Video Quality Question

[Augie5390]

I am trying to convert VOB files from a dvd to either AVI, MPEG or whatever so that I can edit it on Windows Movie Maker. I am using OJOsoft Total Video Converter. My problem is that everytime I convert the VOB the video quality is bad. What should I do to convert into better quality?

[filmboss80]

Convert to a higher quality codec, like DV-AVI.

[guns1inger]

Open the VOBs in AVIDemux or Virtualdubmpeg2. Save as DV AVI as suggested above. Edit in Windows Movie Maker (understanding that any transitions or effects you apply force the video to be resized to 640 x 480, then back again), then output as DV AVI from WMM. Finally, convert to your chosen format using a quality encoder.

[Augie5390]

I downloaded AVIDemux and tried to save it as a DV AVI but it won't let me. It says, "INCOMPATIBALE SETTINGS-At the moment, the DV codec only accepts 720*576@25" What does that mean? Also when I upload my VOB file it says, "This file looks like and MPEG, do you want to idex it?" I say yes.

[filmboss80]

You are in the United States. You need to set for NTSC video, not PAL. The aspect ratio for DV-AVI is 720 X 480. Frame rate should be 29.97 (or sometimes depending on original film source, 23.97 fps).

[45tripp]

Avanti has a working ntsc dv template.

[deadrats]

I am trying to convert VOB files from a dvd to either AVI, MPEG or whatever so that I can edit it on Windows Movie Maker. I am using OJOsoft Total Video Converter. My problem is that everytime I convert the VOB the video quality is bad. What should I do to convert into better quality?

first things first, VOB files are already mpeg files, think of the Video OBject file as a specialized container for mpeg-2 video and any of the following 3 audio formats: LCPM (uncompressed audio), MP2 or AC3.

now on to the editing: what exactly are you trying to do, add effects, transitions, cut, splice, etc? the reason i'm asking is because windows movie maker is probably not your best choice unless you want the final output to be in wmv format.

furthermore, you should not be first converting your video and then editing it, you should always first perform whatever edit you want and then convert to whatever target format you desire.

something to keep in mind: for maximum quality you are better off not converting the source dvd to another format. mpeg-2 is already a lossy codec so transcoding to another lossy codec will almost always result in a lower quality end result.

also, for maximum quality, never resize the file, if it's an ntsc dvd (720 x 480 @ either 4:3 or 16:9 pixel aspect ratio) don't crop and/or scale it to anything else (you will may get people telling you to resize to some idiotic size for some truely moronic reasons, resist the brain dead advice <---get the feeling that i am really opposed to resizing?).

[45tripp]

furthermore, you should not be first converting your video and then editing it, you should always first perform whatever edit you want and then convert to whatever target format you desire.

that sounds close to the 'brain dead advice' you refer to.

if editing is simple cuts, maybe,
if there's disk space to burn,
convert to losseless and edit that.

[deadrats]

furthermore, you should not be first converting your video and then editing it, you should always first perform whatever edit you want and then convert to whatever target format you desire.

that sounds close to the 'brain dead advice' you refer to.

if editing is simple cuts, maybe,
if there's disk space to burn,
convert to losseless and edit that.

ask anyone that does any serious editing, including any pro/semi-pro it's beyond stupid to convert your source to another format, any format, then perform the editing and then convert again.

i'm just dying to hear your reasoning behind the notion to a lossless format, other than maximum compatibility with video editing software that is.

do you really think that any additional quality will be achieved?

[45tripp]

i'm just dying

you certainly are.

i'll let someone else pick up the queries.
suddenly i'm bored.

[filmboss80]

If the source is MPEG1, MPEG2, or Divx/Xvid, for example, editing can only be done on I-frames. Some projects may require upconversion to a format that can be edited on a frame-accurate basis. The dilemma one always faces is whether to reencode for greater editing flexibility or maintain the original at its source, accepting whatever editing limitations that entails. The choice is made on a project-by-project basis, and I have certainly done it both ways, depending on the priorities. To call one perfectly valid method "stupid" is certainly presumptive. I have had many good results where MPEG2 footage was converted to DV for editing, then reencoded back to MPEG2 for DVD Authoring and no viewer could perceive any sort of quality difference -- even watching on an HD monitor. Deadrats, you can all it stupid if you like, but the process has worked well for me, as well as a number of other PROFESSIONALS. (And yes, I have been working professionally at video post-production since 1980--and 16mm film before that.)

[jagabo]

There are several editors that can edit MPEG 1, 2, and 4 video frame accurately, and only re-encode cut GOPs. Obviously, transition effects and other types of filtering require reencoding.

[deadrats]

If the source is MPEG1, MPEG2, or Divx/Xvid, for example, editing can only be done on I-frames. Some projects may require upconversion to a format that can be edited on a frame-accurate basis. The dilemma one always faces is whether to reencode for greater editing flexibility or maintain the original at its source, accepting whatever editing limitations that entails. The choice is made on a project-by-project basis, and I have certainly done it both ways, depending on the priorities. To call one perfectly valid method "stupid" is certainly presumptive. I have had many good results where MPEG2 footage was converted to DV for editing, then reencoded back to MPEG2 for DVD Authoring and no viewer could perceive any sort of quality difference -- even watching on an HD monitor. Deadrats, you can all it stupid if you like, but the process has worked well for me, as well as a number of other PROFESSIONALS. (And yes, I have been working professionally at video post-production since 1980--and 16mm film before that.)

here is why i call it stupid: codecs are programs that compress and decompress either a video or audio stream. the compression part of the codec comes into play when you store the video/audio in digital format, regardless of the media used, and the decompression part comes into play when ever you try to view the contents of the stream(s).

media players, both software and hardware, decompress the audio and video on the fly and then display the image and/or sound. video encoding/transcoding software does the same thing, it takes the video stream, frame by frame decompresses it on the fly, and then performs the encoding to a new format. video editing software does the same thing, it first decompresses the frame then recompresses it using the new compression scheme.

you can confirm this behaviour one of two different ways: examine the source code to any of the open source video encoding/transcoding/editing apps, such as avidemux, virtual dub, cinerrela, etc or install ffdshow and configure it so that it decodes all codecs, then load any video file you want to any editor you choose and notice the icon on the lower right hand of the screen from ffdshow indicating that it is decompressing the file in question. for further proof run a debugger at the same time and examine the memory being used by the video editing app in question.

as far as i can tell, from looking at the source code to open source video editors, as well as examining the behavior of various closed source and open source video editing suites and based on what i know about the way codecs work it seems to me that all video editing apps decompress the file on the fly prior to performing any editing on it.

this also jives with what i know about compression schemes in general (in college i was a physics/comp sci major and one of my hobbies is working on my own compression algorithm). it's just like having a tar archive, you can edit a text file within a tar (or any other archive for that matter) without first decompressing the file in question, same holds true for video/audio streams.

so if the program is going to decompress the stream anyway prior to editing, what's the point in converting it to a lossless format first?

[filmboss80]

Deadrats: no one is arguing about the compression / decompression issue. You are absolutely right on that account. Trust me, you don't need to educate us. But the OP wanted to use Windows Movie Maker (I agree, not a good choice), and we answered his post within that parameter. I certainly would not edit the way he is editing, but I respect that he is most comfortable with WMM. The advice 45tripp and I offered was for the purpose of helping the OP find a workable solution within the realm of his resources. I don't think it's helpful to call that workaround stupid.

[jagabo]

Some processing can be done on frames without fully decompressing them. For example, you can adjust the brightness and contrast of any DCT compression based video without performing the iDCT and re-DCT steps. Enosoft has tools that can do this with DV.

As has been pointed out there are only two reasons to convert to uncompressed or losslessly compressed before editing: for ease of editing (seeking in long GOP codecs like AVC can be very slow) or your editor just doesn't support the format the video is in.

[deadrats]

Deadrats: no one is arguing about the compression / decompression issue. You are absolutely right on that account. Trust me, you don't need to educate us. But the OP wanted to use Windows Movie Maker (I agree, not a good choice), and we answered his post within that parameter. I certainly would not edit the way he is editing, but I respect that he is most comfortable with WMM. The advice 45tripp and I offered was for the purpose of helping the OP find a workable solution within the realm of his resources. I don't think it's helpful to call that workaround stupid.

the point is that windows movie maker works the same way as i described, it internally decompresses the frame, performs whatever edit you tell it to and then recompresses it. i see no point in advising him to first decompress the file to a lossless format if windows movie maker is going to do that internally anyway.

as for the "stupid" comment i made it wasn't directed at the advice given him, it was more of a general comment born out of frustration with a trend to resize video to weird resolutions for God only knows what reason.

as an example, in the days when dvd was the highest quality we had, it was accepted to resize 720 x 480 4:3 or 16:9 video to all sorts of odd resolutions in a misguided attempt to get video to "display properly" on a computer monitor (not realizing that pixel aspect ratio was not the same as display aspect ratio) and so 720 x 480 16x9 would get resized to 852 x 480.

when hi def started become widely available i had hope to see this trend end as hi def video has a pixel aspect ratio of 1:1, just like the pixels on hi def tv's and computer monitors.

what's driving me up the wall is that i keep seeing movies encoded to odd resolutions like 1920x800, 1280x544 and 1920x816 (these are the most common i have seen) and i can't figure out the reasoning between resizing/cropping a 1920x1080 movie to 1920x800.

[jagabo]

what's driving me up the wall is that i keep seeing movies encoded to odd resolutions like 1920x800, 1280x544 and 1920x816 (these are the most common i have seen) and i can't figure out the reasoning between resizing/cropping a 1920x1080 movie to 1920x800.

Aren't they just cropping away black borders? 1920x800 would be a 2.4:1 AR movie. 1280x544 would be a 2.35:1 AR movie.

[deadrats]

As has been pointed out there are only two reasons to convert to uncompressed or losslessly compressed before editing: for ease of editing (seeking in long GOP codecs like AVC can be very slow) or your editor just doesn't support the format the video is in.

actually i was the one that pointed out that there is only one reason to convert to uncompressed or lossless, for the compatibility with all video editors.

as for ease of editing, yes seeking in long GOP in AVC is very slow, but all you are doing is shifting the time it takes to edit the video to the time it takes to convert to lossless and then back. in other words instead of the editing taking a long time you now have 2 transcodes that will take a long time, you really haven't saved any time at all, if anything you have made the project even longer.

[edDV]

As has been pointed out there are only two reasons to convert to uncompressed or losslessly compressed before editing: for ease of editing (seeking in long GOP codecs like AVC can be very slow) or your editor just doesn't support the format the video is in.

actually i was the one that pointed out that there is only one reason to convert to uncompressed or lossless, for the compatibility with all video editors.

as for ease of editing, yes seeking in long GOP in AVC is very slow, but all you are doing is shifting the time it takes to edit the video to the time it takes to convert to lossless and then back. in other words instead of the editing taking a long time you now have 2 transcodes that will take a long time, you really haven't saved any time at all, if anything you have made the project even longer.

Except you can be doing other productive work during the transcode. Serious editing of long GOP AVC (e.g. AVCHD or various h.264) or sometimes even HDV is extremely frustrating. I'm in the digital intermediate camp myself (e.g. Cineform or Apple Intermediate Format). These have the least loss and allow productive editing and processing.
http://www.cineform.com/

[edDV]

...as far as i can tell, from looking at the source code to open source video editors, as well as examining the behavior of various closed source and open source video editing suites and based on what i know about the way codecs work it seems to me that all video editing apps decompress the file on the fly prior to performing any editing on it.

Some of the better editors use the "smart render" concept where unprocessed frames on the timeline are simply copied from the source file to export thus saving a generation. GOP based formats copy unprocessed GOPs and only recode processed GOPS (e.g. a cut within the GOP). This is often done for uncompressed, DV and sometimes MPeg2 project formats (e.g. HDV/XDCAM) where the export format matches the source format.

[deadrats]

what's driving me up the wall is that i keep seeing movies encoded to odd resolutions like 1920x800, 1280x544 and 1920x816 (these are the most common i have seen) and i can't figure out the reasoning between resizing/cropping a 1920x1080 movie to 1920x800.

Aren't they just cropping away black borders? 1920x800 would be a 2.4:1 AR movie. 1280x544 would be a 2.35:1 AR movie.

ah!!! not the "black border" theory again!!! (<---note this isn't directed at you), i thought once we had moved onto hi def video i would never have to hear about that silly idea again.

if the video stream is 1920 pixels wide by 1080 pixels high and the pixels themselves have an aspect ratio of 1:1 and both your hdtv and computer monitor use pixels with a ratio of 1:1 what "black borders" could possibly exist? where would you see these "black borders"?

a 1920x1080 movie has a display aspect ratio of 16:9 and a pixel aspect ratio of 1:1, in what way does it make sense to try and resize it to a 2.4:1 aspect ratio?

same thing with 1280x720, it also has a dar of 16:9 and a par of 1:1 so why resize it to an aspect ratio of 2.35:1?

i have even seen a 1440x1080 movie (4:3 dar and 1:1 par) resized to some bizarre resolution (i don't quite remember what it was but i remember being like WTF).

[filmboss80]

For Augie5390, I think we can all agree that between DV-AVI and WMV formats (for Windows Movie Maker, the app he is using), he'd get best the quality editing in DV-AVI. It looks like he had trouble with AVIDemux, likely because it was set for PAL system DV, not NTSC. Avanti is a good converter also, but the OP already has Total Video Converter. I haven't used that app, but according to its maker's website, it will make DV-AVI video files from VOBs.

Now, of course, there are many other ways for the OP to edit VOB files, as evidenced by the lively discussion here.

What else can we contribute to make Augie's life a bit easier?

[edDV]

Doesn't HDCAM/XDCAM/HDV/AVCHD record with PAR 1.333 = (16/9) / (1440/1080) so that display AR defaults to square pixel 1920x1080?

[poisondeathray]

ah!!! not the "black border" theory again!!! (<---note this isn't directed at you), i thought once we had moved onto hi def video i would never have to hear about that silly idea again.

if the video stream is 1920 pixels wide by 1080 pixels high and the pixels themselves have an aspect ratio of 1:1 and both your hdtv and computer monitor use pixels with a ratio of 1:1 what "black borders" could possibly exist? where would you see these "black borders"?

a 1920x1080 movie has a display aspect ratio of 16:9 and a pixel aspect ratio of 1:1, in what way does it make sense to try and resize it to a 2.4:1 aspect ratio?

same thing with 1280x720, it also has a dar of 16:9 and a par of 1:1 so why resize it to an aspect ratio of 2.35:1?

i have even seen a 1440x1080 movie (4:3 dar and 1:1 par) resized to some bizarre resolution (i don't quite remember what it was but i remember being like WTF).

2.35:1 is the aspect ratio of many Hollywood movies. You can check on imdb.com

[jagabo]

2.35:1 is the aspect ratio of many Hollywood movies. You can check on imdb.com

In addition, the boundary between the movie image and the black borders causes two problems:

If the video is being resized (1920x1080 to 1280x720, for example) with a sharpening resizing filter (like Bicubic or Lanczos3) you will get bright row of pixels at the edge of the image. You can void this by using a Bilinear filter but then the picture won't be as sharp.

If the boundary between the image and picture doesn't fall on an 8 pixel boundary the video will require more bitrate to encode. The encoder won't be able to use vertical motion vectors on those boundaries.