hi, I read the relevant posts on this forum, but I still can't do it properly: when play on widescreen portable dvd player or on my PC (1920x1080), I get the picture that has small black bars on all sides (top/bottom and left/right). Without changing the the aspect ratio on the display, I want to have no black bars, or in one direction only.

My workflow:
avchd (USA Canon Vixia HF200) -> Movie Studio 10 -> avi (1920x1080) ->VirtualDub (resize={Lanczos3, new size=720x405, aspect ratio=same as source, 'do not letterbox', not interlaced} -> avi,
->DVDarchitect5 -> dvd

any advice on the settings I should use instead?
thanks

Shouldn't your aspect ratio be 720x480? 1.212 pixel aspect ratio?

As Jeff pointed out, 720x480 with a PAR of 1.2121 is the general setting for widescreen DV. I wonder if the players just look for that resolution and only resize it appropriately when they see that?

And, for the record, square pixels widescreen DV would be 396 (480/1.2121).

Hope this helps,

thanks guys, I had a hard time entering the aspect ratio as you suggested in Virtual Dub, and I did something like this:

new size = 854 x 480, aspect ratio = 'compute height from ratio = 854:480', framing options='do not letterbox'


but my real problem was the wrong template in DVD Architect: I used:
MPEG-2 720x480-60i, 4:3 (NTSC)
instead of:
MPEG-2 720x480-60i, 16:9 (NTSC)

Now the frame nicely fills the screen.

The avi from VirtualDub looks almost as good as FullHD !
The dvd looks good, it fills the screen almost exactly, but not exactly (playing in PowerDVD8). It feels like it doesn't display in the native resolution of the monitor. I need to work on this...

the problem with converting to avi before going into architect is architect is reconverting to mpeg 2, which I wouldn't like.

Try this just to see how the results work...put the original video into vegas, render to mpeg 2 using the dvd architect (widescreen 720x480 ) and see what happens.

It will then fit properly and will have only been converted once, and the aspect ration will be exactly correct.

the problem with converting to avi before going into architect is architect is reconverting to mpeg 2, which I wouldn't like.

Try this just to see how the results work...put the original video into vegas, render to mpeg 2 using the dvd architect (widescreen 720x480 ) and see what happens.

It will then fit properly and will have only been converted once, and the aspect ration will be exactly correct.

Actually, a quality loss is inevitable no matter what, in this case. This is because the MPEG-2 encoder itself does a very poor job of downsizing, especially on interlaced video. In the OP's case, not only does the video have to be converted (format-wise), but also downsized. And Vegas' MPEG-2 encoder does not use anywhere close to the most appropriate downsizing method (Vegas alone does not offer Lanczos3 resizing at all whatsoever - it offers only bilinear downsizing, which results in major artifacts when resizing pixel aspect ratios). And downsizing the vertical resolution to less than 480 pixels will always result in a loss of image quality no matter what simply because once you downsize the video's vertical resolution to only 405 or 360 pixels, you'll never regain those lost pixels of resolution. As a result, AVCHD downconverted to widescreen SD using Vegas alone will look worse than if that same video is shot in LD (Low Definition) mode (320x240p; there is no 240i).

And the problem here is not the downsizing per se, but the fact that standard-definition video almost always uses non-square pixel aspect ratios while most high-definition video uses square pixel aspect ratios.

thanks guys, I realize downscaling from Movie Studio doesn't work great, and I use Lanczos.
So far the best results I get when:
downscale in VirtualDub,
convert to mpeg (8000Mbps) in Movie Studio,
burn in DVD Architect.

The frame fills the screen nicely now. Quality is not bad, although not as good as SD avi.

I haven't figured out how to avoid resampling in DVD Architect.

I realize downscaling from Movie Studio doesn't work great, and I use Lanczos.

The way you said it I get the impression you are assuming that if Lanczos does not work for you, no other method would. I know this is a common myth on this form that "Lanczos is the best." The reality of resizing is that no method is the best for all cases. Each method has its advantages and disadvantages. Lanczos tends to get good results when resizing photographed images. It is quite unsuitable for resizing text, for example. It is not the best for drawings, such as traditional 2D cartoons and graphs.

If Lanczos does not work for you, try a different method.

Somebody should really do a sticky about resizing Hd to Sd. It comes up about every week, it just shouldn't cause this much grief.
I know there are lots of workflows, but the simple one that never fails, and seems to be used by a lot of people a lot more clever than me, always works.....

1. Edit in Hd, export to square pixels (1440x1080 1.33 aspect pixels become 1920x1080 square pixels) full Hd in a near lossless codec of your choice.
2. Drop it into Virtual dub, apply the Resize filter to SQUARE PIXEL Sd, This seems to be where it goes wrong for a lot of people-
PAL Sd Widescreen is 1048x576 which equals 720 (which is horizontal width) x 1.4568 (which is PAL widescreen pixel aspect ratio).
So,
NTSC Sd widescreen would be 872x480 which equals 720 (which is horizontal width) x 1.2121 (which is NTSC widescreen pixel aspect ratio).

Then export from Virtualdub to Lagarith lossless codec, and then bring it into you DVD authoring program of choice.
All the DVD programs Ive used know exactly what to do with the square pixel files and you shouldn't have to play with aspect ratios at all, just choose PAL or NTSC and they get it right straight away, well this goes for Architect, Adobe Encore, and my personal fave, Nero Vision.

thanks Gerald, looks like the frame size might be off in my workflow: somewhere on the internet I found the wide-screen to be 854x480, while you're saying 872x480. I'll check it out.
And Adam, maybe I didn't make it clear, but I'm happy with Lanczos in VirtualDub. It's the conversion of SD avi to mpeg2 that appeared to be the weak link.

872X480 is not the same aspect ratio as 1920x480.
Do you recommend 'letterbox/crop' or 'do not letterbox or crop'? The % difference is small.

well, although the frame fits the screen nicely, I see a huge difference between the quality of my dvd and sd avi. Maybe it's normal, but I'm disappointed. I set 8MBps encoding in DVD Architect, and I feed it with uncompressed sd avi (872x480) now.
To be precise, my dvd video looks good when shown in its native resolution, but when I make it full screen, it becomes blurred way more than sd avi.

The way you said it I get the impression you are assuming that if Lanczos does not work for you, no other method would. I know this is a common myth on this form that "Lanczos is the best." The reality of resizing is that no method is the best for all cases. Each method has its advantages and disadvantages. Lanczos tends to get good results when resizing photographed images. It is quite unsuitable for resizing text, for example. It is not the best for drawings, such as traditional 2D cartoons and graphs.

If Lanczos does not work for you, try a different method.

And not only that, but there are many methods that deliver poor results on everything. I have shot scenes that when I used a resizing method that does well on text it ended up ruining the rest of the image with severe artifacts, while using a method that does well on the image made the text unreadable. And worse, an in-between method just rendered both image and text unwatchable.

If those images are what you typically work with, then maybe it's best to shoot in the format that exactly matches the capability of the end user media (Blu-ray, DVD-Video). And if one must downconvert (downsize) such tricky scenes, then maybe it would be better converting from digital HD to analog SD then back to digital SD (using the analog composite outs of the HD camcorder connected to the analog composite inputs of the digital SD capture device)?

Stanley, your are using an odd size, it cannot look right. 720x480 is not a suggestion, it is the dimensions it needs to be to play properly on a TV.

You avoid recompressing by rendering to a format that DVDA can use, which is mpeg 2. Please read DVDA help for more information, this is explained clearly there, then come back and let us know what is happening.

Somebody should really do a sticky about resizing Hd to Sd. It comes up about every week, it just shouldn't cause this much grief.
I know there are lots of workflows, but the simple one that never fails, and seems to be used by a lot of people a lot more clever than me, always works.....

1. Edit in Hd, export to square pixels (1440x1080 1.33 aspect pixels become 1920x1080 square pixels) full Hd in a near lossless codec of your choice.
2. Drop it into Virtual dub, apply the Resize filter to SQUARE PIXEL Sd, This seems to be where it goes wrong for a lot of people-
PAL Sd Widescreen is 1048x576 which equals 720 (which is horizontal width) x 1.4568 (which is PAL widescreen pixel aspect ratio).
So,
NTSC Sd widescreen would be 872x480 which equals 720 (which is horizontal width) x 1.2121 (which is NTSC widescreen pixel aspect ratio).

Then export from Virtualdub to Lagarith lossless codec, and then bring it into you DVD authoring program of choice.
All the DVD programs Ive used know exactly what to do with the square pixel files and you shouldn't have to play with aspect ratios at all, just choose PAL or NTSC and they get it right straight away, well this goes for Architect, Adobe Encore, and my personal fave, Nero Vision.

DVD Architect will re-compress that video. So if you don't like the compression algorithm it uses, you are out of luck quality wise.

John

Stanley, your are using an odd size, it cannot look right. 720x480 is not a suggestion, it is the dimensions it needs to be to play properly on a TV.

You avoid recompressing by rendering to a format that DVDA can use, which is mpeg 2. Please read DVDA help for more information, this is explained clearly there, then come back and let us know what is happening.

Therein lies another problem:

VirtualDub does not preserve pixel aspect ratios when resizing. Instead, it converts everything to square pixels. As a result, the resulting 720x480 downconversion is detected by Vegas as a squeezed 4:3 aspect ratio, which is far from widescreen. Vegas (and all other NLEs) must get tricked to see that video as a widescreen 16:9 video. And Vegas detects that video as LFF even though the actual field order is UFF. (Or, if you're using Adobe Premiere, that VirtualDub 720x480 video is detected as progressive-scan video even though it's actually interlaced.)

DVD Architect will re-compress that video. So if you don't like the compression algorithm it uses, you are out of luck quality wise.

John

Yes, but Lagarith is a visually lossless codec, so you are carrying the best quality you can all the way through to the final mpeg2 compression.
You can't want for any better than that unless you just burn a DVD data disc with uncompressed AVI. You wont fit much on though :)

After going through the hd sd thing, and once again reading all the prior posts in this thread,

Boy am I glad i made the decision to only produce high definition.

If some one wants me to do something in Sd I shoot them on my not so old cameras!!!

with blu ray players costing all of 70 dollars now, why bother if you have HD!!

I will also charge more for DVD's than I do for blu rays!!!

Stanley, your are using an odd size, it cannot look right. 720x480 is not a suggestion, it is the dimensions it needs to be to play properly on a TV.

You avoid recompressing by rendering to a format that DVDA can use, which is mpeg 2. Please read DVDA help for more information, this is explained clearly there, then come back and let us know what is happening.
I looked into the help file, but I didn't find any obvious mistake I was making.
The bottom line for me is that both mpeg2 and avi standard definition look great when viewed in the native resolution, i.e. in a section of the full screen 1920x1080. Unfortunately only sd avi looks good in full screen mode. I'm not sure why only avi? Is it something with the upscaling algorithm?
Thanks.

type in mpeg 2 and you'll find the specs

I made the following "quick & dirty" tutorial to prepare AVCHD 1920x1080 60i for DVD Architect. Most of the time it works great for me. I the resultant SD mpg contains excessive moire or flickering, an intermediate render may be required.

YouTube - Vegas Pro 9.0e Quick & Dirty AVCHD to SD Tutorial (http://www.youtube.com/watch?v=hbcgu0KTPGU&hd=1)

Good Luck!
...Jerry

PS: New to this forum - still need to learn how to embed video clips.

Jerry, FYI: when you start reading previous posts about this issue, you'll see that Vegas does NOT do a great job in converting to SD. Lanczos algorithm is usually preferred, and this may be done in VirtualDub.
Good luck.

Jerry, FYI: when you start reading previous posts about this issue, you'll see that Vegas does NOT do a great job in converting to SD. Lanczos algorithm is usually preferred, and this may be done in VirtualDub.
Good luck.

Stanley, I agree 1000%. And I've read much about this subject and done hours & hours of trial runs.

The procedure I posted in my tutorial is labeled "Quick and Dirty" because that's what it is. I've found that for the most part, this procedure creates "pretty darn good" quality SD for DVDA prep - obviously not perfect. I would like to get feedback (often times 'cuz it works for me, it may not work for others).

Sometimes, using the procedure listed above, the output quality will just be lame, many times evidenced in excessive moire and/or flicker. In those cases, more stringent methods must be taken - for example intermediate renders & downsizing using other tools.

...Jerry

PS: One slight error in the tut - I mentioned Project settings should be set to "Interpolate for fast motion, Blend for slow motion" - I've since found out that advice is only true for Progressive renders. The important thing is Deinterlace is NOT set to "none".

I'm trying to downsize a load of m2t files from a Sony Z5 but these don't work in VirtualDub. Is my best bet to downsize using the camera's built in facility (and if so should I crop, squeeze or letterbox?!).

I am mixing these files on the timeline with widescreen SD PAL DV. Final output is to SD DVD and I was hoping to be able to zoom in on the HDV footage, which I guess I will not be able to do if I downconvert. Any other workflow recommended for m2t?

The footage is of people doing exercises on trampolines and it in tests it looks really dreadful on DVD. Curiously, I rendered the timeline using the MainConcept DVD Architect PAL Widescreen video stream preset, but DVDA wanted to recompress the video again. Did I do something wrong?

I have shot HD with the FX1000 and rendered out with Vegas to SD with excellent results. Downsizing via the camera does not produce great results. I believe, and some others do to, that the best results come from resizing in post. I believe that my camera and yours shoot extremely nice SD widescreen, and I shoot almost exclusively SD. The results are fantastic. I have been asked if some of my work was HD by potential clients. (It is doesn't look HD to me, but hey, I'm just making a point).

I usually use 720x480 Widescreen project settings in Vegas when converting HD, rather then using the native HD dimensions, but I've done it both ways, and see little difference.

As an aside, I see people shooting HD and downconverting a lot here. I know sometimes you must, but I just don't. Some cameras reputedly do not shoot great SD (the XH-A1 comes to mind), but my cameras do. The Canon HV30 does well also.

I think the key is the quality of the footage to begin with. Some folks (not you Ian) shoot HD expecting it to be a magic bullet and to produce "amazing" footage (just because it's HD). It does not. Poorly shot/lit HD footage looks as bad as poorly lit SD footage. Properly framed shots in sufficient light, the use of closeups obtained by moving the camera rather than by zooming and knowledge of basic shooting techniques are what create great video. And don't forget to lower the gain! Great quality footage will be very forgiving in post.

Some scenarios, particularly dance recitals, plays, speeches in poorly lit situations are often no-win when using a 1/3" chip camera. But weddings and for almost all other corporate work I find SD with these Sony cameras produces images that is excellent.

Oops, sorry Ian...I went on my rant and forgot to address your question.

I'm thinking you might bring your footage into Vegas using the project setting that match your HD footage...then you can zoom in as needed. Adjust the other footage with the pan and crop tool, or using the aspect ratio plug in...I think then you'd be good to go. Render out to desired output, and it will look fine.

Hi Jeff - it was a good rant - I enjoyed it!

Since my post I have been doing a few unscientific experiments. I certainly agree that downsizing with the camera is not of an acceptable quality so that idea is canned. I'm actually happy about that as it means I can resize the 1440x1080 footage on the timeline (note that I only want to introduce a very slow crawl in to subtly emphasize certain things).

At the moment I am working with project settings matching the m2t footage. The PAL SD footage actually doesn't look too bad. I'm currently rendering out to AVI using the 1440*1080 Sony YUV preset. I will then see what's best for downsizing for the final SD DVD and I am guessing that might be VDub with Lanczos.

Thanks for the advice.

For DVD try rendering out to your normal SD DVD settings for MPEG 2. You might find the results surprising, unless you've tried it before.

For DVD try rendering out to your normal SD DVD settings for MPEG 2. You might find the results surprising, unless you've tried it before.

This usually works pretty well for me. Per the tutorial I posted earlier, here are some suggested "tweaks":

1) Make sure your project properties has a "Deinterlace Method = Interpolate" (Blend works fine, too) as Vegas uses this to resize when rendering (even though we are not actually Deinterlacing). Also, Rendering Quality = Best.
2) Add a Sharpen FX set to 0.000 to each clip in the timeline. This is actually a bug in the Sharpen FX, but improves the quality of the final render.
3) Right-click each event in the timeline and set "Properties->Reduce Interlace Flicker" If the final render still contains moire patterns or excessive flicker, go to Plan "B" - intermediate renders &/or resizing or addng Gaussian Blur. However, normally Plan "B" is not needed.
4) Render to the highest bitrate you can fit on your DVD - you may need a bitrate calculator to determine the actual settings.

Good Luck!
...Jerry

I work with Edius 6 .
And i use also the VDub with Lanczos for downscale to SD AVI (lossless)
When i put the SD AVI in a SD-Projekt in the Timeline, I have to change the settings of the clip to 16:9
After VDub the Clip will be identified as a 4:3 Ratio. But that is just a click to do in Edius.

Before the converting to mpeg it is necessary to use the "sharpen" effect. In Edius i use it up to nr.10.
Without doing this, the DVD out of VDub looks very fuzzy.

Do you still do a separate AC3 audio render in this method?

Do you still do a separate AC3 audio render in this method?

Not sure whether you're replying to my post, but the general rule is if you are rendering to mpeg-2 in preparation for DVDA, you should render w/o audio and then render a separate ac3 with the same primary name (and in the same folder).

It's been a while since I've tried this, but I'm pretty sure you can render an mpeg-2 with audio, and import it to DVDA. But then DVDA will re-render if for you - not very efficent.

...Jerry

Not sure whether you're replying to my post, but the general rule is if you are rendering to mpeg-2 in preparation for DVDA, you should render w/o audio and then render a separate ac3 with the same primary name (and in the same folder).

It's been a while since I've tried this, but I'm pretty sure you can render an mpeg-2 with audio, and import it to DVDA. But then DVDA will re-render if for you - not very efficent.

...Jerry

Hi Jerry, actually I've always used your quick and dirty method because, welll, it's quick and dirty. But I'd like to raise the bar this time because I have a big presentation and want it to look as good as possible. I was referring to this post by Gerald Web, using a method I believe Peronne initially came up with. Here:

Somebody should really do a sticky about resizing Hd to Sd. It comes up about every week, it just shouldn't cause this much grief.
I know there are lots of workflows, but the simple one that never fails, and seems to be used by a lot of people a lot more clever than me, always works.....

1. Edit in Hd, export to square pixels (1440x1080 1.33 aspect pixels become 1920x1080 square pixels) full Hd in a near lossless codec of your choice.
2. Drop it into Virtual dub, apply the Resize filter to SQUARE PIXEL Sd, This seems to be where it goes wrong for a lot of people-
PAL Sd Widescreen is 1048x576 which equals 720 (which is horizontal width) x 1.4568 (which is PAL widescreen pixel aspect ratio).
So,
NTSC Sd widescreen would be 872x480 which equals 720 (which is horizontal width) x 1.2121 (which is NTSC widescreen pixel aspect ratio).

Then export from Virtualdub to Lagarith lossless codec, and then bring it into you DVD authoring program of choice.
All the DVD programs Ive used know exactly what to do with the square pixel files and you shouldn't have to play with aspect ratios at all, just choose PAL or NTSC and they get it right straight away, well this goes for Architect, Adobe Encore, and my personal fave, Nero Vision.