I generally defer to Spot on this sort of thing, but I'd sure like to hear his reasoning as to why he deinterlaces everything. I'd also like to now what technique he uses for deinterlacing.

Almost everything we shoot ends up on either computer displays or on an HD display, or both.
We have a variety of means of de-interlacing, one of them is proprietary, but done in Vegas exclusively. In some situations, we use Mike Crash' filters, and sometimes we'll go outside of Vegas for deinterlacing.
[edit>(didn't mean to hit Enter so fast).
Interlaced isn't inferior, it sure has its place, and will for a long time, but until standards are forced for how PsF is read, and displays quit somewhat randomly interpreting signals one way vs another, progressive has some advantages when the target display isn't controlled.
Sometimes the content is delivered SD for computer and set top television, other times, it's 720p for larger display. If we know we're shooting for a 1080 display, we don't bother with interlace at all, and acquire using progressive formats.

Interesting, as Vegas has 2 settings for this:
The Project Settings De-Interlacer & your target rendering template (whether you select fielded or progressive).

My rule of thumb was never de-interlace interlaced source HDV when rendering to another interlaced format.
I had problems in Vegas Downcoverting HDV(1080i) to SD(480i)

Going HDV(interlaced) -> HDV (Interlaced 1080i) I can leave my project settings to Deinterlace=None (usually, depends on FX's used).

Going HDV(interlaced) -> SD (Interlaced 480i )
Project Settings - Deinterlace = None ----> Video has poor jerky motion.
Project Settings - Deinterlace = Blend ----> Video is nice, good looking motion.
Project Settings - Deinterlace = Interpolate ----> Video is nice, good looking motion.

Note that the target format is Interlaced. The project settings deinterlace settings "None/Blend/Interpolate" affects the 2 fields before they are re-interlaced (I guess).

All I know is in Vegas when downconverting HDV(1080i) -> SD(480i) if I don't use either Blend or Interpolate in the project settings the SD rendered video doesn't look anywhere's as good as the original when there's motion in the video.

I thank Laurence for pointing this out to me. As it has made a hugh difference in quality when downconverting from HDV(1080i) ---> SD(480i)

A successful post based on the number and speed of replies. Thanks all.

My summary: If you have control of the display device and it can do interlace, stay that way. If you don't have control, progressive may give your more consistent quality across a variety of display devices.

One more question: opinions on the quality of deinterlacing by DG's smart deinterlacer vs. Neo HDV?

Motion-adaptive deinterlacing is the bee's knees when it comes to quality with minimal effort.

This is quite well implemented in ReVisionFX's inexpensive ReelSmart FieldsKit.

FieldsKit is also highly tweakable for when you want the absolutely best results with tricky footage, see the settings window about 1/3 down in this review.

That review also shows you a sample clip (with bouncing balls!) deinterlaced six different ways, in a large frame so you can really compare them in detail.

Not bad for $89.95, I certainly haven't seen anything else that compares.

For really easy stuff, any app will do a good job. For the rest, it takes more ambitious tools like this one.

Here's something I'm trying:

Shooting progressive1920 x1080 25p, editing progressive, then for DVD output I started with the Vegas 24p widescreen template, changed it from NTSC 720 x 480 23.976 fps to PAL 720 x 576 25fps and saved it.

The short clip I tried it with looked good, particularly on LCD - on a CRT TV there was still a slight "wavering" on some pans - I wouldn't call it twittering - so I assume that's something that either the playback device or the TV screen is doing.

Anyway, I'm going to render and burn a longer project this way to make a more considered judgement...

Just burned a 25P DVD, progressive from camera to disc ....

First time I viewed it from disc, on PC screen, shock horror, it was twittering like a demented canary!
Strange, I thought, and went back to view preview in DVDA where it was rock solid.

I thought something must have happened in the burning to disk process, then remembered that I was viewing the disc using Windows Media Player.

Changed to a different PC with Power DVD and it was once again rock solid!

So - WMP seems to have a problem displaying progressive, unless there's a setting somewhere.

Then I viewed it with a set top player on an "old type" CRT TV. It wasn't as unstable as the WMP playback, but there was some patterning - sort of moire effect, caused largely by a plain brick wall of a gymnasium in the background - lots of lines. Line twitter was not a noticeable problem.
So, quite encouraging.

A few tips if I may:

1) Don't try to fix what isn't broken. Video has always had problems with moire, it's the nature of the beast. Back when broadcasters cared, if you turned up for an interview wearing stripes you got sent to wardrobe. To see a good example grab any SD camera and zoom in slowly on a res chart or fly screen.

2) Don't use WMP, try VLC. It lets you specify a variety of de-interlace methods so you can see what might happen with different LCD or plasma TVs.

3) The Region50 systems (aka PAL) doesn't support progressive as such. That's why there's no 25p templates in Vegas. The HDV standard doesn't support it either. That's why previously I mentioned 25PsF (Progressive Segmented Fields). So even though your DVD contains 25p it will be sent to the display as 25PsF, that's why you can get line twitter problems. If you use a plasma or LCD TV and it weaves (merges) the two fields then no line twitter problems. That's not so common it seems as there's no flag available to tell the TV it should handle it as Progressive. If you're lucky it might just figure it out. If not it'll assume it's interlaced and try to de-interlace it using bob or some other method which can produce interlace artifacts.

4) This has been mentioned before but it is worth saying again. You MUST specify a de-interlace method in your project settings.

5) Downscaling 1920x1080 to 25PsF can result in a 50i signal going to a CRT with too much vertical resolution. This can cause problems. We went through a very long discussion over this with the V1P. It wasn't a problem with the V1U as the Region60 version had less vertical res. With the EX1 delivering still more vertical resolution the issue gets more likely to be an issue. In the end this is nothing new. It's the same problem we've been dealing with for many years with HiRes stills from DSCs . Progressive video is after all just a sequence of images the same as one could get from a digital still camera.

Some HTDVs have truly woeful upscalers. Even the best SD ends up looking horrible The V Series Bravia would have to be the worst I've been forced to watch.

One final trap. When displays get things wrong they can even create noise that isn't in the source or make it look worse than it really is.
Probably not a bad idea to calibrate your expectations by watching a couple of regular DVDs on the display device before judging your downscaling efforts. You monitoring device could lead you to wasting a lot of time fixing problems that you don't have.

Bob.

farss,
I use a few different customized VLC shortcuts to run VLC.

On my systems I use this command when playing back HDV (1080i) Mpeg2 video.
VLC does a nice job of de-interlacing the source for playing back on the computer monitor and playback is pretty smooth.
At least for me this VLC setting works great for interlaced hd-mpeg2 video.
The --zoom=.5 stops the 1440x1080 video from blowing up to large on the computer screen.
The --vout-filter=deinterlace --deinterlace-mode=linear uses linear mode, one can also use blend, I find the linear to produce the best results with 1080i video with a lot of motion.

VLC Command: For windows add the switches to the standard shortcut (omitting the vlc below).
vlc --zoom=.5 --vout-filter=deinterlace --deinterlace-mode=linear4) This has been mentioned before but it is worth saying again. You MUST specify a de-interlace method in your project settings.Thanks to you and others previously posting this the videos I now export from Vegas look great. I seem to favor the interpolate method.

Not sure if it was clearly stated, but the "Interpolate" basically drops a field, and I see it works best on fast motion action. Page 220 of the V7 manual:

"If you choose the Interpolate option, a single field is used at a time, which is good for low detail, high-motion material."

Interpolate=for high motion content
Blend=for low motion content. Training DVDs for instance, usually do very well in Blend mode

>> Interpolate: a single field is used at a time <<

Isn't this a contradiction in terms? Is it interpolating or is it dropping a field? Two different things.

Or is it perhaps alternately dropping field 1 then 2?

I guess the only way to really see what's happening would be to build a test pattern.

>> Isn't this a contradiction in terms? Is it interpolating or is it dropping a field? Two different things.

When you drop a field the resulting frame has to be resized (since it's only half the height). It's probably during the resize process the interpolation takes place.

Ah, ok. That makes sense.

I'll try changing from "blend" to "interpolate" then for my half-HDV-height (540px) web movies.

"I'll try changing from "blend" to "interpolate" then for my half-HDV-height (540px) web movies."

It would be interesting to compare to your multi-step process to see which one results in better output.

I found this article: "Interlaced vs Progressive Editing" :
http://www.animemusicvideos.org/guides/avtechbeta/interlacedvsprogressive.html

Myself, I routinely shoot HDV (Sony Z1), live-capture the m2t by HdvSplit, and reframe and downscale (to SD or Web) in post (Vegas). I originally used default Vegas settings ("Blend") but noticed the result was soft. By experiment, I found Vegas's "Interpolate" deinterlacing to give better detail even in low-motion shots. That was the opposite of what I expected (is it just me?). Then I discovered motion-compensated deinterlacers (eg AviSynth's TDeint) gave noticeably better results. Did lots of experiments to come to this conclusion.

As part of a workflow, can use the original footage as "offline" media then replace it by Vfapi-AviSynth as "online" media prior to rendering. I wish Vegas could read directly from AviSynth ".avs" files or even had a mo-comp deinterlacer on board, e.g. operative when rendering in "Best" mode.

Recently I used TDeint with mode=1 so each field becomes a frame, doubling the source framerate, hence e.g. enabling better slomo in Vegas. When rendering to interlaced at original framerate (and no rate changing involved), I hope Vegas turns each frame into a field (can anyone confirm this?).

Can't imagine a better way to squeeze out the max spatial and temporal resolution from such a workflow (reframing) in a reasonably convenient-ish way, at the cost of not unreasonably increased render-time.

wish Vegas could read directly from AviSynth ".avs" files It can.

Download VFAPIConv.

Run VFAPIConv and then drop the AVS file onto the app, and then click on Convert to create a "signpost" AVI file. (Use the drag/drop method rather than "add file."). Accept the default for the output file and navigate to that folder to retrieve the signpost. Also, if you have problems with VFAPI not accepting your input file, reboot your computer and try again. It is a little temperamental sometimes to set up, but if you succeed in creating the signpost AVI, everything works great from that point forward.

Other things to try if VFAPIConv doesn't accept the AVS file: make sure to NOT open files directly from the "My Documents" folder and instead start with the drive letter (in Explorer) and drill down from there to the "My Documents" folder (or just save your AVS somewhere other than My Documents). It appears that VFAPI doesn't deal directly with this special folder.

If you get the file into VFAPI, but the converted file doesn't show up, try double-clicking on the file name to bring back up the dialog that lets you specify input/output files and then uncheck all boxes, and then immediately re-check all boxes. Again, this piece of software is a very old hack that has never been updated, but it works once you generate the AVI signpost.

Close VFAPIConv.

Open that signpost in Vegas. Edit and Render.

This works great for getting VOB and MPEG-2 files to edit and preview at lightning speed, even in older versions of Vegas.

John, thanks for taking the trouble to clarify this, it may be useful to others, though I already know about and use VFAPIConv. What I meant was that it would be better still if that extra step was not required, especially when there were lots of source files. I search online for ways of running VFAPIConv from command line (as can be done with DgIndex), to enable a Dos batch file to automatically run it on all files in a given folder/directory, but so far have been unable to find anything.

For HDTV:
Since deinterlacing ALWAYS produces artifacts can someone please explain why the broadcast industry chose 1080i and 720p as a standard in television and video instead of 1080p and 720p?

John

Since deinterlacing ALWAYS produces artifacts can someone please explain why the broadcast industry chose 1080i and 720p as a standard in television and video instead of 1080p and 720p?

I don't know why they picked one of the standards to be progressive and the other interlaced, although I suspect it was an attempt to provide a way to deliver film in a format that still looked like film.

However, as to why interlacing was kept, it is for the simple reason that it looks better for certain things. For instance, during a football game, sometimes sports stations will show replays from an earlier game and will convert it to 24p to give it that "removed in time" look that the cadence of slower frame rate and progressive provides. It definitely does the job, but it completely loses its immediacy. By contrast, having 60 separate events every second, instead of 24 or 30, gives you that "here and now" feeling.

So, you ask, why didn't they simply give us 60p? I don't know for sure, so this one is a guess, but I think it simply would have cost too much. The whole reason for interlacing in the first place was a combination of persistence of vision and cost. Those factors haven't changed, even as technology has marched forward.

Perhaps someone else has a better (or better-informed) explanation.

It was the bandwidth/bitrate of a 1080p signal that was the determining factor. You can cram about 19Mbits into the width of a standard analog TV channel. 19Mbits isn't quite enough to do decent looking 1080p. Curiously, this was essentially the same reason that 480i was chosen in the first place.