DVD Encoding
 

I have one question thats been buging for a while. Once you have created a Cineform.AVI from your project..lets use 720x480 30fps NTSC/Wide Progressive for example. And you take that file into Encore to create your DVD what transcode setting should be used to output the best resaults. Will Encore degrade the cineform file during DVD encoding sine Cinefrom does not do the final transcoding when authoring?

And does the same apply when authoring Blu-ray discs?
I guess I do not understand if we are going to go thru all this work to have Encore or what ever destroy the High Quality CF file?

You always have to lose something when going down to the distribution data rates of DVD and BluRay, yet you are getting the best results by starting with the highest quality master file.

Bruce, many of us use CF for "Mastering", since we have multiple target types for distribution. Usually these are DVD, BR, or Flash trailers. Since we live in an age of rapidly changing distribution options, the high final quality CF "Masters" allow us to downres to the highest quality that particular medium is capable of.

As such, it appears that in the near future, the "full" 1920x1080 res "Masters" could be used directly for "downloadable HD distribution". So, I guess the bottom line for many, is having the highest possible quality masters at the best "price/benefit/editing performance" point possible.

Cinefom is on the cutting edge for that scenerio.

Thanks Dave/Stephen
Question #1 if I deinterlace when I capture 1440x108060i footage Do I need to keep selecting Deinterlace for all my outputs or is my footage deinterlaced when captured forever if I select that option.

Ok I undertand the HQ master.

#2 If you have PPCS4 Suite and you have captured all your video using CF, then In theory if you do not always need a master CF.avi. will you get the same quality DVD output using the dynamiclink and taking it into Encore right away for final transcode?

#3 What transcod setting do you feel give you the best out put for DVD and BD in Encore? if you have the highest quality to start with do you feel CBR or VBR is better and do you feel set the quality slider to the max?

Stephen I sent you an email asking if you would share how you compressed your flash video for your site, I was very impressed with the quality and speed of how fast it loaded over the net and how smooth it played.

Thanks

Bruce,

I am not expert for Cineform and I don't have Encore, but I can say that transcodings from Cineform to Blu Ray or DVD are for most people very important if not the most impotrant stages in whole proces. I think this stage is little underestimated on this forum, but you have many other forums about mpeg encoding and encoders.

As long as you are in Cineform (or other similar codec) you don't lose anything visible. But, when you come back to mpeg2 you must lose something and you can lose all benefits you got with Cineform workflow if you don't do it well or too sparingly - so whole Cineform process don't make sense.

So, you shold do transcodings back to mpeg2 with best encoder you can get and with best settings it has. And you shouldn't think about rendering times, but much more on quality.

VBR vs CBR - if space on disc is not your problem, than for example VBR 20 - 25 -30 can't be better than CBR 30 mbps, but VBR 25 - 30 - 33 can be better than CBR 30 mbps. Who will know - it all depends on your video, but for ordinary video with all given examples you will hardly notice any difference on TV (if source was from HDV).

Bitrate - some people say, and I think that it makes sense, that it is good to set higher bitrate for Blu Ray than it was in your original video before you transcoded it to Cineform (for example, if you had HDV with CBR 25 mbps, it wold maybe be good idea to set something like VBR 25 - 28 - 33 mbps). Final video with higher bitrate never can't be better than original, but with higher bitrate you give more space and chance to encoder to successfully estimate I, P and B pictures.

Bruce, apparently the email function is not working here. I also sent myself an email to test and haven't received it either, sorry.

As Ivan stated, your final encoder is certainly important, but it seems to us, that generally speaking, we gain better results using CBR at the highest practical settings for that particular format (and most others). That might not be true if you have very high motion sequences, though.

There are some qualifiers here.

One, is that we bump everything up to square pixel (1920x1080p from HDV with HDLink) and edit everything as progressive in that size.

For downresing to Flash, we've found the Abode CS3 Flash encoder works well with square pixel material done in High Quality (700kbps) using the On2 VP6 codec. The trailers are 480x270 and look very nice indeed and run smoothly. The new encoder in CS4 is even better and allows us to output in HD if desired, though we aren't using that size yet.

Another beast altogether is doing MPEG2's for DVD! It seems to almost be a mixture of skill, luck, black magic and good codecs to get consistently good material. With square pixel CF masters, we've found that often times, we can get better mpg's from TMPGEnc (version 4.6.3.268) by just using the standard MPEG2 Video setting, setting it to "Quality Encode mode, CBR, 10 bit DC component precision, progressive display mode, and using "High Motion search precision". Save the settings as a preset and you're good to go.

Our mpg's done that way, seem to have pretty consistent quality, though the new CS4 encoder may now rival TMPGEnc (for ouput quality). We really haven't found H.264 to make much difference at these lower resolutions, though it does at BR sizes.

For DVD, the rule seems to be, use the highest quality settings you can, without "pushing" into the range of not playing on some players (by using higher bitrates than 8Mb for video).

The truth is, DVD sucks compared to HD, but if you start from high quality masters, you'll usually end up with pretty decent DVDs.

Thanks Ivan and Stephen!!!! This is a lot to digest and good info. I need to output Both BD and DVD for clients because they have not seen the benifites of BD, But with everone buying LCD tvs if you do not give them the best single the picture will look crappy!!

Stephen I was asking how you created such awsome flash files for your website?? They load so fast abd run so smooth and are excepent quality. If you want to take this off line you can email me at bagrubes@hotmail.com and Ivan I would alos like to hear from you..

When speaking about Blu Ray and bitrate, I also think that stretching 1440 to 1920 is not a bad idea, and on one forum I have read this about stretching, which I think makes sense (it is not my words but directlly interpretation from one senior member from this forum):

”The effects of a 1440 to 1920 stretch are small and wouldn't be noticed on a consumer camcorder. Bit rate is more important. Recording at 1440x1080 saves 25% of the pixels and 25% of the bitrate. To get the same MPeg2 quality at 1920x1080, one would need to record at 33.3Mb/s.

If you stretch HDV to 1920x1080i for Blu-Ray, you should compensate with 33 to 35 Mb/s bit rate to maintain quality.”

If you calculate a little , you will find that 1920/1440 = 1,33, and 25x1,33= 33,3.


When speaking about DVD bitrate, I never noticed problems with video bitrates of CBR 9200-9400 kbps in last 10 years on very different DVD players (cheap and good ones). I think we can speak about 20-30 different DVD players here.

I gave my DVDs to my family and friends and nobody never complained about problems with playing of such a DVD video. I know that 8000-8500 kbps is a practical limit for SD mpeg2 and I lose a little space on DVD, but I always make it with higher bitrates for me, just for safety and not to think about that any more.

Only once I noticed problems with bitrates of CBR 9800 kbps, but only on one and very cheap and old DVD player (not mine).

But, if you are working DVDs for different clients, you should maybe be more carefull and do it with bitrates under the 9000 kbps (just for safety).

If your clients have LCDs, you must do your video in 16:9. It’s a little joke, but sugest them to buy plasma instead of LCD (because in most cases SD looks better on plasma). You shuldn’t be scared that DVDs will look crappy if you make them like Stephen suggested – I tried something like this and they look wonderfull.

I don’t know why they make DVDs progressive (I didn’t try yet), but I tried interlaced SD from interlaced HDV and it also looks great – even visible better than from my old and very good SD cam. Picture with such a DVD on plasma is somewhat cleaner with less artifacts and very sharp. But sharpness is not so important – you can always sharpen in post in your NLE or on your TV if you want. But one thing is important – when you scale down (from 1920 or 1440 to SD) you should scale with HDLink engine, not with your NLE engine).

And some additional questions for Stephen about Blu Ray and Tmpgenc (or Vegas):

- is it good idea to use CBR fro Blu ray mpeg2 ?
- what is the highest practical and proposed bitrate when you make 1920x1080 Blu Ray
from the HDV source ?
- do you use 10 bit DC component precision for Blu Ray mpeg2 and with which bitrates (I
noticed that in most standard made templates for Blu Ray DC component precision is
set on 9, and I know that it is good to put it on 10 if your bitrate is bigger (for example
more than 8000 for DVD) –and I don’t know this bitrate “border” for Blu Ray )?

Thnaks Ivan! this is very useful informaion! I am still learning about the wole pixel thing>
Where is the stetting for 1920x1080 square pixel when Importing using HDlink? Or is that native when you select 1920x1080? and do not select de-interlaced?

One more question. You say you should start with the best possible master. Should I use a Full res CF.avi then take that to Encore or what ever or should I use CF to compress to 720x480 first then take that to Encore or what ever?

In HDLink, when you select 1920x1080 – it is native to get square pixels.
If you select none (for HDV video) you will get 1440x1080, PAR 1,33 (and like your original HDV video, non square pixels).

Selecting “deinterlacing” in HDLink have nothing with pixels and PAR - if you select this - your output from HDLink will be progressive. If you don’t your output will be interlaced like your original.

Some propose to do that (select deinterlace) – I don’t do that because in most cases I don’t like look of my interlaced video when I make progressive video from it (but this is only me, and the way in which I shoot the video).

When you read my previous posts about DVDs – everything is certainly right and tested by me and I do like that for years without problems.

When speaking about HD, I am still new here and I am still learning (although, mpeg2 is mpeg2 and there should be not so much difference between SD and HD regarding best encoder settings). But some differences must be.

For example, thinkings from others like “If you stretch HDV to 1920x1080i for Blu-Ray, you should compensate with 33 to 35 Mb/s bit rate to maintain quality.” - I didn’t prove yet by myself and I am not sure is it the truth or not and maybe you waste to much space on Blu Ray disc because of that (if your source is 25 mbps HDV) – so I put some additional questions to much more expert users like Stephen (or others).

But, you can see also from my previous posts that I don’t care too much about rendering times or space on disc – I would do everything to get the best possible quality – and I always want to have some “reserve” in that sense. If you get more “reserve” (higher bitrates, best settings but higher rendering times , more discs) than you shouldn’t care too much if your encoder is the best or not, or does somebody give you good advice or not – they are all good when you go above some borders.

In past, I read so much about which mpeg2 encoder is the best (Tmpgenc, Vegas with its Mainconcept, standalone Mainconcept and many others) but I never get real and consistently answer. It seems that people measure quality of encoder in terms of speed, and what can it do when you go with low settings and low bitrates – but most people agree that there is not so much difference between all of them when you go with high settings (apart from speed).

Bruce, since this is an open "real names" forum and we can learn from each other, I'll try to answer your questions here where I might have some experience or knowledge (however faulty that may be). If I'm all wet with something, someone with better knowledge is free to correct me!

Ivan, I'm only as much of an expert as most of the rest of the people on this forum! Like you and almost all others, I've spent many hours trying to wade through the morass of the MPEG2 world and burned my share of useless platters.

One thing I believe is true about HDV, is that when you are upsizing to square pixel, you are not really "stretching" anything. HDV is "compressed" to 1440x1080i and uses Long-GOP MPEG2 (15 frames long) to fit on the little DV cassettes. By using CF, we are merely "restoring" the video back to it's original, normal aspect-ratio, which is at the square pixel CMOS chip capture size. CF also converts it to a format which is much easier to edit and does not use MPEG2's method of compression (thus not passing on block errors to each successive generation).

We don't need to go into detail on that, as there are many, many threads on this forum with specific info on the processes involved.

All that is to say the following: The ideal would be to capture our video right from the chips, before it is compressed and information is thrown out)! Since most people do not have an easy way to do that (unless you can afford the cams or equipment that CAN do that...), we try to preserve our HDV material in the most cost/benefit ways possible. CF has given us one way where we can take our CF-compressed material (at +100Mbp/s decompressed!), and output it at the highest rates available to us with each distribution format. At the highest Bluray rates (usually 35-40 Mbps), our HDV material done this way will still look very good. It is only when we start to compare it side by side to true uncompressed HD on large screens that we start seeing the weaknesses.

So to answer Ivan, the rule seems to be just use the highest possible rates on whatever format you are outputting to. Since BR goes back to MPEG2, the same problems we have with DV are still there...

As to interlacing, ALL plasmas and LCD TV's are inherently progressive, as are ALL modern computer monitors. Interlace has to be converted to progressive before being displayed on them. Why not just start with progressive and not have to depend on the conversion quality of the target equipment? Sure, you'll find some interlaced TV's still in use, but the future is PROGRESSIVE. Sometimes high motion scenes can be a problem with progressive, but if your video is not normally high motion (like sports, flashing lights, fast camera moves), it seems to me progressive is best. Although HDV recorded onto tape is usually recorded as interlaced (at least with Sony) no matter what setting you use, CF can change it to progressive quite nicely at the beginning of the edit path.

Ivan, we are still only "testing" BR output, as until there are some changes in the BR scene, we are not going that way unless we have to. Our target audiences are worldwide, so we're still sitting on the fence as to our final HD distribution methods. I personally do not like BR for many reasons (even if it does allow us to show off our material better). For small producers and distribution, it sucks bigtime, so we are still waiting for a "better solution" to appear. My bets are on "direct download", but if things take too much longer for that to truly get here, we'll have to do the BR thing too. In that case, YOU'LL be answering MY questions!

Things are changing so quickly these days, we're glad we used CF's intermediate codec for our workflow. Most of the newer compression schemes still have major editing probs, so CF still claims most of the higher ground, and I can't see that changing much in the near future. Even the new 3D scene has been covered by CF, and they are the only real show in town (for standard PC editing) in those formats.

Ivan, lead the way! You're the BR man, so we'll let you burn the platters for us... :) and answer Bruce's questions!

Sorry Bruce, got long-winded and forgot your question.

For DVD's (or BR, I would assume), just remember that Encore is really only just "packaging your video" to watch. Though it will compress it for you if needed, it always seems much better to me to not trust it to do that, and compress it yourself first so you can check it out. Since TMPGEnc, or the new CS4 encoders (you can't use CF for that final compression, as you're going back to MPEG2 for it) are very fast with their compression/downsizing, you can actually gain time that way. The Encore burns are very fast then, since it doesn't have to compress video for you.

(Note: If decoders on TV's, DVD/BR players and other equipment could decode CF, we'd just have to "downsize" our CF masters for DVDs.......but they can't.)

No this is a good discussion and I wish more people would jump in on it I was hoping to hear for Dave and the Cf poeple also but I know they are working on getting CF up and running for Cs3 and cs4.

It is so hard to find information all in one spot that contains this kind of information.

Ivan Quote(Some propose to do that (select deinterlace) – I don’t do that because in most cases I don’t like look of my interlaced video when I make progressive video from it (but this is only me, and the way in which I shoot the video).


Ivan at what point do you make the Progressive transformation in you work flow?

Also I have a short clip I made slow motion with PP cutting the speed by half I tried De Int. and Progres. And I can not get it to look good and smooth?

First, thanks to Stephen for very informative post. Although we don't think in the same way, he answered my questions.

To Bruce:

It seems you are more interested in SD than HD. I have read whole forum (I am new here) and from HDV to Sabout downresizing D there are basically two different options here:

1) To start with best possible master (Full HD) and then make everything from that. Most people from forum and people from Cineform support that option (and Stephen I think)

2) To directly compress from 1440x1080 to SD and work with that. People from Cineform didn't say anything against that, and only some people here claims that this way they have sharper SD video than with option 1)

Because of two different approaches to the same problem, I tried both on few 1-2 minutes examples and couldn't find any significant difference between the two. But my only tools were my eyes, histograms in Vegas and my BR/TV setup. Both methods are giving very nice and similar looking results to my eyes.

Option 1) is more practical if you want the same looking video for all your formats (same transitions, text, etc), and well, it is supported by people from Cinefrom. It is much more time consuming if you want only DVD than option 2), but it gives to you only one master for all formats (BR, DVD, and all other).

For all your rescalings (up or down) you must be sure that are done with HDLink rescaling engine (not with Vegas or Encore or what ever).

Option 2) is practical only if you want only SD and under that. It is much faster if you want only DVD and nothing more. But, for me it is hard to find such a situation where you want only SD, when you have HD cam.

So, when you think about both options, it seems that option 1) is winner.


About progressive vs interlaced. I have two cams, one is SD, and other is HDV. Both have PAL 25p progressive mode.

If I shoot in interlaced I don't do progressive transformation in my work. Plasma, DVD player or PS3 are doing that for me - and if video is properly filmed it looks very good with very rare interlaced flickers . I mainly shoot interlaced 50i. Very rare I shoot progressive 25p.

I do it that way because I didn't like the look of my progressive 25p video. It was sharp and without flicker, but very often, I could see visible strobe motion with it. I am more sensitive on strobing than on rare flickering in video. And I believe this is the case with most home made videos (you know, everything is not planed, children are running around, weddings, parties and so). For documentary or some other works, progressive is I believe better.

I tried to deinterlace my HDV video with Vegas. It has two options (blend fields and interpolate fields). With blend fields video is not sharp, with interpolate fields video is sharp but strobbing even more than when I filmed directly with 25p.

The true is, I didn't try to deinterlace with Cineform yet. Maybe situation with Cineform deinterlacing would be different. There are some good threads about Cineform deinterlacing on this forum where people from Cineform are explaining what method does Cinefrom use to deinterlace.