DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   Techniques for Independent Production (https://www.dvinfo.net/forum/techniques-independent-production/)
-   -   Deinterlace workflow (https://www.dvinfo.net/forum/techniques-independent-production/66543-deinterlace-workflow.html)

Jeff Cottrone May 3rd, 2006 09:15 AM

Deinterlace workflow
 
I just finished my first short film and am about to upgrade to the new Adobe Production Studio that will allow me to move back and forth from Premiere to AE without exporting and thus saving the slight loss of quality. I'm also about to buy Magic Bullet (and maybe DVfilmmaker) and have a question on this workflow in another post, which is:

1. Capture in Premiere (to DV-AVI)
2. Basic rough edit to remove bad takes in Premiere (to DV-AVI)
3. DeARTIFACT with Magic Bullet (to uncompressed AVI)
4. DeINTERLACE with DVFilm Maker (to uncompressed QT)
5. Color-correct in AfterEffects (to uncompressed AVI)
6. Edit in Premiere (to uncompressed AVI)
7. Finalize (opticals, letterbox, broadcast spec) with MagicBullet (to uncompressed AVI)
8. Encode to MPEG-2 in TMPGEnc
9. Author to DVD in DVD-Lab

Three questions:
1. I'm assuming with Adobe's Dynamic Link that I don't have to export from Premiere to use Magic Bullet. I can just hop over to AE and do the deartifact, correct?
2. Did the new version of MB become better than DV filmmaker's ability to deinterlace? I suppose this is subjective, but if I have to export from AE to deinterlace with DV filmmaker won't that export cause a slight loss in quality? Wouldn't it be better to also deinterlace with MB to avoid that?
3. As for the color correction step...I have to color correct many different clips, shot months apart, but now sit next to each other on the time line. If I export at any time during this process, won't all my clips merge together? So how would I color correct each clip differently? So, if I deinterlace and deartifact within the Adobe line (without exporting), won't that keep my clips separate, so I can color correct each one independently? Or should I color correct first, then deartifact and deinterlace?

Any help would be greatly appreciated!

K. Forman May 3rd, 2006 09:20 AM

I'm not totally sure, but I think that exporting to avi then to quicktime and back to avi, is really going to degrade your video badly. Just exporting repeatedly will be bad enough, but changing codecs that many times can't be good.

Alain Bellon May 3rd, 2006 10:08 AM

Quote:

Originally Posted by Keith Forman
I'm not totally sure, but I think that exporting to avi then to quicktime and back to avi, is really going to degrade your video badly. Just exporting repeatedly will be bad enough, but changing codecs that many times can't be good.

It won't affect quality if you use lossless codecs. You can use YUV codecs which will compress the video but will not lose a single byte from it. (For PC there is a HuffYUV codec that I love to use for this purpose)

If you do this, it doesn't matter how many exports you do, you will preserve the quality intact. ( aside from the quality alterations from doing deinterlacing or color correction )

Carl Jakobsson May 3rd, 2006 10:53 AM

The workflow is similar to mine when working on the pc.
However, I always edit the whole thing first without any filters, cropping, motion, speed-changes etc. When I'm totally satisfied with the edit, then I start to color correct, deinterlace etc. This is also good because otherwise I tend to get stuck on details when i comes to color-correcting for example. To put it another way, for me EDIT and POST isn't the same thing. First EDIT, then POST.

When I'm satisfied with the editiing I export clip by clip uncompressed and process them in After Effects/VirtualDub/Vegas, then uncompressed back to Premiere.

In Premiere I make a new track with the processed clips placed exactly above the unprocessed clips. This way I can always go back to the unprocessed clip and export it again, if I decide to change an edit (eg a clip is too short). I then color correct the processed clips seperately (this is kind of an answer to question 3).

The codecs I use are HuffYUV or MSU Lossless Codec depending on what mood my Windows is in at the moment... :)

A general tip is to always use 16-bit processing if you have that option.
Always deinterlace first, before doing anything else. Especially if you plan to crop, rotate or change the speed of the clip. The only exception is when you want to use some plug-in that uses fields in its calculation.

The weakest link in your flow is probably TMPGenc. Otherwise, it sounds quite reonable!

Jeff Cottrone May 3rd, 2006 11:19 AM

codecs
 
"The codecs I use are HuffYUV or MSU Lossless Codec depending on what mood my Windows is in at the moment... :)"

How do I acquire these codecs? Is it something you import into Premiere and it adds to the list and you just select it?

"A general tip is to always use 16-bit processing if you have that option."

When would I select 16-bit processing? Not sure.

Also, I can't export clip by clip. I have 18 billion clips. I use lots of cuts. And b/c of a crazy shooting schedule we shot different parts of the same scene months apart and the settings weren't exactly the same. So I guess my question is can I color correct before I deinterlace and deartifact? I think I have to. UNLESS the new AE (7.0) with Dynamic Link will let me import my project from Premiere and retain all the separate clips.

Joshua Provost May 3rd, 2006 12:00 PM

Jeff,

The workflow you reference is from a post of mine. The sequence and formats are designed to maintain maximum quality at every step. I still use the workflow for short films. I have a feature to edit, and due to the time involved in deartifacting and storage involved with uncompressed (UC) formats, I may have to alter this a bit.

Some notes. First, going back and forth between AVI and QT loses no quality as long as you use UC formats, in this case UC AVI and QuickTime with None compressor. This is full frame 4:4:4 RGB data, like a sequence of BMP files. The only reason QT is used is because Maker was having issues outputting from UC AVI to UC AVI and was giving me some strange color changes which I didn't like. I believe this is fixed in the latest version.

Also note that once I deinterlace I am working with full frame 24p, there is no pulldown involved, only the full frames are saved to disk (this saves a bit of storage back to balance out the UC large file sizes, but not by much). Unfortunately, while Premiere Pro 1.5 can edit UC AVI 23.976, it does not properly output it, but does so as 23.98 and you can lose sound sync. Oddly, After Effects has no problem with UC AVI 23.976 in and out. Therefore, once I am in UC AVI 23.976 world, I use QT to go from Premiere->AE, but AVI to go from AE->Premiere. These are the kind of quirks you find out only throw many iterations of trial and error.

As to sequence, you must deartifact while the footage is still 60i, and you must set AE to output actual 60i with the right lower or upper field order. If you don't AE will deinterlace your footage, even though you have MB deinterlacing turned off. It must be done prior to any changes, because once you make changes, the deartifacting algorithm is no longer accurate. So it is quite critical to deartifact prior to deinterlacing. In turn, I think it is important to deinterlace prior to color correction or effects so the deinterlacing algorithm of your choice can do its job on as near raw camera data as possible, before you go messing with the color or contrast.

Many people swear by MB's deinterlacing ability, but do a search on my previous posts to see why I don't personally use it. In brief, MB provides the most detail, but also causes artifacts in detail areas like patterns, fine lines, and natural formations like rocks, leaves, etc. Maker is a selective field doubling, so it's not as sharp, but it doesn't have the artifacts which I found completely unacceptable for nearly every shot I threw at MB.

Of course, there are demos version of these products, and Vegas also does a nice deinterlace, so there is no excuse not to try them out for yourself, with your own footage, and make your own determination.

I'm not sure about the Dynamic Link features, since I don't have 2.0 yet, but it would be nice if you could do multi-layered steps in Premiere, AE, back to Premiere, etc, all nested, but I think there could be problems with this workflow due to changes in interlaced vs. progressive footage and other major changes your footage is going through in this process. And if you do settle on Maker, that breaks the flow because it's standalone.

As for clip-based editing versus one big footage file, I guess I am sort of treating my digital files like rolls of film in a way. I capture whole tapes, and in my rough edit I use the razor tool to cut things up, rearrange, roughly trim clips (giving lots of space at the beginning and end, just in case), toss out unusable footage and bad takes, but then I export everything back to one file, albeit smaller than the original. The whole file gets deartifacted and deinterlaced as one. Even color correction is done on the one big file, with keyframes inserted in AE to break up sections that have the same lighting conditions and can be corrected en masse. The result is one master deartifacted deinterlaced color corrected file, like a nicely processed roll of film that I can take into my final edit, where I'll again use the razor tool to cut it up and adjust clips. I may even rotate and zoom a bit to fix a slightly tilted shot, add effects, transitions, titles, everything it needs. Back to AE when done for the final Letterbox, Optical fades, Broadcast Spec, etc. and that's my final file I archive.

Man, it's a lot of work, and not quite as streamlined as I believe you are shooting for. It's what I am used to, and it does preserve maximum quality. I can't wait to get a DVX100B, so deinterlacing can completely go away. At that point I will be able to deartifact and color correct in a single step and single program, and I certainly will.

I'd be interested to see why someone thinks TMPGenc is a weak link. It may be slow, but considering all the time your putting into the project, an hour to render in TMPGenc is nothing. I haven't seen anything that beats it in terms of quality. I'd be interested to hear or see differently.

Josh
Josh

Jeff Cottrone May 3rd, 2006 12:41 PM

thanks
 
Thanks, Josh, lots of great info there! You just saved me a bunch of trial and error. Although, I'm a little bummed about not being able to retain my clips. I'd rather work with separate clips than solely with the razor tool. I guess I'm looking forward to shooting native 24p to avoid this in the future (assuming I can import from Premiere into AE to deartifact and CC, then go back to Premiere and still retain the clips--don't know if that works yet).

Joshua Provost May 3rd, 2006 01:50 PM

Jeff,

I've never worked with clips, so I don't know how they would be handled. Basically, you are looking to batch process a bunch of clips using the same settings (or different settings in the case of color correction). I suspect it's possible, but I can't help you there.

Josh

Alain Bellon May 3rd, 2006 03:23 PM

Lossless Codecs
 
Here you can get the HuffYUV and MSU codecs.

http://www.free-codecs.com/download/HuffYUV.htm
http://www.free-codecs.com/download/...ideo_Codec.htm

The advantage of a lossless codec over uncompressed AVI is that you don't get huge files to work with. For example a second of SD video with the HuffYUV codec takes around 10MB, while a second of SD video uncompressed will take 30MB.

Damian Clarke May 3rd, 2006 04:34 PM

Hello Joshua, one encoder I recommend for sheer speed and quality is 'Cinema Craft Encoder'. Not as user friendly but it's some machine.

Carl Jakobsson May 3rd, 2006 08:15 PM

Clip by clip isn't that consuming. I've worked with 100 hundred separately exported, that wasn't very timeconsuming. Not compared to how long the rendering took. One option is to export several clips to one clip if they require the same colorcorrection.

Joshua Provost May 3rd, 2006 10:49 PM

Alain, I'll look into those lossless codecs. As long as all my software will support them without issues, it sounds like the way to go when I tackle my feature.

Damian, have you compared CCE to TMPGenc head to head? I haven't personally, but all of the online reviews I have seen head to head showed TMPGenc with better quality IMO. Speed is of no concern to me, only quality.

Damian Clarke May 4th, 2006 01:28 AM

I have TMPGEnc also, but I still prefer CCE for quality. Just personal preference though ;)

Jeff Baker May 5th, 2006 09:28 PM

Dynamic link does not allow you to apply AE effects into the premiere timeline, it only allows you to import compositions made in AE and then edit or make changes to them from within premiere (I also was hoping to do what you want to do, but dynamic link does't go that far yet).

However AE 7.0 now has the ability to Capture video using Premiere and this may be the best workflow for you:

Capture in AE using the Premeire plugin.
Interpret footage to NO FIELDS and choose to REMEBER INTERPREATION.
Select all the clips and apply the no fields interpretation.
Apply MB to a 24.976 comp (only the newest version will work with AE 7.0).
You may be able to start a new Premiere Project at this point and import your AE project with Dynamic link. I don't know, but it would have to be a 24p project.
Otherwise render from AE and then edit in Premiere
Download DebugMode's FRAMESERVER or other frameserving software.
Frameserve your project and open it in your favorite MPEG Encoding application.
Encode for 24P DVD with 2-3 pulldown flags inserted.
All Done! Here is a link that may be help for the AE capture method:

http://www.digitalvideoediting.com/a...fterinter=true

Jeff Cottrone May 11th, 2006 07:45 AM

render times
 
I've heard about crazy render times for MB, but come on...please tell me I did something wrong. I'm only doing a test with the demo of a five minute clip from 60i to 23.976. I selected Best Quality and am deartifcating and deinterlacing at the same time. I wasn't sure what to select for the options (motion detection and detail pattern), so I moved the sliders both a little to the right, thinking I'd go for better quality. I don't know if it matters, but my timeline is 1hr even though the clip is only 5 minutes. I only have 1 gig of RAM on this machine (2 more coming). The render time is 131 hours. Even if I cut that by a third (with the additional RAM), that's 43 hours, and my movie is 40 minutes, so times 8 is 260+ hours. Is this typical? If so, I guess I'll give DVfilmmaker a try.


Oh...and about that HuffYUV codec...I downloaded and installed it like it says. But how do I select it? What are the exact steps? I looked in the export movie options and the Adobe Media Encoder options and can't find it.


All times are GMT -6. The time now is 02:11 PM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network