Since you didn't check in on this discussion, I assume you missed it.
You'll find a wealth of suggestions and discussion.
http://www.sonycreativesoftware.com/forums/ShowMessage.asp?ForumID=4&MessageID=738430

Also, Nick's research into deinterlace and lanczos resize is relevant to your question, although your output will be MPEG-2 for DVD.
http://www.bubblevision.com/underwater-video/Vegas-YouTube-Vimeo.htm

@farss:

Edit on a HD timeline. Then with de-interlace set to interpolate and Project set to Best nest that project into a SD project and add however much sharpness you want there. Encode from the SD project.

Wow I never thought about doing this before. I shoot in HDV, render to MPEG using the appropriate NTSC widescreen template, then burn a DVD using DVD-A. Following your method, Is the SD output for a DVD pretty good? Any better than rendering from the HD (in my case HDV) project?

I don't pretend to understand all the technical details but I'm pretty good with "monkey see, monkey do" if I can follow the workflow : )

Kimberly

"If you find you still have problems with aliasing add blur to the HD child project."I'm aware this is the standard recommendation from this group, but I've had better luck with right-clicking on the video events and checking the Properties->"reduce interlace flicker"

My recommendation would be to try both methods and see what works best for you & your footage.

...Jerry

@amendegw
I thought that the last time I saw your post you said that Gaussian Blur of 0.001 was too much so I did a little experimenting.

How about putting two copies of the video on the timeline, one above the other, and applying 0.001 blur to the top one and also turning its opacity down. This would reduce the magnitude of the blur but not reduce its range unfortunately, which is presumably what you want.

I wonder if we would get any joy from SCS by requesting an extra decimal place in the value boxes of the plugin.

Hi Bob, I tried your method. I used Jerry's flickering curtain hula dancer test clip for the test. Tell me if this is correct. I dragged the AVCHD test clip on the timeline matched the format automatically. Then made sure Project settings were set to Best and interpolate. Rendered out to mainconcept mpeg-2 sd widescreen 720x480.

Dragged the new mpeg 2 clip on to a new timeline, matched the format automatically and rendered that out to mainconcept mpeg-2 sd widescreen 720x480. I was not happy with the results.

I did not try adding a blur filter being there was so much artifact and moire going on I felt blurring it to camoflarsh the problem is not what I want?

I went to Jerry's HD to SD Challenge link www.sonycreativesoftware.com/forums/ShowMessage.aspForumID=4&MessageID=738430
It had allot of good information but was way over my head and some of the software was much too expensive.

All I'm looking to do is take AVCHD clips and have them look normal after they get resized to mainconcept mpeg-2 for a sd DVD.

Can it really be that difficult. If some one has a simple way to accomplish this with out reading through hundreds of forums or buying expensive software please let us know.

95% of my work is delivered on SD DVD so how about I just shoot SD mode with my Sony HDR-AX2000 720 X 480 60i 9 Mbps 16X9 mpeg-2. This should work great with Sony Vegas. Easy to edit, mainconcept mpeg-2 render should be a great match. has anyone taken this approach?

Jerry K

"@amendegw Ha! Peter, you caught me! I thought I could remove that comment before someone would notice.

I guess what I really wanted to say was... by the time enough blur is added to eliminate the flicker/moire, the image has lost too much resolution - I've had better luck using the "reduce interlace flicker method" I just rendered a version of the Hula Dancer clip that illustrates this point. The left side has Gaussian Blur keyframed starting from 0.000V/0.000H to 0.003V/0.003H. HERE The left side has Reduce Interlace Flicker applied (no keyframing). Look at the Flicker/Moire vs the the resolution on the grass skirt. At the beginning (with 0.000V/0.000H) the skirt image is better than the "Reduce Interlace Flicker" however, by the time the flicker/moire is reduced to a managable level, the skirt has become very fuzzy. Hence, my observation that the "Reduce Interlace Flicker" method appears to produce better results.

Now, I must say neither method is perfect, and I suspect the results may vary based on the source clip. Try using both methods and see what is better for your particular footage.

Good Luck!
...Jerry

Edit: Here's a screenprint from the last few frames of the video - where the Gaussian Blur has reached the .003V/.003H levels.

Gotcha!

Yes, I agree in your results, deflicker looks good.

@Jerry_K,

Here's a couple of suggestions. First, try the "Quick and Dirty" method (a couple of things I forgot to mention in the tutorial: 1) Make sure "Video Render Quality=Best", 2) Up the minimum VBR rate to something higher than 192,000, maybe 4,000,000, 3) Render Video as "Stretch Video to fill frame size" and 4) Set the AC-3 render to: Dialog Normalization to -31 dB; Line mode profile and RF profile set to none.):



If that doesn't produce good enough results, here's the best method I've been able to come up with (without using software external to Vegas):

1) Create a Sony Vegas project my matching the project properties to your HD clip(s) – in this case the AVCHD.zip clip.
2) Make sure the Project “Rendering Quality=Best” and the “Deinterlace Method = Interpolate”
3) Drag the HD Clip to the timeline.
4) Right-Click the Clip(s) to set Properties->Reduce Interlace Flicker
5) Add a Sharpen FX set to 000. This is optional - it may produce excessive moire/flicker in the final render
6) File->Render-As (this will resize the project to 720x480 NTSC Widescreen)
....a. Video for Windows (*.avi)
....b.Template = “NTSC Widescreen” – but changes must be made:
........i.Click Custom
.......ii.Select “Lagarith Loseless codec” from the “Video Format” dropdown.
.......iii. Save this template as “Lagarith 720x480 NTSC Widescreen - Interlaced”
.......iv. Click “OK”
....c. Make sure “Stretch Video to Fill Output Frame (do not Letterbox)” is checked.
;;;;d. Choose a filename (with an AVI extension) to render your output and click “Save”
7) Close this project and start a new Vegas Project. Match the project properties to the AVI file you just created.
8) Drag the AVI Render you just created to the Vegas Timeline.
9) Right-Click the Clip(s) to set Properties->Reduce Interlace Flicker. Add Sharpen FX = 0.000
10) File->Render As (this will convert the format from AVI to MPG)
....a. Save as Type: MainConcept MPEG-2
....b. Template: DVD Architect NTSC Widescreen video stream (make sure minimum bitrate is > 192,000 (maybe 4,000,000)
....c. Name your File name with an MPG extension.
....d. “Stretch Video to Fill Output Frame (do not Letterbox)” should not be checked.
....e. Click “Save”
11) File-Render As Dolby Digital AC-3 and render the auto with the same name as the video, but with an ACS extension
12) Import the newly rendered files to DVD Architect and create your Standard Definition DVD.

Note: If you get to step 6-a-ii and do not find the Lagarith lossless codec - you can find it here: lags.leetcode.net/codec.html

If that's not good enough, use VirtualDub & resize using the Lanczos 3 resize filter to resize to 720x480 WS.

If that's not good enough, buy Cineform HD Link or Sorenson Squeeze or one of the other software products designed to do this.

If that's not good enough, punt!! [grin]

Good Luck!
...Jerry

There's just one problem with that challenge.
The result is being viewed on a progressive display.
There are TWO issues that need to be wrangled:

1) Aliasing. This works the same for video as it does for audio. Go over the Nyquist limit and you may be in an ocean of grief. I've had a client give me some audio they'd somehow managed to "create" that had harmonics in it over that limit. I never could find a way to get rid of the problem. None of the filters in Vegas or SF would kill the problem and the music had something in it that could strip the enamel off your teeth.
Same thing can happen with video. It really has to be wrangled in the camera in the first place. This is why the video experts test cameras for this very problem and if they don't pass they just stop testing and say you can't use the camera to shoot content for them.
In the audio world the aliasing problem is wrangled by low pass filering before the A>D converters and today is done digitally, the A>D converters generally are oversampled so this can be done. In the video world oversampling isn't pratical and there's also the moire problem on top of the aliasing problem so an optical low pass filter is used in front of the imager.

2) Line twitter. This is a problem unique to interlaced video and it is on top of the aliasing problem.
Imagine you ruled 480 lines 1mm high on a piece of card that were alternating black and white. You took a NTSC digital video camera with some "perfect" lens and very carefully aligned those lines with the scan lines in your camera. You should be able to record those 480 lines and display them on your TV.
Now if the camera and the display is all progressive not a problem. If the display is interlaced what it will display is all the black lines in one field and all the white lines in another. What you get is a blinking mess.
Of course aligning those lines on the paper with the sensor is darn difficult. Imagine though what happens if you pan up ever so slowly, the field with all the black lines and the field with all the white lines would alternate and your interlaced TV would create an even worse blinking and twittering mess.
To get around this problem cameras when shooting interlaced video vertical resolution is limited to around 70% of the theoretical maximum. So for SD NTSC the maximum vertical resolution is around 340 lines and your piece of nicely ruled paper is going to appear as just grey.

The line twitter problem in video cameras is wrangled digitally using line pair averaging which effectively reduces vertical resolution.



Now, if you have a good HD camera and shoot interlaced video and downscale that to SD you should not have too many problems. Your good camera will limit vertical res to 70% of the max and when you downscale the outcome should be the same. It may look a tad soft but it really shouldn't matter if you shoot HDV or AVCHD or XDCAM EX all of this is independant of the codec. To make it look a bit sharper you should add some form of detail enhancement after downscaling. You can certainly run into issues if your HD camera is adding too much "Detail" which is why I have it dialled down in my EX1. That said even our lowly HC5 is pretty clean after downscaling to SD.


I have to admit to a failure to disclose that of late I've not had much luck using GB to wrangle a line twitter problem. I did shoot a job in 720p and it was a nightmare to get a good SD 50i DVD out of it. As Jerry has said I needed more GB than I was happy with but unlike him I was not happy with using the Reduce Interlace flicker switch either. In the end I spent ages masking out the problem walls and adding enough GB to kill the problem and left the talent alone. Thank goodness the job was only 3 minutes long.

I should also add that the above explaination is far from the whole story and probably misses a lot of the technical fine points. There's much more accurate info on the web but the really good oil from the likes of Alan Roberts from the BBC oftenly has too much maths for me to follow which is why I've taken a long time to get a more layperson's understanding. I really should have paid attention in maths class.

Bob.

All the theory of 30p, 60i, odd and even field is all very nice but meaningless to most of use unless it fixes the problem and in this case it does not.

All I know is I bought a HD Sony camcorder for $3,500 hundred dollars.This camcorder takes beautiful video with out any Aliasing, artifacts or moire until I down convert it in Sony Vegas.

I'm not blaming Sony Vegas for the problem reason being other systems have the same problems and I believe it's just the nature of the beast when down coverding HD.

Now with all that said we need to know if there's a fix for this issue? Is there some after market software out there that can fix this issue. Is it affordable? Is it easy to use?

I just want to be able to take HD footage and down convert it to SD with out all the Aliasing, artifacts or moire.

If this is possible, simple and affordable please let us know.

Thanks, Jerry K

It sounds like you think this is a single issue. It is not, but a constellation of considerations for which there is not a single "fix." Believe me, we would not have gotten into discussing the complexities of deinterlacing and resizing if a one-size-fits-all solution already exists.

Vegas uses bicubic resampling and blend or interpolate deinterlacing. These are the same methods that were in use fifteen years ago, and are perhaps the worst, generally speaking, for the issues you raised in your first post.

No, there is not a self-contained software that will convert 1080i AVCHD to gorgeous DVD video with the push of a button. You need to work at it. Newer tools like lanczos, yadif, and others can be used in various combinations to achieve the best "look" for a given piece of video. But it falls into the category of an art rather than a science.

And regardless of which combinations of tools one uses, you will always lose 83% of your visual data in the downsampling from HD to SD. Get your feet wet, and maybe you will be able to improve on the research done in the two threads I linked previously.

"All I know is I bought a HD Sony camcorder for $3,500 hundred dollars."

Can your camera output in SD? That is one solution some people recommend for down conversion.

If you also wanted an HD version, perhaps you could edit in SD and treat it as a proxy for an HD version.

ok, we hear you but you don't seem to be hearing us. Originally you said :

" If I downconvert with Vegas sharpness filter at "0" the video is sharp but very annoying to view with all the Aliasing, artifacts and moire. If I create the DVD with out a sharpness filter the video is very watchable but flat."

I suggested a fix, applying sharpening after downconversion. Have you tried doing that?

Also are you shooting progressive or interlaced?


If you want a perfect solution then Snell and Wilcox or Teranex make the best boxes for this but they cost a staggering amount of money and even then I still see some aliasing etc problem on SD DVB broadcast.

Bob.

if your audience is more concerned watching the moire patterns it doesn't say a lot about the content - that is, unless you're displaying herringbone material.... ;-)

without spending oodles of money (bob can direct you where ;-)), you're going to have to live with certain anomalies - as he (bob) points out, all you have to do is watch sd tv to see the problems.

and buying a ($3,500) camera without running comprehensive tests on its abilities for your needs beforehand is rather putting the cart before the horse.

there has been quite an ongoing discussion regarding avchd > dvd in nearly all the forums i've visited, eg:

http://www.dvinfo.net/forum/what-happens-vegas/493756-avchd-mts-sd-dvd-via-vegas-pro-10-a.html

and lots of suggestions, workarounds, whatever.

i shoot exclusively in hdv, but often use a friends avchd camera as a 2nd one when needed. i simply dump the footage on the tl, edit, render to dvd and haven't had any complaints (i can see some anomalies here and there occasionally - but i certainly wouldn't lose any sleep over them). if i need to do more with the footage i transcode to mxf. end of story.

if you want to try for perfection, it comes at a high price, either financially or time-wise....

Bob has said it already but I will repeat it explicitly:

To reduce moire, aliasing, etc. apply blur (low-pass or antialiasing) filter BEFORE you downconvert.

If it is too flat, apply sharpening AFTER downconvert.

You can do this in two steps by generating an intermediate file or you can do it in one step by nesting projects.

Jerry K said,

All I know is I bought a HD Sony camcorder for $3,500 hundred dollars.This camcorder takes beautiful video with out any Aliasing, artifacts or moire until I down convert it in Sony Vegas.
==============
Musicvid replied,

It sounds like you think this is a single issue. It is not, but a constellation of considerations for which there is not a single "fix." Believe me, we would not have gotten into discussing the complexities of deinterlacing and resizing if a one-size-fits-all solution already exists.
============================
Leslie said,

there has been quite an ongoing discussion regarding avchd > dvd in nearly all the forums i've visited, eg:
==================================
Which brings me to a point I haven't posted in a few years, but used to quite regularly (to deaf ears).

This problem, as Leslie correctly points out, is a discussion all over the video forums. The answers, although never perfect, are all workarounds for the particular NLE software being discussed.

Look at the length of Jerry's challenge thread and the related threads or Musicvid's mammoth thread.. Look at Jerry's and Nick Hope's respective web pages they have devoted to these problems. All of them workarounds involving other software. Discussion after discussion and post after post of each individual's workarounds. And most involve freeware !!

So here is my simple question:

Why isn't the Sony software development team working on this stuff as an improvement to Sony Vegas?

I bought Vegas 2, then 3, and then 4.

I didn't buy Vegas again until I bought an HD cam. Then I bought Vegas Pro 8 because it had something BIG and really really useful. The ability to edit and render HD files. I haven't bought Vegas since.

My complaints used to be and still are that basic needs are ignored in the software upgrades in favor of bells and whistles. Example: The Credit Roll generator has NEVER changed. The Titler only changed recently. Those problems are DWARFED by the one we are discussing here in terms of pervasiveness and numbers of threads and posts.

And how many discussions have we had about how to apply the deinterlace setting (None, vs Blend, vs Interpolate) and trying to figure out the quirks in those settings and whether you have to apply it even if it seems illogical to under a given circumstance? Endless. Has Sony changed anything after reading all these discussions about those settings (for years now)? NO!

And if the Mainconcept rendering engine isn't the best around why are we stuck with it forever? Or will we see in the future something like: Buy Vegas 28 with new and improved HD to SD rendering engine!

And why is it that I have to use Sorenson Squeeze for rendering decent Quicktime video? Sell us a quality Plugin for Vegas if it's a money issue. It has to be cheaper than what I spent on Sorenson Squeeze.

We don't need bells and whistles in Vegas. We need some REAL changes.


................................end of rant.
John

John,

I certainly simpathize with your rant. I'm a little less outraged (after all, what else would I do with my life - [grin]) fwiw, here's the question I posted at the recent "Creating DVDs..." Webinar:

"What is the recommended method for downrezing Vegas HD projects to SD DVDs - recognizing that resizing often results in flicker and moire patterns as well as loss of sharpness. And are we getting any help in future versions of Vegas / DVDA to assist us in this process?"

They promised to post answers to all questions after NAB. I'm waiting with bated breath.

...Jerry

I understand the frustrations but a call for more effort at solving the problem is fine if it is solvable. Unfortunately as far as I know it is no more solvable than going faster than the speed of light or travelling backwards in time.

Here's my (further) understanding of the problem.
The SD video sample clock runs at 13.5MHz. Full raster resolution equates to a bandwidth of 6.25MHz and that is exactly the Nyquist limit.
The original design of digital audio used a clock of 44.1KHz and a bandwidth of 20KHz, just a smidge under the Nyquist limit. I don't recall the man's name but he was one of the true audio gods of the day and he argued long and hard that the sample rate should be 48KHz to get more distance between the Nyquist limit and the bandwidth.

So going back to video land there's a problem. We need a low pass filter with a corner frquency of 6.25MHz and an infinite Q i.e a brickwall filter. There's a problem, such a thing doesn't exist. From many, many years ago back in the Sonic Foundry days the developers did explain the problem. The code to get closer and closer to the ideal filter involved running a loop, the more times the code goes around the loop the closer you get. Say 10 times gets you 90%, then 100 times gets you 95% and 1,000 times gets you 98%. Render times obviously blow out something fierce.
So to get a reasonable result it a reasonable time they use a less than ideal filter which does not completely stop at 6.25MHz and that does carry a risk of some aliasing. You can really see this in high resolution images from a good DSC if the images contain fine detail. That's what started the discussion around 10 years ago and nothing has changed since except now we're all shooting high definition moving images which does add another set of issues on top of the original problem.

Now I'd be the first to give SCS a good kick in the butt if I thought they were being tardy but I cannot justify doing that. If nothing else no one has managed to do better. The guys using products from Adobe, Apple and Avid are having the same issues. No one has managed to change the laws of physics. Again tonight watching my SD CRT TV I see big budget TVCs with aliasing and line twitter problems. As I said before even with $100K hardware boxes there is no perfect answer.

Bob.

I hear what you are saying Bob, but how does that solve the resizing problem. And do you think that the deinterlace settings (None - Blend - Interpolate) are NOT in need of improvement?

Or that the MC rendering engine is sufficient?

Or that Vegas QT rendering is adequate?

John

Or that Jerry's question to SCS does not need an answer?

Bob,

I must defer to your technical expertise on this subject - I know nothing about Nyquist limits, relationships to "full raster resolution", etc.

However, the thing that encourages me that great strides can be made in a software only solution to this problem is that in the HD to SD Challenge thread, David Newman (HD Link) and Perrone Ford (Sorenson Squeeze) both posted very good results - albeit with pricey software.

So, I guess we are where we are. Several very knowledgeable people (including you) have posted many techniques to improve this process. Unfortunately, we do not currently have a perfect solution - Sigh!!

...Jerry

How much of this issue is related to the slight PAR difference between HD and SD Widescreen? Has anyone tried these without the stretching, and just crop the HD to the SD widescreen ratio, then render with no stretching?

"How much of this issue is related to the slight PAR difference between HD and SD Widescreen? Has anyone tried these without the stretching, and just crop the HD to the SD widescreen ratio, then render with no stretching?"In my testing, I see very little difference in image quality.

...Jerry