CineVision encoding well lower than set target bitrate

[asweepe]

I've been in the DVD/BD/UHD design and authoring business for over 20 years now. My company still uses CineVision for Blu-ray AVC encodes. Overall we have no issues and it yields great results. We've been doing a lot of anime work lately and I've noticed the encoder is encoding well below the set target bitrate. For instance if I set a variable rate of 29 mbps and a max of 35 mbps my final encode will have an average of 22 mbps. I understand that most anime is very simple, solid colors with not a lot of movement so often times the lower bitrate is just fine. The issue is that my bit budget gets all screwed up and I can't fill the disc, which to some clients and consumers is an issue.

Aside from doing a CBR (which can't be segment re-encoded in CineVision) is there a way to force the encoder to respect the target bitrate that is set regardless of the complexity of the video? I've tried several tests with different quantization settings as well as adjustments to the AQ.

I realize that CineVision is no longer supported or sold by Sonic or Rovi and I was given access to download the entire Sonic knowledge base before they closed up shop. There are no articles in the knowledge base that reference this issue.

Any help would be appreciated! Thanks much!

[kolak]

It's not very good encoder at all, but such a bitrate difference sounds strange. I assume you do 2 pass encode?
Other option is to ditch CineVision. I think you can get good deal on Cinemacraft HDe today. It's by faaaar batter encoder (and very fast) with very good segment re-encoding and tons of other features. IF you do a lot of work you would benefit and get return quickly.
If you're interested pm me.

[mp3dom]

Is this happening on every source or just from some of them? Maybe you've already reached the optimal bitrate with such average bitrate. You can try to see if there's the option to use stuffing bits.

Cinemacraft is probably still "supported" but no more developed. The encoding engine is still at... 2014/2015 maybe? And it's "suffering" now from the new kind of sources coming from 4K downlscales or new 2K datacine, where there is way more definition and "high frequencies" to store. SiriusPixels is more updated but probably way expensive. In such circumstances, I would stick with x264 as alternative.... not perfect but not bad either.

[kolak]

All h264 engines are from that time. Not much has really changed.

There are only minor updates based on Sirius website- rest is the same code as HDe. SiriusPixels (knowing Ray) will have price which today has no place Not worth over HDe at all, which you can get at "good" price.
If you can give HDe 25Mbit+ for average then you should not have much of a problem with any progressive source (just don't use default settings).

[mp3dom]

They've added adaptive matrices and deblock (both fixed on HDe) which could improve a little (but guess not so dramatically), plus other kind of filters to limit the LPF effects, mosquito noise around edges and grain retention. Mosquito noise is very visible now with HDe even with maxed-out bitrates, especially on new anime sources with very thin and solid lines. Unless you apply some sort of LPF which somewhat "soften" the lines (if you apply it gently) or even show the classical haloing (if you apply it in stronger way). That's indeed what I also see from commercial blurays coming from HDe/Sirius encodes... complex scenes shows very strong chroma LPF (less visible by naked eye) to lower complexity in some way and better encode the luma.

[kolak]

Problem is not with HDe, but with its users.
They use default settings and just press buttons (settings in HDe can make huge difference in final quality). No one cares much about quality these days.
BD authoring/encoding pays little money. All titles are done in factories like Technicolor etc. All good, independent authoring houses are gone. It's all about $, nothing else anymore.
You don't really need anything better than HDe.

[asweepe]

Thanks for the replies. I am doing a 2-pass encode and this is really only happening on animation source and I just can't figure out why. I'm fairly certain that CineVision doesn't have a stuffing bits option, but I'll dig a bit further.

I've talked with Ray and Bill at Sirius Pixels several times over the last handful of years about new software, including very recently, and at nearly $18k for a license of HDe it's just not in my budget if CineVision still does the job. Kolak is correct that it's all about money. I consider my company a good, independent authoring house and we work on hundreds of DVDs and BDs every year for many medium and small distributors, but our prices have been driven down so low because distributor's profit margins are so small that clients will literally go to a different authoring house to save $100 despite the quality of the final product being inferior. We invested in upgrading to Scenarist UHD and ATEME TITAN three years ago to offer 4k authoring after a few inquiries and we've still only completed one 4k BD because it's not in most distributor's budgets.

The fact is if you're a distributor that's not a major studio or don't have design and authoring in-house then you're doing anything you can to save money and remain profitable.

I'll keep running some tests on the animation source that I have and see if I can't make CineVision do what I want it to. Thanks all!

[mp3dom]

If it's 18K, then they dropped the price a lot these years.
I don't remember CineVision very well because I used it years and years ago, but depending on the version of the software, you should have options for adaptive quantization or choose where to spend more bits (bright/dark part of the image) and so on. But this depends on the source as well... if you have a source so flat and without any small amount of grain/noise, the target bitrate could not be met by the encoder.

[asweepe]

If it's 18K, then they dropped the price a lot these years.

They certainly have dropped the price, but that's for the software only and it's still pretty expensive considering I spent less than that on TITAN three years ago to encode 4k. I do understand though, they are really the only premiere encoder left.

I don't remember CineVision very well because I used it years and years ago, but depending on the version of the software, you should have options for adaptive quantization or choose where to spend more bits (bright/dark part of the image) and so on. But this depends on the source as well... if you have a source so flat and without any small amount of grain/noise, the target bitrate could not be met by the encoder.

That has largely been my assumption. Most of the anime is large swatches of solid color with some lip flap here and there. There is very little motion or texture. I just assume that the encoder is telling me there is absolutely no need for a higher bitrate. The issue becomes the consumer. Anime fans most definitely analyze the bitrates and if the bitrate is low and the disc isn't full my clients will get a phone call asking why it wasn't encoded at a higher bitrate if there is still room on the disc. Then I have to try to explain to them why a higher bitrate isn't necessary. It's a tough spot to be in because most of them think higher bitrate always equals better quality and in some situations that is just not the case.

[kolak]

If this is for HD only then this is still a rip-off in today's reality.
CineVision is just Mainconpcet engine (full version with adaptive quantisation). It never was well working product (not just quality but usability etc.).

[benwaggoner]

If you're looking to inflate bitrate (which makes my soul hurt to contemplate) You could try turning off codec features that add efficiency. Using CAVLC instead of CABAC, reducing consecutive B-frames, that sort of thing. Or heck, use VC-1 if you have that option.

I'd argue that detuning compression to impress people who don't know about compression trying to judge quality with a ">" instead of eyes seems rather silly.

[mp3dom]

Originally Posted by asweepe

Most of the anime is large swatches of solid color with some lip flap here and there. There is very little motion or texture. I just assume that the encoder is telling me there is absolutely no need for a higher bitrate.

That's not always true. It's true that there's lower complexity with solid colors but it's also true that recent anime have lots of digital fx and gradients, plus the contrast between solid lines and flat colors make this a "high frequency" zone that needs higher bitrate to avoid mosquito noise (jpeg-like effect). I would suggest you to analyze your file to see if effectively the encoder is maxing out the bitrate distribution (but this require additional software that analyze macroblocks' qp, framesize, motion vectors and so on... but I guess it doesn't make too much sense for this particular issue)

Then I have to try to explain to them why a higher bitrate isn't necessary. It's a tough spot to be in because most of them think higher bitrate always equals better quality and in some situations that is just not the case.

True, but in that case using stuffing bits it's no good either. It's quite easy to see the use of stuffing bits so you'll get a call as well and they'll ask you why you have used stuffing bits rather than real bits
My suggestion, in this case (which cost 0$) is to give x264 a chance not for better encoding (at least, not only that) but to see if x264 too will encode your source at a lower bitrate. If that's the case, then you can guess your source is "too easy" to compress.

[asweepe]

Originally Posted by benwaggoner

If you're looking to inflate bitrate (which makes my soul hurt to contemplate) You could try turning off codec features that add efficiency. Using CAVLC instead of CABAC, reducing consecutive B-frames, that sort of thing. Or heck, use VC-1 if you have that option.

I'd argue that detuning compression to impress people who don't know about compression trying to judge quality with a ">" instead of eyes seems rather silly.

Thanks for the tips and tricks. I completely agree with you about inflating the bitrate, but we have had to explain over the years why the bitrate wasn't higher because of consumer complaints. Usually it's not higher to accommodate multiple tracks of uncompressed audio, but sometimes it's for reasons like my original post. Some consumers are just smart enough to think higher bitrate equals better quality in all instances. They don't always realize things like banding are often in the source and won't be remedied with a higher bitrate.

[asweepe]

Originally Posted by mp3dom

That's not always true. It's true that there's lower complexity with solid colors but it's also true that recent anime have lots of digital fx and gradients, plus the contrast between solid lines and flat colors make this a "high frequency" zone that needs higher bitrate to avoid mosquito noise (jpeg-like effect).

True. We do work on titles that have more complexity. In those cases the overall average is brought down by the simple areas of the video, but I don't concern myself too much with those titles as the areas that need the extra bits usually get them.

True, but in that case using stuffing bits it's no good either. It's quite easy to see the use of stuffing bits so you'll get a call as well and they'll ask you why you have used stuffing bits rather than real bits
My suggestion, in this case (which cost 0$) is to give x264 a chance not for better encoding (at least, not only that) but to see if x264 too will encode your source at a lower bitrate. If that's the case, then you can guess your source is "too easy" to compress.

Also very true. Damned if you do, damned if you don't in that situation. I may run some tests with x264 and see where I'm at. Thanks!

[Emulgator]

I would always suggest to add noise/grain, but then I guess I'm thrown out by Anime/CGI fans ;-)
...or give them reason to proudly denoise their full-bitrate purchase ?