1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Problems w/MPEG2 encoding. Bad interlacing artifacts...

Discussion in 'MPEG-1 and MPEG-2 encoding (AVI to DVD)' started by nitelife, Feb 14, 2003.

  1. Minion

    Minion Senior member

    Joined:
    Sep 28, 2003
    Messages:
    5,623
    Likes Received:
    0
    Trophy Points:
    116
    What are you useing to encode your DV AVI files to Mpeg2?? The encoder has a very Big impact on the Quality after encodeing to Mpeg2...There are only 4 encoders that I believe are even worth useing, They are Tmpgenc ,Mainconcept Encoder ,Canopus Procoder and CinemaCraft encoder ,Most all others are pretty much Crap, and the settings used in the encoder are very important also and useing a High enough bitrate to Achieve the Quality you want, and Progressive and Interlaced Video needs different Scanning Patterns when encodeing to Mpeg2 and selecting the wrong one can really degrade the Quality...I use DV Source Files Occationally and Never had a problem with Quality after encodeing so I wonder whay you are haveing Problems...well i hope you get it worked out...Cheers
     
  2. BJKESLER

    BJKESLER Guest

    The only encoder that is affordable is TMPGENC and its not perfect regardless of the claims of some. TMPGENC can be tweaked but the standard settings are as crappy as the others if not worse. Nero probably has the best out the box no tweaking required encoder producing the best result. None of the affordable ones have reach Digital 8 tape quality yet.
     
  3. Minion

    Minion Senior member

    Joined:
    Sep 28, 2003
    Messages:
    5,623
    Likes Received:
    0
    Trophy Points:
    116
    I would not say that at all..Nero is well Known in the Video Comunity as haveing probably the Absolute Worst Quality encoder ever created, It is definately a Newbie encoder but even a Newbie should notice the Absolutly awefull Quality of the Mpeg encodeing done By Nero...If you use the Standard Templates in tmpgenc but set the "Motion Search Precition" to "High Quality" then you will get very Good results ,Tmpgenc is Probably the Highest Quality Mpeg1 encoder out there But it"s Mpeg2 encodeing is Only Above average..For High Quality Mpeg2 the Best and Fastest Is Probably CCE SP 2.67 but it is also the Most Complex to use but if you Know anything about what settings do what in an encoder then it should be simple to set up no mater the encoder...Cheers
     
  4. trx

    trx Member

    Joined:
    Jan 17, 2004
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    11
    Minion, what capture program do you use to create interlaced AVI's? I've tried Roxio DVD builder (junk), DVIO and WinDV and they all seem to produce progressive AVI's. Also, depending on what TV you have these interlaced artifacts can be very noticeable. I have a 56" DLP TV in which everything is displayed in 720P so it's closer to a computer monitor than an old NTSC SDTV. Any loss in quality is noticeable on this TV. I quickly learned how bad Roxio DVD builder was at MPEG-2 compression.

    Want to see sample DV frame with interlaced artifacts and the same frame using a deinterlacing filter? Here it is: http://trx.punknet.org/gallery/album02 The first two images shows before & after shots of deinterlacing. The deinterlacer I used does not drop resolution and minimizes blur as much as it can.

    Bernie, do you have any before and after shots showing the quality you are loosing? I haven't done much encoding but I don't see much loss in quality. Yes, there will be a little since you are re-compressing video from one format to the other and at a considerably lower bit rate but it shouldn't be that noticeable. I also have uploaded 3 shots of video frame showing the original, one compressed with CCE and one compressed with Roxio DVD builder. I can't remember the CCE settings but I'm sure they weren't optimal since I was still learning at the time I did it. As you can see Roxio looks like crap.
     
  5. Minion

    Minion Senior member

    Joined:
    Sep 28, 2003
    Messages:
    5,623
    Likes Received:
    0
    Trophy Points:
    116
    Most All DV Is interlaced Bottom Field First unless you have a 24p(FPS) Film DV Camera But those are Extremely expensive, accept From What I hear that the Microsoft DV Codec does Deinterlaceing with DX9 ...Much of the Time it is Hard to tell if a frame is interlaced or Not unless there is Moovement but one way to tell for sure is to use a Tool To analize the Frame for it"s Picture Structure...There is one such tool in AVISynth but it is easier to just Load the DV AVI file into Tmpgenc useing the WiZard which will automaticly Analize the Files for it"s Frame Structure and if it is Interlaced it will then analize the Frames for the Correct Field order which in Most all cases will be Bottom field First...There are all Different Kinds of deinterlace Filters, Some of them Remove One Field some of them Blend Both Fields together and some Only Blend were there is Moovement, and Not all Deinterlace Filters make your File Progressive either..In most Cases you should only really see Interlace artifacts on your TV when the Field order is set incorrectly as 30fps 60fields/ps DV and Broadcast TV signals are displayed in the Same way so in theory if you get interlace artifacts on your TV set when displaying Interlaced DV Material then you should also get Artifacts when Watching Broadcast TV...Most any Capture program will Capture Interlaced AVI as Long as the vertical resolution is above 240/288-Pal, Like Virtual-Dub ,Virtual VCR or any other VFW or Direct Show capture program....
     
  6. BJKESLER

    BJKESLER Guest

    http://www.dvdrhelp.com/comparison
    Compares Nero 5.5 with TMPGENC and Cinema Craft 2.5 and the results "the differences arent so big between all these encoders". My tests show the same. I know some people are unable to discern differences in video qualities on their TV; just as some can't discern differences in audio qualities. If the VCD stuff was so hot then why do the prerecorded movies look so crappy. I'm sure they have access to the best hardware and software. I am unable to invest the time others do on this problem, especially when they still come up with an inferior product. I'll wait for the affordable technology to get their.
     
  7. Caveman3

    Caveman3 Member

    Joined:
    Feb 7, 2004
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    11
    I'd like a piece of this one...
    Pardon the rant... I will include some info as well...

    I have been fool'n with this stuff for some time... waiting for the right price on hardware and software to archive and edit/play family (mostly Hi-8) videos for >10 years. With all the DV buzz starting a couple years back and over the last year very low cost DVD (4.7GB) burners AND the availablity of good? mpg encoding tmpg, procoder home, etc I thought the time had finally arrived to archive the tapes on DVD...
    I have found that the analog to DV process (even back a few years) can capture Hi-8 with little loss but I have not found an acceptable technique to create DVD for play back on a standard interlaced TV. I agree completely regarding the limited support from even "pro" hardware/software vendors. How can one hope to keep interlace setings correct when some (pinnacle/adobe) say lower=fld2=B=even and others (canopus, Ulead?) say lower=odd and lower=A. In any event, I have burned disks with swapped settings all to no avail.
    A recent test:
    1) Capture analog via pro-one (lower first DV) (DV import would be the same but a bit better qual).
    2) Check by capturing a FRAME in adobe photo shop and using it's "dumb de-interlacer and observing that deleting the odd field shows the picture taken earlier in time (therefore lower/even was captured/placed first).
    3) Converting to 2 mpgs one set "A" first and one "B".
    4) Captured again in photoshop and learned which encoder setting resulted in a file that matched the original.
    5) Used that file and burned DVD... Still bad! tried the other... also bad!

    -Don't trust any posting of "screen capture" proof... many applications process the image more than the user knows!
    -I guess we could snap photo of a TV screen playing DVD and use that to make a case...
    -De-interlace (even smart) is not the answer as it is still a change/loss from original (remember DV camcorder is generally interlaced). In any event even if a real smart de-interlacer were used (shifting just areas of motion the correct amount) the result would look bad on an interlaced TV (in motion scenes)!
    -One solution alluded to in the thread.. shoot video w/o motion!

    Any others??
     
  8. Caveman3

    Caveman3 Member

    Joined:
    Feb 7, 2004
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    11
    Don't know if any are still following this thread but I have looked at the mpeg coding (and options) a bit more and do have some ideas that may help with some (many?) of the problems noted. I'll try to be brief but still say enough to communicate the ideas!
    The mpeg "spec" allows video to be encoded in several very different ways (the ways can change within one file, see PROs below)...(the following discussion is for interlace 30 not progressive or 3:2 pull down). I have seen it stated that not all stand alone players well support all the stream types!...
    Here are a list of some items perhaps more critical to solving "our problems" than the TFF flag (that's top field first (or not) flag).

    Interlaced video can be encoded by splitting the fields into 2 pictures (odd lines and even) or encoding the frame of both odd and even. Not all players can handle the field encoded format (I've read). On a "still" video the "frame picture" is much better qual/bit rate as the features in the picture are "smooth" and easy to encode while the field pictures would be rough ad coarse (those edges will be lost/softened by mpeg) plus twice as many motion prediction etc because of the missing lines (even then odd). With motion the "frame picture" will have very rough/comb edges that will also be lost/softened in mpeg... in that case a field picture would probably be easier (higher quality) to compress than the "frame" style. This problem requires much human labor to solve (ie select variable encoding types depending on video, I think the PROs actually do a bit of this) or very clever software! This is perhaps at the root of the noted quality limitations.

    Another issue is encoded order Vs. display order:
    When "B" frames are used the orders are not equal. Also the GOP (group of pictures "IPB" chain) can be "open" (it is not required to start with an "I"). I was making VERY bad quality DVDs when viewed on TV that looked "OK" on the computer. The problems were not simply de-interlace/field order. I had been (w/o planning to) encoding open GOPs and think that my APEX player could not cut it and so was flashing up pictures out of sequence! (talk about tough on the eyes!). I'm still testing the performance of Frame Vs Field pictures etc. but wanted to share this much sooner rather than later as I can now also make Decent quality DVD...

    CHECK your players ability to handle Picture fields and open GOPS!

    Perhaps this will spark more info sharing comments...

    additional notes: The apex seems to deinterlace on frame hold so even real jittery stuff looked "ok" on pause.
     
  9. trx

    trx Member

    Joined:
    Jan 17, 2004
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    11
    I was capture more DV video an noticed something that may add to the confusion. I've noticed that in some of my captures I did not see any interlacing artifacts but it was evident in others. In one capture there was one point where the camera was stopped and restarted later. In the first video segment the video was artifact free and in the second segment there were artifacts. What makes this odd is this was in a single capture to AVI (one AVI file). It looks to me that for some reason the field order was reversed when the camera was restarted. As stated earlier DV is bottom field first so the fields are recorded in this order: BTBTBTBT... So is it standard that the recording must always start with a bottom field *and* end on top field? It seems like this may not be the case from what I saw. If the first recording was possibly stopped on a bottom field and then the second was started on a bottom field then there would point where there were there were two bottom fields. I guess it could look something like this: ..BTBTB|BTBTB... (| is where the recording stopped). How is the DV codec to know that that the bottom field was repeated? Would it be safe to assume it will interpret the second bottom field as a top field and cause the fields second video segment to be out of order? I know this probably doesn't explain all the scenarios but it this situation is seems like the source video could have caused this problem since the first segment was perfect and had the most motion. I'll have to try more captures from different spots on the tap and see if I can get different results.

    Does anyone know of any video editors that allow you to edit fields instead of full frames?

    Caveman3, although you don't think anyone should believe any screen captures I think it helps to shed some light on the problem. Some people can post their 'facts' all they want but I won't have much faith in it unless they back it up with some sort of visual evidence, state the source of their information, or provide specific steps to reproduce what they say is possible. The frames I captured were of the DV source and the deinterlaced source in VirtualDub. The interlacing artifacts where the same in all programs I used (Windows Media Player 9, Media Player Classic 6.4 and VirtualDub). I agree even smart deinterlacers (like the one I used) are not the best solution but it's what I'm using for now until someone has a documented and repeatable solution.
     
  10. Caveman3

    Caveman3 Member

    Joined:
    Feb 7, 2004
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    11
    Trx, No offence to your pictures was intended (I didn't even look at those!). This thread was started (and has a title) relating to interlaced work not de-interlaced. (Oops, I take that back... I guess I was responding most to Bernie and BJKESLER's comments!). Anyway, what I was trying to get across is that the interlace "artifacts" are actually "FEATURES" of an interlaced stream. For those of use wanting to preserve as much quality as possible from our footage while transfering to DVD for durablity and easy of playing back/distributing we want to keep the data interlaced. You are correct that Progressive (single time) frames are better (higher qual/bit rate) to compress into mpeg. What I recently learned was that the mpeg format allows "field pictures" (60/sec)although not all stand-alone players may handle (I haven't tried much and was hoping to hear from some experts). So, as you note AVI is frame based (that does not really mean "progressive" as frames can be progressive (representing on moment in time) or interlaced (representing 2 1/2 resolution pictures taken at twice the frame rate). My understanding is that most mpeg software attempts to compress by frame (Yes, even interlaced streams!). When the fields are different (motion) the Frame picture will have VERY sharply defined features line to line... mpeg/jpeg etc is not good at preserving all that data so on display the edges will be softend (We WANT the sharp jaggie lines in the frame captured from mpeg) ... the field mode encoding can male a big change to that quality but I'm not sure waht compressors can do it, If they can properly select the mode to achieve decent bits rates AND if the resulting disks can be played on most/many/any players.
    As to your stop/start issue I "suspect" that the system would not shift order it would simply have a blank field interlaced in with the last image...

    I think I may start a new thread specifically relating to "INTERLACE ALL THE WAY" tape to DVD...

     
  11. Bernie

    Bernie Member

    Joined:
    Mar 17, 2003
    Messages:
    6
    Likes Received:
    0
    Trophy Points:
    11
    I find it interesting the amount of disagreement that continues about how much difference the brand/make of MPEG2 encoder makes on final DVD quality—even in this thread. In my experience, the vast majority of the more scientific, less biased studies indicate that there isn’t much difference in the results from different encoders when equivalent settings are used. Conversely, the more-manufacturer-biased and less-experience-based studies always seem to state a HUGE difference. I have tried eight different popular consumer level encoders, ranging in price from free to $1000. In my experience, using equivalent settings—measuring apples to apples—the differences were VERY slight.

    On the other hand, Gusto, in response to your question, I have tried recording to a dedicated standalone DVD encoder/author/recorder connected to a video source through a IEEE1394 (firewire). And interestingly, this DID seem to have a significant impact on my results. The degradation problem is significantly better; not perfect, but far better than the results from any variation in the software and hardware of computer-based encoders/burners. I’m still trying to figure out why this is the case because, as I mentioned before, I have used some pretty beefy computer-based hardware/software, and even with optimized settings, achieved less improvement. Anyone know why this might be the case?

    Actually, I might even be inclined to use the standalone encoders/recorders more exclusively if it weren’t for the limits in editing, and setting up menus and chapters currently inherent in all of these units.
     
  12. Bernie

    Bernie Member

    Joined:
    Mar 17, 2003
    Messages:
    6
    Likes Received:
    0
    Trophy Points:
    11
    I apologize in advance for the length of this message, but you know how it is when you hear from others on a topic that you care about. . . . . .I just had to respond to you, trx and Cavemen.

    Caveman, always good to know when even more folks have a first hand understanding of our concerns. You said that you have also NOT found, “ …an acceptable technique to create DVD for playback on a standard interlaced TV.” That makes six people just in this thread alone that seem to be seeing eye-to-eye on poor results in consumer level MPEG2 encoding products. Of course, I would say: …poor results not just on an “interlaced TV” but on ANY TV—from my grandmother’s rotary dial TV to my LCD panels to my Sony XBR progressive hi-def flat screen. The "pros" in the marketplace would say we were all crazy: "Just look at 'Lord of Rings' on DVD and tell us that quality DVD is not attainable!"

    Caveman, I do think we do have some minor differences of opinion, though. First and foremost, I do not see that “shooting only video without motion” is any sort of a solution. I don’t know about you, but most of the time I do not have direct control over the amount of motion in my videos. Plus, if I wanted video without motion I would just shoot photos! :)

    Also, I am not so quick to discount the utility of viewing still captures of video in helping to analyze video results. I agree, many applications modify the stills more than most users realize. However, I would not go so far as to say stills cannot be trusted and/or useful. Case in point are trx’s video stills: trx, I am a “show-me-or-I-am-not-buying-it” type of person and I really appreciated your posting of your test results screen shots. 1) They ARE an accurate reflection of some of the degradation that so many of us have experienced and have tried so hard to describe, and 2) I think your steps toward alleviating the problem are worth a test. It is refreshing to hear from someone who is precise, thinks twice before stating something as “fact”, and is willing to back up his beliefs with further supporting information and unbiased logic. Again, there is so much misinformation spread so many of the “pros” in this marketplace.

    Caveman, makes a valid point that some might say this de-interlacing method of improvement may be still too “lossy” for those looking to make the best possible reusable archives. But while this may be true, I think the whole concept of using MPEG2 encoded DVD’s/VCD’s for best reusable archives is a farce. Even though disc media outlasts tapes, they simply do not maintain the quality of the original. I, for one, am not throwing away original tapes—Hi8, D8, MiniDV, or even VHS—until a totally different, less lossy technology comes along. However, I am still hopeful that DVDs may be used for delivering a decent quality end product where the generational loss and further editing are NOT important. To that end, I continue to take interest in the hints passed along in this thread.

    I would still like to believe that creating DVDs with “close-to-original” quality video—not for use in later editing—is achievable. And because I am pursuing this goal, I do not agree that interlacing “artifacts”—or really ANY artifacts for that matter—are “features.” This is not to say, Caveman, that I was not interested in your ideas about the idea of going “interlaced all the way.” Normally, keeping the highest quality as long as possible in the editing process is desirable. But, at this point, I am interested ONLY IF this approach leads to less degradation and less artifacts in the end. And unless I misinterpreted what you were saying, you made it sound like the strength of this approach is definitely not artifact reduction. Thus, I must say that, at this point, I am more interested in the results attained thru de-interlacing as they do appear to make a more immediate positive impact. I can’t wait to recreate your tests for closer review. I will certainly post my test results. I hope you will find them as interesting as I have found yours.
     
  13. Caveman3

    Caveman3 Member

    Joined:
    Feb 7, 2004
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    11
    Another long one!

    I really haven't been satisfied by the "no motion" technique either ;)
    (but it would eliminate all this interlace trouble!).

    Interlace start to finish will/DOES get the best results!... read on

    As it is tough to visually compare interlaced frames most of the encoder comparisons I've seen were deinterlaced for review... The trouble is that deinterlacing is a job that is often tougher than mpeg encoding so those comparisons are not very good for checking interlaced encoding performace.

    I have been misunderstood! I intended to say that what some call artifacts are actually features (an integral part of) the interlaced format...They only become MPEG ARTIFACTS when messed with in encoding or DEINTERLACE ARTIFACTS when poorly deinterlaced for display in progressive format.

    My point about interlace "artifacts" should perhaps be clarified...
    Recording a scene containing motion, via an interlaced format recoding device will (and should) result in frames with sharply defined "jaggies"...That is indeed a feature/fact of the format. It has been noted that tape to DV/avi back to tape works pretty well. From that we can conlude that the DV/avi is a "good copy" of the data on tape. My goal is to get the data to mpeg/dvd (and back to a display, or at times back to a dv/avi file) without significant change in the data....

    I don't have the following worked out in practice and so don't have any sample pictures to post...but can still use it as a "thought experiment"...

    Capture DV/avi from in interlaced source
    mpeg the file
    Grab a frame from the original DV/avi file.
    Grab the same frame from the mpeg encoded file, invert one in "photoshop" add them together and attempt to get "null".

    Obviously if the mpeg encode has been deinterlaced this test will fail badly, so that ends the deinterlace route to "quality".

    I think I was a bit harsh in my first posts to this thread...
    I am burning my best original material to data DVDs as dv.avi files, but I have also been saving complete copies of the tapes at approx 1Hr per DVD-"DVD format" (this is ~the max data rate allowed by DVD format). I think that the data is in pretty good shape and would get decent marks on the compare to original test above.
    Have you (Bernie/others) tried high data rate interlaced start to finish, played on a stand-alone DVD player/TV?

    I DO FIND that this results in the best quality!... I only consider deinterlace when I plan to "distribute" for PC only and so then do not make "DVD format" disks.

    To maximise the encoder performance of interlaced material, I still think that high motion scenes should be encoded with "field picture structure" but have found little info about software/hardware support for this feature of the mpeg specification. Do you know what I'm talking about? Do you have info to share regarding "picture structure"?

    Thanks for continuing this discussion!
     
  14. Caveman3

    Caveman3 Member

    Joined:
    Feb 7, 2004
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    11
    Oops,
    I meant to add ...
    Bernie, have you taken a look at the mpeg files generated by the hardware DVD deck? You if not, please do and share your findings! Use a tool like bbmpeg tools bbvinfo.exe to view the picture/frame structure, IPB count etc (I'd be very interested if it uses any "field pictures" but suspect that it does not. I would be VERY surprised if it did not use interlaced frames... Feel free to clip a section of the txt report and post it here...

     
  15. GeorgeK

    GeorgeK Member

    Joined:
    Feb 22, 2004
    Messages:
    2
    Likes Received:
    0
    Trophy Points:
    11
    I bought my first DV camera recently. So, once I did, I attempted to use a variety of $99 - $499 DVD authoring programs to capture my DV stream and put the information onto a DVD. The results were invariably awful--any time there is motion, vertical lines become teeth on a comb. After reading all the postings here and doing a bunch of experimentation, I came up with two viable schemes. (1) Take your AVI file and deinterlace it with DVFilm Maker. This is an affordable ($100) program that makes a de-interlaced AVI from an interlaced one. The secret here is not in deinterlacing but in doing it with a motion-compensated algorithm. You can then throw the new AVI file into any of these DVD-authoring-for-dummies programs and you won't have the comb artifacts. (2) Capture your AVI file and manually MPEG2 encode it. Capturing is easiest with the automatic MovieMaker function in Windows XP, but all methods give the same result. I spent a great deal of time researching MPEG encoders and concluded that the inexpensive ($48) TMPGEnc Plus is suitable, in addition to being reasonably priced. Again, the important thing is that it has competent motion compensation in its algorithm. Once you convert the AVI to MPEG with it, you can again finish production with any authoring package. But if you're going to do this, then DVD Lab ($99) seems the best, since it is neither for dummies nor geeks, but for intelligent folks in-between. It has good flexibility and control, yet is not intimidating.
     
    Last edited: Feb 22, 2004
  16. Bernie

    Bernie Member

    Joined:
    Mar 17, 2003
    Messages:
    6
    Likes Received:
    0
    Trophy Points:
    11
    GeorgeK, thanks for taking the time to read through this thread and jumping into the fray. I read your comments and Caveman’s latest entry with much interest. First, Caveman, I would like to respond to some of your comments. You said, “…DV/avi is a "good copy" of the data on tape. My goal is to get the data to mpeg/dvd (and back to a display, or at times back to a dv/avi file) without significant change in the data....”

    Even though I have pretty much given up on coming back from mpeg/dvd to dv/avi with any sort of quality, I think that’s a good way of stating the goal concisely. I liked your above “thought experiment”; it is a great example that would certainly help clarify opinions. Using that same test, I think the result that I would hypothesize would be different than yours. Perhaps I am a bit too cynical but based on my experience, instead of hypothesizing, “…if the mpeg encode has been deinterlaced this test will fail badly…”, I would simply say, “…the mpeg encode, it WILL fail badly…period.” GeorgeK, you’d be the lastest one to support me on this view, I am sure. So the question in my mind becomes, “Which fails worse: the deinterlaced MPEG encode or the interlaced MPEG encode?” As I stated previously, I am not as quick as you, Caveman, to conclude that the deinterlaced encode will fail worse. Isn’t it conceivable that even if the deinterlaced version is technically “less accurate”, there is a chance it might still produce a better looking DVD result (less comb teeth, less strobe, less blocky, etc.) because it made the subsequent MPEG2 encoding go easier? I think this is what TRX was alluding to earlier in this thread—that MPEG2 encodes from deinterlaced files actually enhanced image detail for him in some ways. I can’t wait to run more tests, it’s just a matter of time.

    I do completely accept, and agree with, your practice of, “…burning my best original material to data DVDs as dv.AVI files, but I have also been saving complete copies of the tapes at approx 1Hr per DVD-"DVD format" (this is ~the max data rate allowed by DVD format).” For others reading this thread, I think this is about as sound of advice as it gets.

    Though, keep in mind what I consider to be the limitations: Even at the max MPEG2/DVD data rate—which I highly recommend using—there will still be significant differences from the original. And regarding this very-best-quality-saving practice of putting DV/AVI files to data DVDs: it only allows about 14 to 15 minutes of video per DVD using current technology (or at least that’s been my experience). In my books, that’s a pretty restrictive limit. But hey, when you absolutely need to get your video off tape and not lose any quality, what other reasonable cost options do you have? It’s certainly not the current MPEG2 DVDs!

    Caveman, I have been reading a ton of literature but have not come across as much as I would like regarding “field picture structure.” Furthermore, none of my software packages have made much mention of this, either. I’ve got a ton of things on my “to do” list already, but I’ll see if I can get my hands on some analysis tools like the ones you mentioned. You’ve got me curious as to whether or not I could learn something from the decent quality DVDs generated from the standalone deck. At a minimum, it would be nice to know if there is anything technically different about these.

    GeorgeK, you’ve also got me curious. Do you know what technology is used to produce the “motion-compensation” about which you are referring? Please pardon the reservations, but where I have seen the term “motion-compensation” used, it has been abused by sales and marketing folks. Often vendors of lower end video products call their blurring and filtering algorithms: “motion-compensation”. Some vendors believe “motion compensation” and “deinterlacing” are the same thing. While higher-end vendors often consider only complex, predictive field/frame weaving as truly “motion compensating.” I guess I prefer to stick with terms that are less open to differences in interpretation. I will read up on the products you mentioned and see if they go into more detail about what they mean. If you already know, I would be interested. I am always hopeful I might stumble across something that provides a significantly better end result.
     
  17. GeorgeK

    GeorgeK Member

    Joined:
    Feb 22, 2004
    Messages:
    2
    Likes Received:
    0
    Trophy Points:
    11
    Concerning the deinterlacing technology used, no, I don't know the technical details. But, in the final analysis, it doesn't matter--the proof is whether it gives you a good video or not, as visible on your screen. So DVFilm Maker and TMPGEnc Plus evidently do things right, since the output is good. The rest of the packages I tried, do not. There may also be viable solutions via very geeky techniques (i.e., AviSynth etc.), but I figured that if I as an engineer cannot make heads or tails out of a technique in the first 1/2 hour of study, hopefully I won't have to use it.
     
  18. jantiff01

    jantiff01 Guest

    ** I've tried several MPEG2 encoders for making DVDs. The ones I've tried seem to be of only moderate quality; some are pretty poor and one (which I won't mention) is so bad that it's unusable. I tried capturing some live broadcast video via FireWire; in DV format it was great but the 3 MPEG2 encoders I tried out to encode the DV files for DVD left a lot to be desired. When viewed over a studio monitor, all the high-frequency detail was gone and it looked dull, flat.
    ** I've heard a lot about Canopus ProCoder and TMPGEnc Plus. How good a job do they do overall? If it comes down to one or the other of those, which has better performance?
     
  19. cdulde

    cdulde Member

    Joined:
    Feb 24, 2004
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    11
    From a posting I recently did on Canopus' forum - similar question I had...

    I did some testing and thought some might find this valuable...

    Encoder, Time & Quality
    --------------------------------------------------
    ProCoder Express 2.5:1 Excellent
    MainConcept MPEG Encoder 1.25:1 Very Good
    Let's Edit (internal encode capability) 4:1 Very Good
    ULEAD Movie Factory 2.0 3:1 Good
    Roxio EZCD & DVD Creator 6.0 3:1 Good

    Other comments:
    Similar settings were used on all encoders to put 2 hours of video onto one compliant DVD. Bitrate works out to be something like 4500000 bps.

    Subjective comments:
    ProCoder Express is by far the best of those encoders I worked with (used Highest Quality mode) (I didn't use TmpGen as I felt it would be too slow from what I read in other forum postings - although quality is supposed to equal PCE). MainConcept was a very close second to PCE. It is a lot faster and produces nearly the same quality, but I felt the MainConcept MPEG had a 'haze' on it that muted the colors. Not a big deal, but enough for me to stick with the longer encoding time of PCE. The encoders for ULEAD and ROXIO are general purpose, quick and dirty, get it on DVD types. I had varied results using them and frankly, felt they're not in the same class as the other three. The Let's Edit encoder that comes built in to that software (I believe it is an earlier Canopus product called SoftMPG), produces quailty of about the same as PCE. But it is s l o w...

    From another user:

    TMPGEncPlus 6.5:1 Excellent

    I used the highest quality MPEG-2 video settings at CBR = 8000 with linear PCM audio on an Athalon 1800+.

    Granting system processor and encoder setting differences 6 hours encoding for 55 minutes of video is incredibly slow. However, it does look great, I'm very familiar with the program and not at all pressed for time so for now it is the ticket.

    Here's the link if you want to check out the thread, but you may have to register to gain access...

    http://forum.canopus.com/showflat.php?Cat=&Board=ProCoderExpress&Number=158300&fpart=&PHPSESSID=
     
  20. cdulde

    cdulde Member

    Joined:
    Feb 24, 2004
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    11
    Clarification on last posting....

    ProCoder Express is from Canopus (www.canopus.com)
    MainConcept is from MainConcept (www.mainconcept.com)
    Let's Edit - is a non-linear editing package that has MPEG encoding built in and is from Canopus.

    the TIME listing is a ratio of Encode Time to Playing time.

    So, 2.5:1 equates to 2.5 hours of encoding time to 1 hour of playable video.

    Chuck
     

Share This Page