John, since the video is being downrezzed from 1440x1080 to 720x480, will there be an advantage to redering using the "best" setting rather than "good" setting? (I understand "best" increases rendering time)

Make sure you do the following:

1/ select the widescreen version of the SD DVD mpeg2 template.
2/ make sure you use "best" rather than "good" since there is resizing.
4/ make sure you have a deinterlace method selected. It doesn't matter what it is, but you need to select one in order to avoid squiggly edges during movement.

make sure you have a deinterlace method selected. It doesn't matter what it is, but you need to select one in order to avoid squiggly edges during movement.Laurence, are you sure? That's not my workflow working with interlaced footage. Wouldn't this depend on the source video and/or if you have processed a de-shaker script etc.
For normal output, using a Sony HC3 cam I simply leave everything in tack.

As far as I can tell, unless you render the video as (custom) progressive the de-interlace setting appears to me to be ignored. So if you select a standard template your going to have fielded video either way.
But for a quick dvd I put my HC3 into downconvert mode and record straight into a Sony dvd recorder via the firewire connection. My videos don't look jagged. they look like standard interlaced video. I admit they do look a little better if the dvd player converts the dvd to progressive display mode on the hdtv monitor.

I've always been under the impression that Vegas will do its resizing on each individual field and then re-interlace the results.

I shoot 16:9 HDV and deliver 4:3 SD DVDs all the time. Even here in "wealthy" Carmel and also in Pebble Beach (which actually IS wealthy), most people still watch 4:3 SD, although that is finally starting to change.

I posted my technique for doing this a long time ago in this post:

My workflow for HDV to SD projects

John, thanks for the workflow, great tips in there. Would you be willing to email me the mask you referenced? I would love to use it in my workflow.

Thanks.

"I've always been under the impression that Vegas will do its resizing on each individual field and then re-interlace the results."

Yeah, but you have to select a deinterlace method for it to do it this way. If you don't select a deinterlace method, it will resize the comb edges and resized interlace comb is the squiggle that people complain about. If you select a deinterlace method, and resize interlaced footage from one size to another, it will separate the fields, resize the fields separately, then fold them back together at the new size. It doesn't make any sense that Vegas would care about the deinterlace tab, but numerous experiments have shown me that it does. Don't worry about which deinterlace method you choose. It doesn't actually deinterlace so it doesn't matter.

Now some people will say that the picture seems to look sharper if you don't choose a deinterlace method. It will on static parts where there is no movement. The reason is that the static parts of interlaced footage are just like progressive footage and the two fields in this case show more detail. The problem is that as soon as there is any motion, the resized comb is going to look like squiggly edges and much worse than the extra clarity of the static parts, so it is by far the better option.

If I was designing Vegas, I wouldn't have done it this way. I would have Vegas make up it's mind about whether to separate odd and even fields before resizing based upon what type (progressive or interlaced) of field pattern was selected. That is not how Vegas works though. It decides based upon the tab you select in the deinterlace method. It shouldn't be this way but it is.

Conversely, if you are resizing progressive footage, make sure that the "select deinterlace method" tab is unchecked. If it is it will split your video into fields before the resize and you will lose half your resolution. Don't take my word for it. Try it out for yourself and you'll see.

"Yeah, but you have to select a deinterlace method for it to do it this way. If you don't select a deinterlace method, it will resize the comb edges and resized interlace comb is the squiggle that people complain about. If you select a deinterlace method, and resize interlaced footage from one size to another, it will separate the fields, resize the fields separately, then fold them back together at the new size."

Hmmm. i can't make my brain wrap around that. A field is not interlaced, by definition. One single field is progressive. It resizes the same way that a full progressive frame would. There's no need to deinterlace because there is nothing to deinterlace.

Resize each field individually and you're resizing two progressive images. "Shuffle" them back together afterwards and you have essentially the same thing you started with, only smaller. This shouldn't have any effect on combing or any other interlaced phenomenon at all.

When or if I do de-interlace and select (custom) progressive as the fielding I definitely notice a difference in the encoded video between none, blend & interpolate. If I do have to make a progressive encode I usually select interpolate which seems to give the best results. Usually my settings are "None", because I want to retain the fielding information.

I have heard of many that do have to de-interlace their videos when converting from HDV->SD.
Could be the source videos are different from mine or the source videos have already been altered from their original encoding. I don't know, but haven't had this problem using my Sony HC3 and always leaving the de-interlace method to "None", both for the HDV re-encodes and down conversions. Normally my project settings for de-interlace are none.

When I first got Vegas I recaptured video that was shot from the dashboard of a truck, tons of movement and highway lines, many fast turns. My project settings for de-interlace were "none" and that's how I exported to another HDV file or back to tape (smartrender off). Then I created a SD 720x480 16:9 video @ 9000kbs also with de-interlace = "none". Burnt a dvd in DVDA which wasn't re-encoded. The video played back on a standard 36" 4:3 TV interlaced had no jaggies where the highway lines or panning occurred. They were there, to a keen eye, so small you had to single step through each frame to notice the jaggie. At full frame rate playback you can't see them. Because those jazzies were so small and tightly woven was what made me decide to always shoot in the HDV mode. Vegas did an excellent conversion to SD (interlaced).

I think it may also depend a lot on the source video, settings in the cam and all.
Lucky for me I don't have this interlacing problem with my conversions, at least not yet.

I've found if you don't de-interlace correctly the video can cause slight nausea to the viewers.

Next time I make a SD video I use Laurence's suggestion. As I see it the video should come out the same either way, just as long as I don't change the fielding to progressive and keep the video as interlaced.

"Hmmm. i can't make my brain wrap around that. A field is not interlaced, by definition. One single field is progressive. It resizes the same way that a full progressive frame would. There's no need to deinterlace because there is nothing to deinterlace."

I may be using some incorrect lingo. Here's what I mean:

Check the "deinterlace method box" and Vegas will separate the images shown on the even and odd lines into two separate images, resize each of them, then recreate a new image at the new size where the odd lines all come from one resized image and the even lines all come from the other.

If you don't check the "deinterlace method box" Vegas will resize both the even and odd lines together as if it was one single image.

The above two statements happen regardless of whether the footage is interlaced or progressive. If I was designing the software, I would resize the even and odd lines as separate images any time the video was defined as interlaced and resize the image as one any time the video was defined as progressive. That is not how Vegas works however. It should work that way but it doesn't. This isn't a big deal as long as you know what is going on, but it does confuse a lot of people.

When I first discovered this, I was totally confused. Sometimes video resized properly, sometimes it didn't and I had no idea why. At the time I wasn't going from HDV to SD. I was resizing 4:3 to 16:9 by cropping the top and bottom 60 lines than stretching the remaining 360 lines to 480. When I did this, sometimes it would look exactly like it was supposed to and sometimes there would be horrible squiggly lines on any parts of the video where there was sideways motion.

At one point I submitted a bug report to Sony, then sent another report saying it was fixed, then another saying it was broken again. I was just totally confused. Anyway, I finally figured out what was going on: that Vegas determines it's resize method based upon whether or not the "select a deinterlace method" tab is checked and not upon whether or not the video is actually interlaced or not.

To test this out and see for yourself, just find a little piece of interlaced HDV video with a quick pan in it and render just the pan to SD. Do it once with the "select deinterlace method" box checked and once with the same box unchecked.

In order to see the difference, you'll have to view the view the downrezzed video through an interlaced TV or SD monitor. You can do this very easily if you have a Canopus ADVC-110 or similar firewire to NTSC (or PAL) convertor. Just start a new SD project at the rendered SD resolution and set up the preview so that it is at "best" and so that it is being previewed though the firewire box. Stop on any frame with motion and you'll see the typical "flickering between two positions" look of properly interlaced footage.

If the footage was resized without the "select deinterlace method" tab checked, not only will the fast pans have squiggly lines at the edges which run up and down, but there will no jumping back and forth between fields on stopped frames because the motion won't be properly separated into fields.

If this is confusing, I'm sorry that I can't explain it better, but it is something that I am absolutely certain about.

Would you be willing to email me the mask you referenced?

Your email is hidden in your user preferences. However, I uploaded it to Yousendit and you can get it here (for seven days):

HDV to SD mask

I am not sure I am understanding the discussion about deinterlacing (although I fully understand the concepts). However, I recommend that you NEVER deinterlace if your goal is to watch the results on a TV set.

"I admit they do look a little better if the dvd player converts the dvd to progressive display mode on the hdtv monitor."

Yet another thing that makes it confusing. Field order problems that look horrible on an old fashioned interlaced TV don't always look so bad on a newer progressive mode LCD or plasma TV. I have an HD TV in the living room and an old SD CRT TV in my bedroom. I remember looking at some DVDs that didn't look too bad on the HD TV but looked absolutely awful on the old SD CRT TV.

Here's a general rule that is good to follow with Vegas:

If your source material is progressive, always uncheck the "select deinterlace method" box. If you check it, it won't look too bad, but you'll lose half your up and down resolution any time you resize.

If your source material is interlaced, always check the "select deinterlace method" box. If you uncheck it, your video may look sharper on static shots, but it will mess up all your moving parts with squiggly edges running up and down the frames.

Whether you like interpolation or blended fields is up to you.

Laurence,
Very interesting.
Thankyou for posting this information related to the Vegas De-interlacing settings.
I found your suggestions to be dead on with those de-interlacing settings when converting from HDV to SD. I played the videos back on a SDTV 36" interlaced TV via a panasonic dvd player connected 480i (Not progressive, the TV is a SDTV).

Following your advice I took my HDV test videos which contain panning of vertical uprights (fencing) and vechicles in motion. I exported to both Lower Field First & Upper Field First 9 test videos each one using the 3 selections for de-interlaceing. Upper or Lower field didn't seem to matter as Vegas stored the new videos in the correct field order.
Using none for de-interlacing I saw jaggies & motion wasn't good at all.
Either Blend or Interpolate made a world of difference, my preference being Blend which didn't appear to have much shimmering when paused.

I moved from VMS to Vegas, in VMS the default de-interlace setting is Blend, for all the templates. I'm assuming this is why my HDV->SD conversions came out so nice in VMS. All my Vegas templates the de-interlace settings appear to default to "None".

Now I need to go back and check the SD dvd's I've made from Vegas. I think maybe viewing those dvd's on a progressive player to a HDTV may have made the difference that I wasn't picking up, not sure.

Thanks again for pointing this out. I have to admit when leaving the de-interlacing to "None" the videos looked pretty bad on the motion side. I haven't come across, or don't think that this is occuring when rendering out a new HDV video file, HDV-> HDV (UFF). All those HDV videos appear to look correct viewed on a 1080i HDTV (60" Sony LCD Projection, interlaced) and the HDV -> AVC/H264 videos look identical to the HDV source. When I've done everything in HDV or avc/h264 I've left the de-interlacing to None.

Keep in mind that when you are going from 1080i to 480i SD, there is no actual deinterlacing going on even though you have the "select deinterlace method" tab checked. Thus which method you choose in this case makes no difference whatsoever.

Choosing "blend fields" or "interpolate" does make a difference however if you render to a progressive format.

My recommendation is to select a deinterlace method any time your source material is interlaced, but only worry about which one you select when you are rendering to progressive.

Also, be aware that you are going to have some ghosting if you render 60i to 24p regardless of which deinterlace method you choose. This is neccessary in order to keep some semblance of smoothness in the motion. 60i to 30p works as you would expect it to.

Another tip: don't bother with the Mike Crash smart deinterlacer. A simple blend fields or interpolate looks better. If you do use the Mike Crash deinterlacer, forget about resizing at the same time. It looks terrible.

Keep in mind that when you are going from 1080i to 480i SD, there is no actual deinterlacing going on even though you have the "select deinterlace method" tab checked. Thus which method you choose in this case makes no difference whatsoever.OK, thanks. Rendering out some HDV -> SD now and remaking some dvd's.

"All those HDV videos appear to look correct viewed on a 1080i HDTV (60" Sony LCD Projection, interlaced) and the HDV -> AVC/H264 videos look identical to the HDV source. When I've done everything in HDV or avc/h264 I've left the de-interlacing to None."

Where that will get you in trouble is if you zoom and crop a bit. HDV source to HDV end product looks fine without a deinterlace method if you don't resize, but as soon as you do something like crop out something at the edge of the frame, look out! You've got the squiggles! It is easiest just to leave the "select deinterlace method" tab checked any time you have interlaced footage.

Conversely, NEVER check the "select deinterlace method" tab when you are working with progressive footage. If you do, you'll lose half your vertical resolution every time you change the size even just a little!