RB 0.70 OPV oversize

[cd2k]

I'm getting some outputs as much as 4.57GB. I have CCETargetSectors=2260000. With 0.69 lots of outputs as little as 4.1GB. Now too big.

[jdobbs]

That sucks. I'll cut it down a little. In the meantime, I'd suggest you run the output through DVD-RB with ReJig to shorten it a little. The problem I've run into is the lack of granularity of the Q selection. For example I've done DVDs with a Q of 21 that came out to 3.89GB -- then I lowered it to 20, only to get have it go over (4.41)...

I think I may be stuck with accepting undersizing.

[Khauron]

Same here. Tested with clean 0.70-VIP and movie "Starsky and Hutch" R2 PAL. Had CCETargetSectors=2255000 in [Options] and oversized a bit. Never happened before.

[jdobbs]

I've backed that code out and it will come out in the next version. The older version tended to undersize so I adjusted it. Bad idea.

[ShadowKnight]

I apologize if this is a stupid question... but would there be any way to break down each VTS into smaller segments of the whole, so that the Q value could vary more independantly, and possibly create closer target sizes? I dunno exactly how you'd do it, but it would be nice if it could be done.

[jdobbs]

That's exactly what you want to avoid. The Q generally reflects the quality. Changing Q between segments would make noticable differences as you watch it.

[tylo]

Yep, one Q-step may sometimes make quite a big difference in size, so you must settle with the Q that creates below target size. The good part is that normally Q-step sizes are big when Q is low. So you will get good quality anyway.

Alternatively, if oversizing a percent or two, a fast run through Shrink won't reduce the quality at all, and you can burn it in the same go...

[jdobbs]

That's exactly where I saw it. It was a change between a Q of 11 and 12 as I recall.

[ShadowKnight]

I'd like to chime in here and say that my encodes so far have been a lot better with the new sizing. Is it possible to keep the code in as an option? You could put a warning that it may oversize on it just in case. I haven't had any problems with oversizing, but then again I'm also not changing my ccetargetsectors in the ini. Maybe that has something to do with the oversizing? I'm not doubting that its possible it can oversize, but it seems to be pretty good for me so far.

[ron spencer]

Yes....can we leave it as an option if this is true...it is really easy to run through DVDshrink if need be

[Sir Didymus]

... or to just manually reencode one or two cells, and rebuild, if needed...

Me too almost happy about actual 070 OPV behaviour... :)

[ShadowKnight]

I was wondering if the changed predictive method was changed in the newer versions, and I would still like to suggest it gets added in as an option.
I've run about 5 movies (mostly episodic discs) through DVD-RB v70, and they have all come out to anywhere between 4.20 and 4.34.
I haven't had time to check the newer versions yet, I will tonight and post back any differences in size that I notice with these discs.

[jdobbs]

I backed out the new predictive code because it was oversizing too often. If it makes sense to do so, I can reintroduce it.

[XMEN3]

oversizeing for me too with new cce2.7...never before with 2.5

[jdobbs]

Hmmm... I haven't really tested it with v2.70 beyond just making sure it worked. Have you changed the CCE TargetSize from the default?

[XMEN3]

CCETargetSectors=2258000 but with 2.55 was always around 4.690.000

[VamPYR]

Never had oversize issue with RB.
Currently using RB 0.72, CCE 2.70.01.05, AviSynth 2.55, DGDecode 1.10.
mostly get between 4.31G - 4.34G, maximum size are 4.36G (rarely happen)and never go over 4.36G.
use default setting, unmodified source.
cheers.