I also have been using very high bitrates and not yet had a problem... but all this talk is making me a little nervous. I think I've been doing 9800/8000/192 kbits for max/avg/min. I've got lots of motion in the video(pans and the like) and I'm trying to prevent the artificting from being visible. Perhaps I should change my strategy a little....

-Jayson

In DVDA 2 No. But why encode in DVDA?
There's every imaginable tweak for the same encoder if you encode out of Vegas.

the issue you see is that burned dvd's (vs. commercial dvd's) often have playing problems at high bit rate ... Canopus and others will tell you the same thing ..

Constant bit rate means constant into the vbv buffer - but the actual bitrate may vary a little (in some encoders you can see big spikes even)

In this thread DVD media test results, I quoted one of the few good studies ever done on DVD media compatibility (i.e., how many set-top DVD players will play successfully a particular brand of DVD media). The author, Ralph LaBarge, claims that one of the things that improves compatibility is to lower the bitrate. His reasoning is that the reflectivity of DVD-R and DVD+R is less than what is achieved through the conventional mass duplication process, and therefore the player must correct more "missed" bits. If the bitrate is high -- so the theory goes -- and the error rate is also high, the player might not be able to keep up, and the disc may not play, or play with skips.

I have never been fully comfortable with this theory, primarily because I don't see why the player would have a tough time "keeping up" with the errors (error correction is a virtually instantaneous computation). Therefore, if the player CAN keep up, it would seem to me that the bitrate wouldn't matter at all.

Lowering the bitrate will fit more minutes onto a one-hour single-sided, single layer DVD. Given that I don't believe the above-stated compatibility theory, fitting more minutes onto a DVD is the only reason I know why you would intentionally reduce the bitrate below 8,000 kbs.

FWIW, the DVD spec (mpeg.org) requires that all DVD player should be able to play a continuous video stream of up to 9,800 kbs. Note that this link is to the MPEG.ORG site.

No matter how good the encoder you need to consider how a DVD stream is played out. High bitrates means the player is working flat out, no spare time for error correction. All optical media is full of errors, as the disks age or it gets played on crappy players the risk of noticable errors during playback increases. Layer switching (you'll need to deal with that for DL DVDs) is another area where things can go wrong. Before and after the layer break you should throttle the bitrate back to give more space in the read buffers. Hopefully that'll prevent a noticable pause or lockup of the player.

What'll further confuse the issue is something I've seen a lot of. Higher quality encodes at the maximum bitrate look woeful on poor players and big prices or brand names don't guarantee quality. I've watched the same DVD on a brand name player and my reference machine. The difference is staggering. For all the world you'd swear the encoder was a heap of junk yet it was done on a hardware encoder fed 4:2:2 via SDI.

So if you're aiming for the highest possible quality outcome there are many factors to consider. Preparation of the source material for encoding is one, the capabilities of the encoder is another and the quality of the media along with the player also needs consideration.

I'd also suggest that DV25 compression and mpeg-2 encoding are perhaps not the best of friends. I don't really have the technical knowledge to back this up but I'm suspecting the way a lot of cameras seem to introduce a certain 'edginess' to the video is counter to what encodes well. SP footage which is a little softer than what you get from most DV cameras even after DV25 encoding seems to convert to mpeg-2 beautifully, this kind of ties in with the results the big end of town can get coming off 35mm at very low bitrates. I'm just extrapolating here, could be way off the mark.

Bob.

quote: "FWIW, the DVD spec (mpeg.org) requires that all DVD player should be able to play a continuous video stream of up to 9,800 kbs. ."

this is ONLY for commercial stamped/pressed DVD's ......NOT burned dvd's ..

let me assure you after making 40-50 thousand burned DVDs and authoring 1000's .. i can tell you pretty well what does and doesn't work ..

anyone in the duplication business with experience will agree with me ....

max bitrate depends on the media, the encoder used , the burner and the authoring app...

take a detailed look (per frame) of bitrates done CBR and you will be surprised -- they are not constant ...

some other good reasons are also posted aobve ..

quote: "DV25 compression and mpeg-2 encoding are perhaps not the best of friends"

true -- for dvd's

DVD's require frame mode encoding but DV is field based -- if you encode DV field mode it will (often) look better -- but will not play on many many players..

also the matrix used for mpeg encoding is very often for progressive sources , not interlaced ...

couple other reasons also -

not to say you can not get really encodes from DV, because you can -- but there are a couple factors getting it from being 100% perfect ..

btw -- myself and others maintain the largest database anywhere of dvd media compatibility ... as well as lengthy tests and studies on the subject ..

this is at www.videohelp.com (where i also am one of the moderators)

btw -- myself and others maintain the largest database anywhere of dvd media compatibility ... as well as lengthy tests and studies on the subject .. this is at www.videohelp.com (where i also am one of the moderators)

Couldn't find www.videohelp.com (no such site), but Google refers to vcdhelp.com and the newer URL for that site, dvdrhelp.com. I am very familiar with this site. It is an amazing and excellent resource for all sorts of things. Unfortunately, the database of DVD media presented there is based on user's self-reported experiences. As you point out in your post above, "max bitrate depends on the media, the encoder used , the burner and the authoring app...". The problem is that most of the posters don't have any way to determine which of these things caused the problem, and I suspect that many problems may have resulted from something other than the media. However, while this makes it difficult to figure out what the best media might be, if a certain media causes problems for an abnormally large number of people, you can definitely use the dvdrhelp.com database to spot the really awful stuff.

As for this: "the DVD spec (mpeg.org) requires that all DVD player should be able to play a continuous video stream of up to 9,800 kbs.," you responded:

"this is ONLY for commercial stamped/pressed DVD's ......NOT burned dvd's."

Why is that? The requirement itself has to do with the circuitry, which is designed to handle a certain number of bits per second. That circuitry has no way to distinguish between bits from one type of media and another. Bits is bits. Thus, the circuitry, if designed to be in conformance with the spec, will handle 9,800 kbs. Of course, because of the lower reflectivity of burned media compared to commercially pressed media, the error rates with most DVD players will be higher when playing DVD-R or DVD+R, and this can lead to various problems, including inability to play at all.

As for farss' contention that: "High bitrates means the player is working flat out, no spare time for error correction.", error correction is a trivial task, requiring virtually no compute time whatsoever. It has been a long time since my EE days, but I remember enough to know that the error correction probably takes less than a fraction of one percent of the decoder's horsepower. In fact, now that I am starting to remember the math and how you implement the decoding, the error correction takes place ALL the time, whether there are errors or not, and there is probably no increase in CPU load as the error rate increases.

Also, the rest of the circuitry actually has less to keep it busy when it is dealing with a lower bitrate. Think about it: The DVD can hold a fixed number of bits (4.48 Gbytes, using one measure, or 4.7 Gybtes, using a slightly different measure). If you encode at a really high bitrate, and fill up the disk, then all those bits must be read out in one hour of playing time. If you encode at a really low bitrate, then those same bits must be read out over two hours or more. Far fewer bits per second.

I have searched many times for a scientifically valid test, that holds all variables constant except bitrate, and which then demonstrates that a disc containing the same material at a lower bitrate played, whereas a disk containing higher bitrate material didn't.

I suppose one could take a few minutes of video, encode it at several different bitrates, and then put the resulting files on a single DVD. If the clips were short enough, the slightly different positions on the disc probably wouldn't matter much. If the bitrate theory is correct, the low bitrate files should play, and the high bitrate ones would not.

I have not yet seen the results of such a test published anywhere, so I continue to maintain my skepticism.

http://www.videohelp.com/ same as VCDHelp, SVCDHelp, DVDrHelp, and others ...


all the same site ..... :)

just updated php - so it was down for a few minutes ..

just updated php - so it was down for a few minutes

Ah, now videohelp.com works. Like I said before, GREAT site.

quote: "That circuitry has no way to distinguish between bits from one type of media and another. Bits is bits."

totally true - but the laser DOES differentiate and so does the circuitry .. the way the disk is read is different for dye based media and non dye based ...
All disks carry this information in a ROM part of the disk...

a rather crude case is svcd's .. they have a much lower bitrate than dvd's , and the dvd player should be able to play full bitrate svcd (some actually do) based on the circuity is capable of it .. but they don't ..

I think many are missing my main point here , which i brought up twice already .. is that CBR is not alway so constant and that it can spike over even the max allowed data rate ..

an allowance should be made for this .. and as well for the fact of the media/burner/authoring combination used and allow a little leeway .....

you can run it right up to the max if you want -- just like you can shift at redline every time also ... but if you plan on distribution of a project on disk and want max compatibility -- well help yourself out ..

though sometimes - no mater what -- some burned disks just do not play in some players (another example of circuitry/laser differentiating between burned and non burned disks, a non burned disk MUST always play even at max bitrate (but not over), a burned disk doesn't even have to play AT ALL in a dvd player for that dvd player to still carry the official tested dvd logo).

John,
you make some good points there and I don't know enough about the internals of the decoder chips to comment however I've seen real world tests that do seem to show marked differences with the same files burnt onto different media, I've also seen that burning at a higher speed on the same media cured a skipping problem on playback with a cheap player.

However your assumption that the player has twice as much time to read 2 hours of video maybe a bit flawed. From what I can see the player doesn't spin the disk at half the speed. Rather it reads data in blocks to keep the buffer full. It still has to output video at the same rate but the number of blocks it needs to read per second will drop at lower bitrate as one block decodes to more video. This should mean if it gets a bad read it may have enough time to take another go at getting a good read. But I'd certainly think it reads each data block at the same bits/sec.

Looking at it another way, with lower bitrates then the buffer can hold larger chunks of video for output. What I see happening at lower bit rates is intermittent read access to the media, I cannot measure the data rate off the laser but I'm assuming regardless of bitrate the read data rate is constant.

Bob,

btw -- tmpgenc is famous for it's spiking over the max bit rate ..

"I'd be interested in finding out what tweaks you use with the Mainconcept encoder."

Probably the tweaks I am using should more rightly be called selections and you probably already use these:

Search Method set at 11, it can go higher but I have read several times that higher than 11 really doesnt improve things.

Scan Range - A high(er) number here gives much better quality.

Half Pel Search - making this selection is supposed to help.

I agree with others who said the ability to play high-bitrate DVD's varies from player to player. I have a PIoneer DVD333 that plays them like a champ, but my little portable Panasonic LV50 plays them for about 6 minutes and then starts to protect itself by skipping on audio and video.

If I am doing a PCM video, what's the MAX video bitrate I can use to stay in DVD spec (eg: 9.8MBPS) ?

Your example, 9.8MBPS is a very high bitrate, in my opinion.
If you have a good quality source, anything above 8mbps is probably just wasted bitrate and might even hurt quality.

For maximum compatability with players your DVD might encounter in the future, I would suggest VBR with a max of 7.5mbps (6.75mbps is even better) and minimum of 3mbps. If the video source is poor quality you might need to go with CBR of 8.5mbps and take the chance that some players will object. I'm sure others will have different thoughts because that's just how varied thoughts on mpeg encoding really are.

2Road's recommendations on bitrates are echoed by many other people "in the know" that I have read. My own experience seems to bear this out, namely that you should definitely stay below 8,000 kbs for the video encode, and that for all but the most pristine sources, you probably won't see much difference in quality between 8,000 and 7,500.

He also makes a good point about the header information encoded into the DVD+-R disk that tells the player that it needs to do something different because the disk is physically different than a pressed disk.

A lot of people are under the impression that the highest bitrate they can get on their DVD-R disc, the better. I'm actually of the school that compatibility beats quality any time. I mean, what's the point of giving someone an amazing looking DVD disc, if they can't play it without stuttering.

So, my feeling is, conduct your own tests, and choose the encoder that most pleases your eyes and works with your particular workflow. Example, Mainconcepts is extremely fast, if you need a fast turn around -- then by all means use it.

I've basically standardized to Canopus Procoder at an average 5,500-6,000 kbps using "mastering quality". My source video is always of high quality, if I worked with inferior video, I might need to adjust these settings (again, CCE is very close, but I like the fact I can use filters with Procoder, can't do that with CCE).

The other thing, I always follow this compatibilty checklist :

STEPS TO INSURE HIGHEST LEVEL OF DVD COMPATIBILITY
====================================================

- high quality DVD recorder such as Pionee, Plextor, Sony

- high quality DVD-R media (Maxell, Fuji)

- high quality MPEG encoder (Procoder, CCE)

- 5,500-6,000 kbps (lower bitrates mean lower error tracking problems on home DVD players)

- use of high quality DVD authoring programs, some that get high praise for producing DVD"s that have high rates of compatibility are: Sonic's MyDVD/DVDit, Ulead DVD Workshop, ReelDVD)

- use standard authoring templates and not exotic GOB options

- always encode audio AC3 (dolby digital), easiest way is by buying an authoring program that has this capability

- use a top notch burning program to burn to DVD, Nero is an example

- always include AUDIO_TS with VIDEO_TS folders, do not add extra data files

- don't apply labels to DVD discs

- use proper DVD cases, not CD-ROM cases

- you can do to improve playback compatibility by avoiding filling the disc. The outer edge of a recordable DVD disc is more susceptible to data errors, so if you only use about 90% or less of the total disc capacity you will get better overall results.

- Some independent tests have reported better compatibility results with discs that were burned at 1X speeds. I have never seen a test that focused solely on playback compatibility versus burn speed so I am a little skeptical about this connection. In my own testing I have seen a higher number of compatibility issues with discs burned at 4X and 8X speeds.

---------------------------------

Ralph LaBarge, Author, "DVD Authoring & Production" added to my list and also stated "There is no reason to recommend DVD-R over DVD+R media as long as you use a "name-brand" of either type. There are some players that prefer DVD+R over DVD-R, while there are other players that prefer DVD-R over DVD+R. I recommend that authors have a multi-format drive so that they can burn both DVD-R and DVD+R. Then if a particular player is having problems you can always send that user a disc made using the other format".

I don't know the validity of this comment, I assume 2-3 year old DVD players would have a better chance of playing DVD-R then DVD+R.

Another trick is, turn validation on in your DVD burning software (example, Nero) and as well, as an added check, when finished burning, drag and drop your DVD-R disc to your PC's hard drive. If successful (without errors) then you are sure to have created the most compatible disc possible.

Again, there are two issues -- compatibility and quality, you kind of have to find a happy medium between the two. But if you take your time and keep in mind the checklist in this post, you're going to produce a DVD disc that will play 90-95% of the time, no matter what DVD player .

Darn, hope this stuff is useful, cause it takes a lot of time to write and takes up a lot of bandwidth on this forum.

MB

Mbelli,

Thanks for that fantastic test results post. It was very helpful.
I was wondering:
Have you done any comparative tests for Mpeg 1 or am I misreading your earlier post?
I never seem to get good results with Vegas MC. I have heard people on this forum recommend TmpGenc for Mpeg 1, but I was wondering if you have any insight?

John

Maxell disks means nothing as they could be made by anyone --

unfortunately Fuji's now doing the same thing -- and now we have been getting reports of orders from newegg and others having two different media codes in the same spindles ...

"name brand" has no meaning what so ever... use the media codes only to see who made those blanks ..

adding extra data files is no problem at all -- in fact it is recommended for DVDs under 1 gig and if you have some malita players , a 2gig min is required on the disk . i pad these disks out with invisible data files in a invisible name directory .. nero does the same thing when you select "dvd compatibility" in its options and so does prassi (still the best burning app IMO) ...

not using labels and not filling the disk is excellent advice ..

playing a dvd on a pc to see if it works is fairly useless though as software players will play dvd's way way out of spec (they don't care about VBV buffers at all, or bitrate or GOP structure) ..


and no mater what -- some players will just not play ANY burned disks or most of them (BOSE is one example, some in car players also are very picky, there many many more. price has no correlation (in fact the cheaper - sometimes the more compatible)).

>Maxell disks means nothing as they could be made by anyone --

The point is not to buy the 99 cent Walmart no-brand specials. If you have success with a certain brand, ID it with software like DVD Identifier (http://dvd.identifier.cdfreaks.com). I would presume Maxel or Verbatim, if they put their name on a disc even if not made by them, insure a certain level of quality above generic no names brands.

>Adding extra data files is no problem at all -- in fact it is recommended for DVDs under 1 gig >and if you have some malita players , a 2gig min is required on the disk . i pad these disks >out with invisible data files in a invisible name directory .. nero does the same thing when >you select "dvd compatibility" in its options and so does prassi (still the best burning app >IMO) ...

Good to know, I got Prassi with my A03, can't say enough good things about it, great software.

>playing a dvd on a pc to see if it works is fairly useless though as software players will play >dvd's way way out of spec (they don't care about VBV buffers at all, or bitrate or GOP >structure) ..

Misunderstood. I recommend this link, http://www.customflix.com and click and get their guide called "DVD Authoring Nightmares".

It points out the following:

------------------
Check data integrity.

If you have a disc that you think may be damaged, a good way to check to see if the data is intact is to put it into a computer and copy the entire disc to the hard drive. Your computer requires each bit of data to be readable and verified during this copy process, so if there is a physical glitch in the DVD, this will make the computer stop and tell you that it couldn't copy the disc. Burn a new disc and toss the damaged one. If you are replicating your DVD, make sure to request a check disc before the final run is produced.
-----------------------

That's what I meant, a good thing to do if you skipped data verification on burning and you might not get a chance to view the entire disc through a DVD player.

In regards to VCD, I've given up cause on DVD playback through a good TV set, it always looks awful, on a PC it's not bad, Canopus Procoder seems to do the best job on this as well.

I use SVCD which is amazing and have had great results with a program called DVD2SVCD. However, compatibility is a big issue with SVCD.

Guess it all depends on what you need. VCD is good for quick dirty playback for clients to judge content and MPEG1 is great for PC's because all operating systems/Windows media players can play it.




MB