Vegas incorrectly interpreting aspect ratio of .WMV file

[Melodious]

Crop the original black bars away. Then let Vegas add letterbox bars.

I don't think Vegas allows you to crop and then add custom bars. If I crop to 720x396 (taking off the original bars) and then let Vegas resize to 720x480, it will add bars but won't look as correct as..."cropping"(adding pixels) to 720x540 and letting Vegas resize THAT to 720x480. The video dimensions look different. The latter way is how hello_hello described it as giving you 4:3 worth of square pixels:

The assumption being if the wmv's current 3:2 aspect ratio is correct, adding 60 more pixels of black gives you 720x540, which is 4:3 worth of square pixels, which you'd then resize to 720x480 for a 4:3 DVD. If you wanted ITU resizing..... my brain won't do it today.... anyone?.

[poisondeathray]

If you want to do essentially the same thing in vegas (There will be slight differences because Vegas uses ITU resizing.)

Setup the project to NTSC DV 4:3, but set it to progressive

Interpret the file PAR as 1.0 (right click clip on timeline, properties)

Before rendering out , use a "Computer RGB to Studio RGB" filter, use a NTSC DVD Architect preset (4:3 , not widescreen)

The reason for the "Computer RGB to Studio RGB", is vegas uses studio RGB in 8bit mode, but most non native camera formats (like WMV that you are using) , get interpreted as computer RGB, not studio RGB .

[Melodious]

If you want to do essentially the same thing in vegas (There will be slight differences because Vegas uses ITU resizing.)

Setup the project to NTSC DV 4:3, but set it to progressive

Interpret the file PAR as 1.0 (right click clip on timeline, properties)

Before rendering out , use a "Computer RGB to Studio RGB" filter, use a NTSC DVD Architect preset (4:3 , not widescreen)

The reason for the "Computer RGB to Studio RGB", is vegas uses studio RGB in 8bit mode, but most non native camera formats (like WMV that you are using) , get interpreted as computer RGB, not studio RGB .

Progressive scan looks so much better. I was using lower field first, and that cause the picture to look bad. Thank you for that tip. The Computer to Studio RGB filter isn't necessary because Vegas seems to interpret the WMVs properly. When I add that filter, the blacks look very washed out.

I'm still unable to resize the video to 720x540 (then export to 720x480) without two-tone black bars. If I cut out the original bars by cropping video to 720x396 (then export to 720x480) the aspect ratio is not equal to the original video.

[poisondeathray]

Progressive scan looks so much better. I was using lower field first, and that cause the picture to look bad. Thank you for that tip. The Computer to Studio RGB filter isn't necessary because Vegas seems to interpret the WMVs properly. When I add that filter, the blacks look very washed out.

Whenever there is a mismatch between project settings, asset and/or render settings in terms of progressive vs. interlaced, vegas will apply the deinterlacing specificed in the project settings (default is blend, = blurry) . It also affects interlaced resizing . The quality settings also affect vegas' resizing methods. When set to "best" it will use bicubic, otherwise bilinear (softer)

I don't use vegas much anymore, but I don't think things have changed in the newer versions

How are you determining this? Are you looking at the exported MPEG2 file directly , and with what software ?

[Melodious]

Progressive scan looks so much better. I was using lower field first, and that cause the picture to look bad. Thank you for that tip. The Computer to Studio RGB filter isn't necessary because Vegas seems to interpret the WMVs properly. When I add that filter, the blacks look very washed out.

Whenever there is a mismatch between project settings, asset and/or render settings in terms of progressive vs. interlaced, vegas will apply the deinterlacing specificed in the project settings (default is blend, = blurry) . It also affects interlaced resizing . The quality settings also affect vegas' resizing methods. When set to "best" it will use bicubic, otherwise bilinear (softer)

I don't use vegas much anymore, but I don't think things have changed in the newer versions

How are you determining this? Are you looking at the exported MPEG2 file directly , and with what software ?

Actually, it was just a placebo effect. When I compare the two exported files (one progressive scan w/ no interlacing & one lower field first w/ blend interlacing) I don't notice a difference.

[poisondeathray]

Progressive scan looks so much better. I was using lower field first, and that cause the picture to look bad. Thank you for that tip. The Computer to Studio RGB filter isn't necessary because Vegas seems to interpret the WMVs properly. When I add that filter, the blacks look very washed out.

Whenever there is a mismatch between project settings, asset and/or render settings in terms of progressive vs. interlaced, vegas will apply the deinterlacing specificed in the project settings (default is blend, = blurry) . It also affects interlaced resizing . The quality settings also affect vegas' resizing methods. When set to "best" it will use bicubic, otherwise bilinear (softer)

I don't use vegas much anymore, but I don't think things have changed in the newer versions

How are you determining this? Are you looking at the exported MPEG2 file directly , and with what software ?

Actually, it was just a placebo effect. When I compare the two exported files (one progressive scan w/ no interlacing & one lower field first w/ blend interlacing) I don't notice a difference.

Thre shouldn't be a difference because the file is interpreted as progressive

The problem more commonly occurs when you have interlaced "flags" in progressive content. For example, "30p in 60i" content. This means progressive content gets deinterlaced, unless you manually interpret the file .

Are you still noticing a problem with the letterboxing ?

[Melodious]

No matter what I do, I can't get my video to look as good as hello_hello's export (and his is an 8MB mkv). My MPEG-2s look terrible compared to the original WMV and not as clear as hello_hello's. Vegas makes it look bad.

[Melodious]

Progressive scan looks so much better. I was using lower field first, and that cause the picture to look bad. Thank you for that tip. The Computer to Studio RGB filter isn't necessary because Vegas seems to interpret the WMVs properly. When I add that filter, the blacks look very washed out.

Whenever there is a mismatch between project settings, asset and/or render settings in terms of progressive vs. interlaced, vegas will apply the deinterlacing specificed in the project settings (default is blend, = blurry) . It also affects interlaced resizing . The quality settings also affect vegas' resizing methods. When set to "best" it will use bicubic, otherwise bilinear (softer)

I don't use vegas much anymore, but I don't think things have changed in the newer versions

How are you determining this? Are you looking at the exported MPEG2 file directly , and with what software ?

Actually, it was just a placebo effect. When I compare the two exported files (one progressive scan w/ no interlacing & one lower field first w/ blend interlacing) I don't notice a difference.

Thre shouldn't be a difference because the file is interpeted as progressive

The problem more commonly occurs when you have interlaced "flags" in progressive content. For example, "30p in 60i" content. This means progressive content gets deinterlaced .

Are you still noticing a problem with the letterboxing ?

Yes, even if I don't crop the picture (just leave it 720x480) and make the project NTSC 4:3 (while keeping the interpretation of the file at 1.0 square pixels), I still get two-tone black bars. Vegas bars are black plus the video's grey bars baked into the picture.

[poisondeathray]

Yes, even if I don't crop the picture (just leave it 720x480) and make the project NTSC 4:3 (while keeping the interpretation of the file at 1.0 square pixels), I still get two-tone black bars. Vegas bars are black plus the video's grey bars baked into the picture.

Again, how are you determining that ? What software and what file are you looking ? Because vegas can be tricky unless you have it setup properly with an external broadcast monitor . The small preview window isn't always accurate in vegas, depending on how you have it set up.

What version of vegas are you using ? I seriously doubt it has changed that much

[Melodious]

hello_hello's pretty Avisynth encode: https://forum.videohelp.com/attachments/21111-1383932784/720x480%204-3.mkv

My ugly Vegas encode: https://dl.dropboxusercontent.com/u/26626325/Melodious%20-%20Tom%204-3.mpg

[Melodious]

Yes, even if I don't crop the picture (just leave it 720x480) and make the project NTSC 4:3 (while keeping the interpretation of the file at 1.0 square pixels), I still get two-tone black bars. Vegas bars are black plus the video's grey bars baked into the picture.

Again, how are you determining that ? What software and what file are you looking ? Because vegas can be tricky unless you have it setup properly with an external broadcast monitor . The small preview window isn't always accurate in vegas, depending on how you have it set up.

What version of vegas are you using ? I seriously doubt it has changed that much

Viewing exported videos in VLC Player. Software: Vegas Pro 10.0a

[poisondeathray]

Apologies, you are correct . I've been looking at the wrong thing. Sorry for the confusion

I'll list some instructions for avisynth if you want to use hello_hello's approach for now (But be aware there are different methods of resizing and AR interpretation)

But I'll revisit this in vegas later (I think there should be a way to crop and retain the AR the way you want)

1) Install avisynth

2) Download ffms plugin package , unzip and place the .dll, and .avsi into the avisynth/plugins directory
http://code.google.com/p/ffmpegsource/

3) Open a text file with notepad write the following , save it, change the extension from .txt to .avs . You would change the PATH to whatever the path is, e.g. C:\folder etc... . And the filename to whatever the wmv file name is

FFMpegSource2("PATH\video.wmv", atrack=-1)
AddBorders(0,30,0,30)
Spline36Resize(720,480)
AssumeFPS(30000,1001)

If you didn't need audio, or want to import audio separately you can use FFVideoSource

FFVideoSource("PATH\video.wmv")
AddBorders(0,30,0,30)
Spline36Resize(720,480)
AssumeFPS(30000,1001)

4) You can encode that .avs with vdub for example. Open the .avs in vdub, video=>fast recompress, video=> compression . Audio will be set to direct stream copy by default which will be uncompressed audio. File=>save AVI

You can choose a compression format you feel is suitable, eg cineform is "visually lossless" and works well in vegas. "Lossless" compression formats include lagarith, ut video codec. (Technically they aren't "lossless" in vegas, because vegas treats them as RGB, not YUV)

You will need an .avs for each video file. There are batch avs script creators

[poisondeathray]

ok for vegas, did you have "source maintain aspect ratio: yes" for event pan/crop ? Works ok here , bars same color . Crop height was 396, so 42 on top & bottom each

[Melodious]

Apologies, you are correct . I've been looking at the wrong thing. Sorry for the confusion

I'll list some instructions for avisynth if you want to use hello_hello's approach for now (But be aware there are different methods of resizing and AR interpretation)

But I'll revisit this in vegas later (I think there should be a way to crop and retain the AR the way you want)

1) Install avisynth

2) Download ffms plugin package , unzip and place the .dll, and .avsi into the avisynth/plugins directory
http://code.google.com/p/ffmpegsource/

3) Open a text file with notepad write the following , save it, change the extension from .txt to .avs . You would change the PATH to whatever the path is, e.g. C:\folder etc... . And the filename to whatever the wmv file name is

FFMpegSource2("PATH\video.wmv", atrack=-1)
AddBorders(0,30,0,30)
Spline36Resize(720,480)
AssumeFPS(30000,1001)

If you didn't need audio, or want to import audio separately you can use FFVideoSource

FFVideoSource("PATH\video.wmv")
AddBorders(0,30,0,30)
Spline36Resize(720,480)
AssumeFPS(30000,1001)

4) You can encode that .avs with vdub for example. Open the .avs in vdub, video=>fast recompress, video=> compression . Audio will be set to direct stream copy by default which will be uncompressed audio. File=>save AVI

You can choose a compression format you feel is suitable, eg cineform is "visually lossless" and works well in vegas. "Lossless" compression formats include lagarith, ut video codec. (Technically they aren't "lossless" in vegas, because vegas treats them as RGB, not YUV)

You will need an .avs for each video file. There are batch avs script creators

I'm stuck on step 2. After unzipping, I see no .dll or .avsi files.

[poisondeathray]

Something wrong with your last file, it seems more blurry than it should be

View these at 1:1 and compare the text in the background and the various fine details like textures, wood grain etc.... Was that "Melodious - Tom 4-3.mpg" a direct export, or did you do something else to it ?

The circled area on the histogram (actually a Y' waveform) shows where the black level difference is . The black level difference isn't show in the screenshot (only picked up by the histogram), because of the standard Rec601 conversion for RGB

This is actually a B-frame on my encode, compared with a P-frame on yours . (P-frames are usually higher quality and allocated more bitrate). And the average bitrate for the entire file was lower on my version. Vegas9 export

[Melodious]

ok for vegas, did you have "source maintain aspect ratio: yes" for event pan/crop ? Works ok here , bars same color . Crop height was 396, so 42 on top & bottom each

I have it set to "Lock apect ratio: no". When cropped down, the bars are the same color. But you're cutting the title sequence area. Look at hello_hello's title sequence to speaker portion and notice no cropping was done at all.

[poisondeathray]

I'm stuck on step 2. After unzipping, I see no .dll or .avsi files.

Ok,that's not a compiled version that you downloaded

Try this one here
http://code.google.com/p/ffmpegsource/downloads/detail?name=ffms-2.17.7z

The one I'm using is here , with avisynth 2.6 alpha 5
http://forum.doom9.org/showthread.php?p=1650386#post1650386

[poisondeathray]

ok for vegas, did you have "source maintain aspect ratio: yes" for event pan/crop ? Works ok here , bars same color . Crop height was 396, so 42 on top & bottom each

I have it set to "Lock apect ratio: no". When cropped down, the bars are the same color. But you're cutting the title sequence area. Look at hello_hello's title sequence to speaker portion and notice no cropping was done at all.

Don't you want to process the intro vs . speaker section differently? Wasn't that what all the earlier discussion was about ?

So you want to emulate what that .avs script is doing in vegas ? The addborders (30px top & bottom) , and resize to 720x480 ? Because that will still distort the intro section

[Melodious]

Something wrong with your last file, it seems more blurry than it should be

View these at 1:1 and compare the text in the background and the various fine details like textures, wood grain etc.... Was that "Melodious - Tom 4-3.mpg" a direct export, or did you do something else to it ?

The circled area on the histogram (actually a Y' waveform) shows where the black level difference is . The black level difference isn't show in the screenshot (only picked up by the histogram), because of the standard Rec601 conversion for RGB

This is actually a B-frame on my encode, compared with a P-frame on yours . (P-frames are usually higher quality and allocated more bitrate). And the average bitrate for the entire file was lower on my version. Vegas9 export

I see the difference. Mine is much blurrier. And yes, it is a direct export from Vegas. Here were all the settings for my export:

[Melodious]

ok for vegas, did you have "source maintain aspect ratio: yes" for event pan/crop ? Works ok here , bars same color . Crop height was 396, so 42 on top & bottom each

I have it set to "Lock apect ratio: no". When cropped down, the bars are the same color. But you're cutting the title sequence area. Look at hello_hello's title sequence to speaker portion and notice no cropping was done at all.

Don't you want to process the intro vs . speaker section differently? Wasn't that what all the earlier discussion was about ?

So you want to emulate what that .avs script is doing in vegas ? The addborders (30px top & bottom) , and resize to 720x480 ? Because that will still distort the intro section

I'd rather not waste my time doing them separately. The truth is they do not care about that. In previous years, they sent me videos with badly stretched logos and I processed them as is, and they made no complaint or care. I bet I could even get away with cropping them all for the sake of not having to cut and crossfade each intro. I know that doesn't sound professional, but they aren't going for professional clearly by the standard of the videos I was provided with. All that matters is getting the speaker segments of the videos to look as close as possible to source.

I tried adding 30px top & bottom to make it 720x540, but hello's video is still slightly wider and not as tall.

Here's how it looks out of Vegas with the emulated hello_hello adjustments:

https://dl.dropboxusercontent.com/u/26626325/Melodious%20-%20Tom%204-3%20added%2060px.mpg

[manono]

The truth is they do not care about that.

Maybe not, but you should care and take pride in the work you do. I all the time do extra for people that won't know, understand, or appreciate what's been done for them. And maybe sometime you'll lose some possible business when a prospective client/customer sees the sloppy job you did with this one.

I'd rather not waste my time doing them separately.

If using an AviSynth script it's a simple matter to handle them separately in the same script and the same encode.

[poisondeathray]

I tried adding 30px top & bottom to make it 720x540, but hello's video is still slightly wider and not as tall.

Vegas uses ITU interpretation for all AR calculations, so it will NEVER be the same as using non ITU methods when encoding directly to non square pixel formats such as DVD. But if you export to a square pixel format, then resize and encode with a different encoder, then it will exactly the same in terms of letterboxing as the avisynth method

So if you want exactly the same AR interpretation as non-ITU in the avs script above, one workaround in vegas is to use masks and overlay the letterbox bars (e.g. using media generator, or even imported still from a photo editor) on a 720x540 square pixel project timeline, export a square pixel format, do the final resize and encode with something else like hcenc. This way it will be identical to the non ITU resizing based on the avs script , and the bars will be the same uniform color





For the quality difference, one difference is I only encoded a small loop region, not the entire video. But it shouldn't be that big of a difference, I think something else is going on





You're making a big deal about "not wasting time", but isn't making a bunch of scripts and encoding to some intermediate going to take some time ?

What kind of things/edits/manipulations are you doing in vegas ? For some projects it might be easier to do in avisynth alone (what do you specifically need vegas for on this project ? )





So this is for a carpet company...and pictures of Tom Jenning's shiny head plastered all over this thread... why aren't there any "rug" jokes yet ?? [/joke]

[Melodious]

Apologies, you are correct . I've been looking at the wrong thing. Sorry for the confusion

I'll list some instructions for avisynth if you want to use hello_hello's approach for now (But be aware there are different methods of resizing and AR interpretation)

But I'll revisit this in vegas later (I think there should be a way to crop and retain the AR the way you want)

1) Install avisynth

2) Download ffms plugin package , unzip and place the .dll, and .avsi into the avisynth/plugins directory
http://code.google.com/p/ffmpegsource/

3) Open a text file with notepad write the following , save it, change the extension from .txt to .avs . You would change the PATH to whatever the path is, e.g. C:\folder etc... . And the filename to whatever the wmv file name is

FFMpegSource2("PATH\video.wmv", atrack=-1)
AddBorders(0,30,0,30)
Spline36Resize(720,480)
AssumeFPS(30000,1001)

If you didn't need audio, or want to import audio separately you can use FFVideoSource

FFVideoSource("PATH\video.wmv")
AddBorders(0,30,0,30)
Spline36Resize(720,480)
AssumeFPS(30000,1001)

4) You can encode that .avs with vdub for example. Open the .avs in vdub, video=>fast recompress, video=> compression . Audio will be set to direct stream copy by default which will be uncompressed audio. File=>save AVI

You can choose a compression format you feel is suitable, eg cineform is "visually lossless" and works well in vegas. "Lossless" compression formats include lagarith, ut video codec. (Technically they aren't "lossless" in vegas, because vegas treats them as RGB, not YUV)

You will need an .avs for each video file. There are batch avs script creators

Outstanding step by step guide. I just processed one of the videos using the lagarith compression format, and it looks amazing. I just dropped the avi that I made using vdub and it looks crystal clear in the preview panel. I compare it with the WMV next to it in the timeline and it's a night and day difference. The color of the avi looked very saturated, so I used that Computer RGB to Studio RGB filter, and that brought it back to source.

The only thing that's bugging me is it looks horizontally stretched just a tad. Or maybe that's how the source looks. Tell me what you think. This is a Vegas MPEG-2 render of the VirtualDub AVI:

https://dl.dropboxusercontent.com/u/26626325/Melodious%20-%20Tom%204-3%20Avisynth.mpg

[poisondeathray]

The only thing that's bugging me is it looks horizontally stretched just a tad. Or maybe that's how the source looks. Tell me what you think. This is a Vegas MPEG-2 render of the VirtualDub AVI:

https://dl.dropboxusercontent.com/u/26626325/Melodious%20-%20Tom%204-3%20Avisynth.mpg

I haven't looked at this yet, but remember vegas uses ITU guidelines for resizing and AR. So if you export a MPEG2 for DVD directly, it won't necessarily look the same as encoding that script directly with something like hcenc , cce or some other encoder that accepts avs scripts natively

Also, when importing intermediate AVI files into vegas, you have to go through the same hoops with AR interpreting

[sanlyn]

The video I posted has little or no distortion in the logos or title cards. That's because I re-made new titles from shorter cuts of the titles, with the short cuts rescaled and joined to form enough frames to place a dissolve where the originals were. Then tied the audio to it later. If you're thinking of a single resize to get all the shots in the same AR, it won't work. It's a case of the sadistic creator tying together titles and video chunks of different AR and not giving a damn about the results. I'm with manono: this is just sloppy work. If I were handed stuff like this for my business, I'd refuse to pay for it. It doesn't take hard work to use an expensive NLE to screw up video.

[Melodious]

The truth is they do not care about that.

Maybe not, but you should care and take pride in the work you do. I all the time do extra for people that won't know, understand, or appreciate what's been done for them. And maybe sometime you'll lose some possible business when a prospective client/customer sees the sloppy job you did with this one.

I'd rather not waste my time doing them separately.

If using an AviSynth script it's a simple matter to handle them separately in the same script and the same encode.

You're right. I should care. I just go into panic mode when I run into problems. I have very little knowledge about video editing. This isn't my day job nor is it what I'm going to school for. It's hardly even a hobby. This project is for some side cash. I'm just doing what they ask for, without putting my blood sweat and tears into something I have such very little time to work on. The priority now is maintaining a picture that looks identical to source. I'm devoting my time to getting the high priority job done first. The little subjective details can sit on the back-burner.

[Melodious]

[QUOTE=poisondeathray;2281893]

What kind of things/edits/manipulations are you doing in vegas ? For some projects it might be easier to do in avisynth alone (what do you specifically need vegas for on this project ? )

One adjustment I'm making is using a color corrector to take the "red" out of Tom's face. The person in charge of the project specifically requested that. I had to do this for last year's DVD and she was happy with the results. I'm also inserting chapter markers into the timeline and titling them so that when the MPEG-2 is inserted into DVD Architect, the chapters will be prearranged. Does DVD Architect use a bad recompressor? It's the only DVD authoring program I have and know how to use. This is my menu screen:

[Melodious]

The only thing that's bugging me is it looks horizontally stretched just a tad. Or maybe that's how the source looks. Tell me what you think. This is a Vegas MPEG-2 render of the VirtualDub AVI:

https://dl.dropboxusercontent.com/u/26626325/Melodious%20-%20Tom%204-3%20Avisynth.mpg

I haven't looked at this yet, but remember vegas uses ITU guidelines for resizing and AR. So if you export a MPEG2 for DVD directly, it won't necessarily look the same as encoding that script directly with something like hcenc , cce or some other encoder that accepts avs scripts natively

Also, when importing intermediate AVI files into vegas, you have to go through the same hoops with AR interpreting

I don't know man. Here's a side by side of the original video played in its original size (left) and the vegas mpeg-2 render of the avisynth avi (right).



Looks about the same to me.

[Melodious]

The video I posted has little or no distortion in the logos or title cards. That's because I re-made new titles from shorter cuts of the titles, with the short cuts rescaled and joined to form enough frames to place a dissolve where the originals were. Then tied the audio to it later. If you're thinking of a single resize to get all the shots in the same AR, it won't work. It's a case of the sadistic creator tying together titles and video chunks of different AR and not giving a damn about the results. I'm with manono: this is just sloppy work. If I were handed stuff like this for my business, I'd refuse to pay for it. It doesn't take hard work to use an expensive NLE to screw up video.

That's just an insane amount of extra leg work considering there are 22 videos in total. Maybe if they gave me another grand. I don't know.

[budwzr]

89 posts still have failed to solve the problem? Is editing an SD Mpeg really that complicated?