Tmpgenc DC component precision ???

[kenmo]

Can someone exlpain this setting to me. And which is the better setting 8, 9 or 10bit. I'm guessing 10 but....

Thanks muchly

[puertorican138]

DC component precision is used to make the movie look more smooth (fluid). Set it to 10 for highest quality.

Hope That Helps!!! :P

[adam]

I have to disagree. Geneally speaking, higher dc precision is better but the problem is that it uses up bits, which in most cases would be much better spent on the actual video.

I cannot give a real explanation of what this does, its over my head but TMPGenc's description of making the video "smoother" is really misleading.

According to every single mpeg2 document I have ever read, high dc precision should be used on smooth pictures with little movement and low should be used on intensive movement. More importantly, you should not use a high dc value for bitrates under 4Mbits, so that rules out SVCD and probably even DVD-R. In fact, most commercial DVDs don't even use a DC precision of 10, they use 8 or 9 or a variable amount like what CCE's auto setting can achieve.

My advice is that if you are using TMPGenc, stick to a value of 8 unless you are using an avg bitrate at or above 4Mbits, and then use 9. If you are using CCE stick to 8 unless your avg is at or above 4Mbits, and then use its Auto DC setting.

I have heard that a higher DC precision may achieve better results on HDTV's so take that into consideration, but if your avg isn't anywhere near 4Mbits then its a non-issue, as 8 will always yield superior results.

[SatStorm]

My observation in this matter, concerning TMPGenc's settings, is like this:
For CQ encodings, use 8 bit. You can try 9 bit if you set up to 90%

For CBR encodings, it depends. For example, a CBR @ 9000 looks better with 10 bit than 8 bit. IMHO, it seems that there are thresholds on this: less than 5000kb/s is 8bit. 5 - 7 is 9 bit. 7 and higher is 10 bit.

For 2 Pass VBR encodings, I don't suggest 10 bit at any case. Only if you use an average more than 6000kb/s, you can try 9 bit. But 10bit never...

If you emulate 2 PAS CBR with TMPGenc, 10 bit is suggested always.

Personally, I use 8 bit, because that way the picture looks better (in my eyes) on the lower bitrates I tent to use (for example, an average of 2800kb for 1/2 D1 framesize projects)

I don't know the situation for the new TMPGenc version (3). There are reports that things changed (once again....) with that version. I don't own this version, and I can't report my thoughts an observations about it.

[lordsmurf]

Well ... 10 is out ... we'll get that one out of the way right now.

So 8 or 9 ... hmm.

With MPEG2, I generally stick to DC9.

For MPEG1, I often use DC9 for 1850k DVD encodes.

I've tried to see a difference between DC8 and DC9. I really can't.

DVD recorders use DC10, and I'm not really sure why.

Commercials DVDs seem to use a lot of DC9.

I forget what ATI AIW uses (DC8 I believe, have forgotten).

Do a sample of each and go from there. I think my beef with TMPGENC DC8 was it didn't really do a good job at it. At some point I moved to mostly DC9. I forget what project that was for, sometime last fall.

I don't think PROCODER even let's me pick. I never looked in MC. Don't use CCE.

Most of my encode are MPEG1 at 1850k CBR at DC8 or DC9, or MPEG2 at VBR_CQ 85% of 4000k using DC9.

[SatStorm]

You might see a difference mostly on low bitrates and multipass VBR projects.

Also it seems to me that huffyuv likes encodes with DC 09 and 10 and PicVIDEO like encodes only with DC 8. I don't know how and if this is possible, but I really think that when I capture with PicVIDEO Mjpeg, the 8bit encodes I do are always better 9bits, even at higher bitrates!
Of course, I can always be wrong on this...

[Sulik]

FYI, MPEG-1 only supports 8-bit DC. Any SW that allows you to choose 10-bit DC precision for MPEG-1 has to use 8-bit anyway.

Using more than 8-bit DC is pretty much equivalent to lowering the (0,0) coefficient in the intra quantization matrix. I have yet to see a sequence where it improves the quality at the same bitrate (it mostly applies to I-frames anyway).

[lordsmurf]

FYI, MPEG-1 only supports 8-bit DC. Any SW that allows you to choose 10-bit DC precision for MPEG-1 has to use 8-bit anyway.

Using more than 8-bit DC is pretty much equivalent to lowering the (0,0) coefficient in the intra quantization matrix. I have yet to see a sequence where it improves the quality at the same bitrate (it mostly applies to I-frames anyway).

I sort of thought that, but TMPGENC shows DC9 anyway (though grayed out). Probably why I never see a difference here. Thanks for the head's up.