re-encode with a specified Q (the one used previously for the whole movie) and see if the quality deteriorates again. If it does then the problems are due to constrained bitrate and not the settings. If this is the case, you could try TMPG, but I found this can severely macroblock with fast action scenes.

Hope this helps,
Bob

I'm testing right now (some small clips). It's getting better. IQP 16 and Bias 20 seems to be a good point to start with. D2SRoBa determined a Q of 10 for the whole movie so I expect to get a good result. Thanks again.

Cheers.

Generally, with Qs less than 20, I struggle to see the difference between the source amp; the CCE encode.

I fell into a habit of always doing the conditional pass but like you Bob, movies with Qs under 20 look fine to me. So, if the resulting encode fits under my target, are there any other compelling reason to do the conditional pass?

In higher Q material enabling the extra sizing pass will improve quality significantly.

showthread.ph...g+pass+quality

Nick,

Right,I agree as Q creeps higher and higher, particularly over 20ish, the conditional pass improves the image.

What I was trying to establish is for encodes where Q is UNDER 20 and if the size is within a few percentage points of the target, what does one gain in terms of image quality by doing the conditional pass?

Of course the conditional pass creates a new encode likely be spot on the target size but that I don't believe that adjustment alone improves the visual image quality.

Sorry about that. I didn't read your question properly.

(Mental note READ the question, THEN answer  )

As far as low Q factor encodes go, I personally cannot tell any difference with the conditional sizing pass.  I do not feel the improvement warrants the extra time until Q reaches 30+.  As ever that is with my eyes on my TV at my house.  Others may see things differntly but around 30 is the cutoff point for me.

As you say yourself, the lower Q encodes look fine to you.  They are your backups, so if they're good enough for you there's no-one else you need to please with the quality, so you may as well save yourself some time!

Anyway, sorry for the confusion.

Nick