Work Flow for best image quality with codecs

LatentImage wrote on 8/22/2003, 12:41 PM

I am going to be doing a color correction pass on a DV project I shot in Vegas 4.0 and wanted some feedback on what was best work flow. I have been dong some research on the issues and have come up with the following plan.

The project comes to me with locked (uncorrected) picture on DVCAM in PAL 25 interlaced. I then capture it over 1493 to Sceanalyzer so I can use it's optical scean detection to give me as many cut points as possible. I import the avi's into Vegas (at this point if I'm correct there is no image loss as Scenalyzer exports untouched DVC and Vegas imports the same). I then do a first color correction pass in Vegas where the images will be decompressed/compressed using the excellent sonic codecs with minimal image loss. I save it to uncompressed avi. Then I need to do some noise reduction and here is where it get's complicated.

Vegas doesn't seem to have the depth of complex noise filters I need. So my best choice seem VirtualDub. But VD won't read avis unless they use a vfw codec and Vegas's saved avi's list microsofts dvsd as there codec which is not a vfw codec. So I have three options.

1/ Convert the dvsd avi into a Canopus cdvc avi with the Canopus codec converter which is vfw and VD can read. Is there a loss of quality here as presumably the decompression is handled be the shitty microsoft dvsd codec? How good is the Caopus codec when I use it inside VD.

2/ Install a third party dvsd compatable codec that works for vfw. There seem two candidates Mainconcepts and the Panasonic codec. Which of these is better for image quality? I am inclined towards the Panasonic because it's free. Any feedback on it's quailty?

3/ Install Satish's plugin and use the VD filters inside Vegas. This would presumably avoid all the codec issues but I am inclined against it because:
1/ The render time is so much slower and I have alot to do.
2/ The preview function on the filters I intend to use don't really seem to work and I need to judge by eye.

Assuming I take path 1/ or 2/ I now need to get the files back into Vegas which will compound any issues with any of the codecs. I then will do a final pass in Vegas (possibly with some unsharp mask) and print back to DVCAM tape and send it back to my clients.

Beyond any input anyone has on the impact the workflow options will have on picture quality (I'm not concerned with sound quality in this part of the chain.) I would also love any general feed back on the plan. Will I hold frame accurate sync through htis chain? Is there a much better way I havn't thought of (I have considered importing their edl etc. instead of using scenalyser but they are on FPC and a Mac and the issues I been having with my trial runs make this seem like childs play. I really get the impression alot of time and money has gone into making the major NLE's incompatable. I have consideres Automatic Duck but it is expensive for a one off use.)

Sorry for the long post but I wanted to be a specific as possible. Thank you for any help. I am really impressed by the quality of the Vegas community.

Comments

BillyBoy wrote on 8/22/2003, 2:04 PM

You're doing it backwards and making the process needlessly complicated.

I'll assume you mean video noise. VirtualDub should be used to PRE PROCESS. The first thing you do. I agree the current crop of filters in Vegas in weak in this regard. I do not frameserve. No need or reason to. I get anywhere from 70-90 frames per second running in VD directly. If you frameserve I doubt you'll get anywhere near that speed. In other words the processing in VD is fast, painless and results in a file that Vegas can easily handle. Once you pre process THEN use the excellent color correction filters in Vegas.

Howdie wrote on 8/22/2003, 7:18 PM

well i keep asking the same question and noone answers !! how to get vegas 4 to use system codecs !!

its seems that vegas 4 licenses mainconcept codecs: mpeg, and probably DV...

so how do you add in panasonic codec to the vegas 4??

BillyBoy wrote on 8/22/2003, 7:23 PM

What does system codecs have to do with THIS thread?

You'll get a better chance of getting answers if you start a thread with an appropriate subject.

Your question has been answered in the past. A lot. Vegas should be able to use any CODEC on your system. To see what's there, do this:

a. File/render as, select file type as AVI
b. click CUSTOM then in the window that comes up the video tab
c. click the down arrow under video format.

All the codecs you have properly installed should appear in the list

LatentImage wrote on 8/23/2003, 10:16 AM

Thanks for the reply. My logic, probably flawed, was that the first pass in Vegas brightened the image (mostly levels (gamma) and curve adjustments) accentuating the video noise. This would give VD more to work with when it did it's pixel comparisons for the temporial and possibly spaitial smoothing. So should I noise reduce after having effectivly increased the gamma in the shadows and accentuated the noise?

Second I don't understand your comment about frameserving. What is frameserving or rather how do I get VD not to do it? Thank you for indulging a newby. I have tried to read enough to avoid repetative questions about system codecs.

BillyBoy wrote on 8/23/2003, 12:41 PM

Need to preface by saying every source file is different. However I think most of the time doing it your way; adjusting pixels in Vegas accenturating the video noise makes VD's job harder. In other words you may very well be making it more difficult for VirtualDub to remove the video noise. I would suggest most times getting rid of the noise first works best. This will blur and blend some pixels using the excellent filters avaialbe in VD. Once the noise is gone, then go about improving the image in Vegas with its excellent set of filters.

Again, got to preface every source file is different, but I seem to get the best results removing noise first, then bring the file into Vegas, set levels, then black and white points, then any fine tuning with color corrector in conjuction with color curves.

I'm not a big fan of frame serving. It probably got popular due to it being frequently discussed over in VCD Help. That was back in the days before Vegas and a lot of people were using little more than VirtualDub and similar things. Then frameserving to TMPGEnc for example was a benefit. All frameserving is really is feeding the output of one application to another application so you can take advantage of features not otherwise available in the host application. Awhile back Satish developed a plug-in so you could frameserve from Vegas. The problem is the Vegas engine is much slower than Virtual Dub's which can fly through what it does on the order of anywhere from 3 to 20 times faster than Vegas can. Since I use VD to preprocess I find it much faster that way. Example: Last night I ran a 40 minute file through VD with five filters applied and it only took 12 minutes to process it. While I could have done something else in Vegas, I took a mini break and chatted here... which is one reason you see me pop in and out of the forum as often as I typically do. A busman's holiday. :-)

LatentImage wrote on 8/24/2003, 10:38 AM

Thanks for your help BillyBoy. I'll try preprocessing. One of the reasons I'm using Vegas for this is it's excellent user community of which you are a great example. I've injoyed your color timing tutorials.

If you have a second I know this is not on this topic but what VD filters do you like, particularly for noise, and what settings do you typically use.

BillyBoy wrote on 8/24/2003, 11:48 AM

Usually a combination. I'm kind of partial to Dynamic Noise 2.1 and the 2D cleaner and I think there's one called smart smoother. As far as settings it varies depending on the noise. I usually stay pretty close to the defaults sometimes a little stronger.

Then I bring into Vegas. Sometimes after color/level correction I'll add a bit of the unsharpen or the sharpen filter. Both do sharpening in spite of the name of one that suggest the opposite. Anyhow, if you sharpen it seems to work best if that filter is the LAST in the plug-in chain. Remember that both VirtualDub and Vegas apply filters in the order you set them up. So if you're going to sharpen, probably in most cases the last thing you want applied.

DataMeister wrote on 8/24/2003, 1:11 PM

The unsharpen filter is actually an unsharpen mask. What that's saying is, it is the sharpen filter with a mask applied to unsharpen certain areas. Or in other words to remove the sharpening effect. That's how the name came about I believe. Why they didn't just call it "sharpen with mask" ... who knows.

JBJones