16mm Film transfer image enhance

WayneM wrote on 5/10/2011, 9:02 PM
I did several searches and didn't seem to come up with anything on what I need. Probably means I've missed something.

I have a 40-year old piece of film I shot that was edited from original. I've transferred to DV and now have in Vegas (8.0C). It is mixed color and B&W. Some of the B&W looks good if projected but the transfer to DV (with limitations I have) doesn't have the dynamic range. So I'll be using curves (I think) and probably some color correction.

Is there a trick or best practice when applying the filters and setting the points of how to take a filer from one setting to the next immediately. It seems I did this before but when I set a single point at the end of the area when I want a constant "filter" setting applied it seems I have to set another point and set that either to "null" settings of whatever the next correction is. I hope I'm clear or at least that you can figure out what I'm asking.

Essentially what I would like to do is set a single point of the effects timeline for a filter for each edit-point/scene change. Then I could go back and adjust the image in each and save the settings to be in effect for that scene only.

Any help appreciated!

Comments

musicvid10 wrote on 5/10/2011, 9:10 PM
Look for previous threads by user johnmeyer on this subject. He is our resident expert on restoring film transfers, and we are lucky to have him.
Serena wrote on 5/10/2011, 11:26 PM
I think you are asking about setting a keyframe at each splice and adjusting the filters at each. You can do that but I think it very tedious. Doing the same work I split the video at each splice. If the material is consistently exposed then you can develop the filters and copy and paste event attributes to each similar clip. Presumably in this case the material was not evenly exposed, and copying the filters (and making individual adjustments) is much easier than adjusting the filters at each key frame. You will want to remove colour information from the B&W clips and use curves for grading those. The colour clips will be more involved (as you know!).

EDIT: when I say "at each splice" I assume that the film was edited. Perhaps I really mean "each clip".
farss wrote on 5/11/2011, 4:32 AM
Just a guess but maybe the key the OP is looking for is to R Click the keyframes and change the type to Hold.

Bob.
WayneM wrote on 5/11/2011, 5:40 AM
Thanks much MusicVid. I will look for those postings.
Serena wrote on 5/11/2011, 5:54 AM
Bob, that's one of the things that one has to do if treating the film as a single clip, and part of the tediousness of treating it in that way.
farss wrote on 5/11/2011, 6:00 AM
TBH, I was just focussing on this part of the original post; "Is there a trick or best practice when applying the filters and setting the points of how to take a filer from one setting to the next immediately"

Sounds like he might be apply the FX to the track. I think your suggestion of applying them to each event is a better idea. He's split the clip into events anyway and doing it your way avoids the whole keyframing issue.

Bob.

WayneM wrote on 5/11/2011, 6:13 AM
I had never thought about doing that split of the video. It might work very well for this project which is an image montage to a music track. Certain long takes are split up and sections appear at different places throughout the film. I could split the video at the frame showing the film splice.

I was looking through the documentation, but have not been able to find a way to split individual tracks such as video without also putting a split in the audio tracks. With a 16 minute project upwards of 100 individual splices, having splits in all the tracks, including audio, is a nightmare waiting to happen. I've had that happen before where I move one item and miss another item off screen and discover it later. I'm running Vegas Pro 8. Maybe the kind of individual split in a single track goes by another name? I'll do some searching.

I may want to remove color info on the B&W, but I was shocked that the transfer I did through a Buhl aerial image multiplexed, gave me very clean B&W. This is a test transfer as the camera used was a single-chip Mini-DV but at least with Schneider optics. (Actually the results of the "free" transfer may be so good as I won't bother with an expensive one from DigitalTransferSystems.net.)

Your technique might also be useful because I have been thinking of tinting/toning the B&W to see if the effect.

Thanks for sharing the creative techniques!
rs170a wrote on 5/11/2011, 6:17 AM
I was looking through the documentation, but have not been able to find a way to split individual tracks such as video without also putting a split in the audio tracks.

Click on the video track so that it's the only thing highlighted.
Press the 's' key and only the video track will get split leaving the audio untouched.

Mike
WayneM wrote on 5/11/2011, 6:24 AM
Bob,

BINGO, I knew that was in their somewhere but I wasn't dreging it up late last night. So I set the HOLD on the keyframe at the beginning of the film segment and the settings will stick, right? I looked at that last night and saw Linear, etc and must have looked right over Hold!

The Split technique also sounds useful, but I'm afraid of not having something locked and creating a train wreck :-) Since I may only need 8-20 different settings I couold just go with the keyframe and Save the named settings.

BTW, I did a compilation project with video sourced from high-end DV to mobile phones a coouple years and and discovered a bug/quirk in rendering. I had dona a lot of effects correction to the clips and it rendered beautifully for a DVD. But when rendered for YouTube according to all the info here and at YouTube, the color correction was offset by several frames! It was very noticable and repeatable. I also reported it and never got a response. Again, I'm running V-Pro 8.0C and don't have budget for an upgrade unless there is a special deal.

Thanks for the help, now I have options to tackle this project thanks to the great folks on the Forum!
WayneM wrote on 5/11/2011, 6:31 AM
Actually I think you were right. This is a single transfer to DV of a hand edited 16mm film. It wasn't an A/B roll so the video is a single track and event. I began with Track Effects and the Hold idea would let me go forward that way.

It may not have been clear in my first reply, the the glitch I ran into in rendering for YouTube (MPEG? it has been a couple years) was with using Track Effects to manage the color correction. I never thought to apply the color/exposure correction of those DV clips at the event level!

Thanks. . .

"A mind is a terrible thing to waste. . ."
farss wrote on 5/11/2011, 6:33 AM
In addition if you hald down the Ctl key while you click on the event to select it the playhead doesn't move. Took me a couple of years before someone here taught me that trick.

Bob.
rs170a wrote on 5/11/2011, 6:37 AM
The Split technique also sounds useful, but I'm afraid of not having something locked and creating a train wreck :-)

Each time you do a split, click the newly split video event, shift+click the audio event and press the 'g' key to group them together.
No more train wreck :-)

Mike
WayneM wrote on 5/11/2011, 6:37 AM
Thanks Mike. I could swear that was exactly what I did last night. In my case I was trying to split the two audio tracks without putting a split in the video. I ended up trimming back the two audio tracks to the edit point, then adding the full tracks in again, editing the front ends to the edit point (skipping a section of the music track) and butting the events. It worked fine. . .but tedious :-)

I'll try that again!
WillemT wrote on 5/11/2011, 6:41 AM
You can right hand click the audio track and select Switches/Lock. That will prevent the audio track from being split along with the video.

Willem.
WayneM wrote on 5/11/2011, 6:44 AM
You are kidding me! I've been doing projects on and off since Vegas 1 and NEVER heard of that tip! You can only imagine the number of profanity bursts and Crtl-Z's I have expended.

Another Post-It on the monitor!

THANK YOU BOB. . .

And Thanks to the Forum respondents who have ALL provided spot-on suggestions and insights.
WayneM wrote on 5/11/2011, 6:53 AM
Willem,

Thanks for that VERY useful tip. I have seen the LOCK option but never delved into it. Actually I hope this works for any track because I may want to lock the VIDEO which is the film transfer (no sound/SOF associated) and I need to lock that puppy down tight.

I did most of my editing with Vegas 4 and have upgraded a couple times over the years and am now at Pro 8.0C . As such I seem to have missed several "What's New In. . ." version update documents which I used to always pay careful attention to.

Thanks again!
johnmeyer wrote on 5/11/2011, 9:12 AM
You've had responses from the two people who actually worked with film professionally, farss and Serena. While I took my first movies in the early 1960s, these were strictly amateur movies. My only claim is that I've done a lot of film to video transfers, and built some equipment to do these transfers.

Splitting the video into events is most definitely the way to go. When finished, you select all the video and audio and re-group them. So, I agree with everything that others have posted.

FWIW, here is, more or less, my workflow for dealing with film. Sorry that it is so long, but it is a reasonably complete explanation of what I do, and as most people know by now, I am genetically incapable of a short post.

1. Set the header on the AVI file so it has the correct playback speed. All my transfers are "frame accurate," meaning that I have one frame of video for each original frame of film. If I were to play this back at 29.97 fps or 25 fps (PAL), my 16, 18, or 24 fps film would play back too fast, so when rendering to DVD, I have to add "pulldown."

2. Match the project properties to the film. This makes sure the fps set in #1 matches. At this point I also set the project properties to progressive. I also check that all media I have imported (i.e., the film captures) are set to progressive. This step is necessary for me, but may not be for you. I have to do this because I have always been capturing using the DV AVI codec, and this always gets reported within Vegas as interlaced. Since there is no temporal difference between the top and bottom field in a frame-accurate film transfer, it can -- and should -- be handled as progressive. Whether YOU need to do this step depends entirely on the method of film to video transfer you used (see my end comment below on what to do if you don't have frame accurate film transfer).

While I am changing media/event properties, I also "Disable Resample" for each event (I have a script which does this). I do this because I want to retain the "purity" of the film and not create any blends if I do speed changes. Also, when I do the final render, I want Vegas to create real "pulldown" (going from 16, 18, or 24 fps progressive to 29.97 interlaced), and it will not do this if "smart resample" is selected.

3. I put all my film on the timeline, in roughly the order I expect to use it. I refrain from doing any editing at this point.

4. I serve this footage out to the Debugmode frameserver and then feed this into an AVISynth script that detects scene breaks. This saves a HUGE amount of time. This script runs at almost 180 fps on my computer, about 10x the speed of the original film. It outputs frame numbers to a text file. These numbers are the frames where the scene change occurs.

5. I temporarily change the ruler on the Vegas timline to "Frames," and then copy/past the frame numbers from the text file into the Edit Details Markers window. I then have markers at each scene change.

6. I use a simple script to actually split the events at each marker (scene change). This script only splits events on the first selected track. If I also want to split the audio at the same points, I run the script a second time, but this time with the audio track selected.

7. I assign a "filter package" (this is a Vegas feature where you can bundle together a series of fX and assign them all at once -- including presets) to the first event, and then copy/paste attributes so that these fX are on every single event.

(BTW, I should point out that I save the Vegas project after each step, and after each step I increment the project file name -- e.g., Niagra_Falls_01, Niagra_Falls_02, etc. -- so that I can easily go back to the previous step).

The fX in this package are Color Curves, Color Corrector, and Levels. The first one has a film gamma curve I created years ago and which always seems to get me pretty close to the right video gamma. This curve does most of its work in the first 1/4 of the horizontal axis, and gains the shadows, but leaves the pure black levels alone (i.e., I don't move the 0,0 point).

If someone knows how I can export this preset so someone else can use it, let me know and I'll post it. Here's a screen grab of what it looks like:



The other two fX are set to their default (i.e., they do nothing). The purpose for including them is to eliminate the need to add them later on. This saves a lot of steps. (As an aside, it would be interesting to hear various people's recommendations for the easiest/fastest/"best" way to add an fX to an event -- you can click on the button; drag them from the fX corral; copy/paste attributes; and a few other ways, I think.)

8. I then color grade the footage. You DEFINITELY want to do this before you start editing. You may want to create a few presets as you go along because some scenes will be similar to earlier scenes and you don't want to re-do everything from scratch. This is usually the most laborious part of doing the film editing, and is where I wish I had someone like Bob (farss) or Serena looking over my shoulder to tell me what "best practices" in the film industry are. I've just had to do this by the seat of my pants.

9. Finally, I edit the footage (yeah!!). So, all the prior steps are unique to the film process. At this step, I can begin to treat the footage just the same as I would anything captured by a video camera.

10. I then frameserve my edited project out through Debugmode frameserver into a restoration script. You can read about this here:

The power of Avisynth: restoring old 8mm films

and view the author's result, using pristine film:

http://www.vimeo.com/13173031

and my result, using absolutely horrendous film that had been stored in a basement for fifty years, outside a film can (yes, I actually DID clean this film prior to capture):



11. In the final step, I render to my final file format and, if the result is to be put on DVD, I add pulldown (i.e., duplicate fields so that the result will play at NTSC standard 29.97 interlaced frames per second). Vegas will do this (add pulldown) if you disable resample for all events, but I like the control I get doing it through an AVISynth script, so once again I serve out of Vegas into an AVISynth script, and then render to MPEG-2 for DVD out of that script.

I've never posted my entire film workflow before. I hope this helps someone. I have many additional tricks and tips for dealing with specific problems. Just ask ...

John

P.S. I almost forgot to mention something alluded to above: if your film to video transfer is NOT frame accurate (e.g., it was transferred by pointing a video camera at the movie projector screen), you may be able to turn it into a frame-accurate transfer. Someone just posted about this in the film transfer thread at doom9 that I referenced above, and a few of us got intrigued and collectively came up with two different methods for dealing with (and eliminating) both the blended fields and the flicker, both of which are the inevitable, unwanted by-products of this inferior method of transferring film. Here is a link directly to the point in that long thread where this problem is first described:

Recovering Original Frames

The script posted by Didée did a remarkable job on the flicker. You will have to tinker with the "SRestore" framerate settings until you get exactly the correct decimation: too slow an fps and you'll leave too many blended frames; too fast, and you'll start to drop frames.

Download the before and after clips; the results are (to me, at least) pretty amazing.

WayneM wrote on 5/11/2011, 11:44 AM
John,

Saying Thank You seems inadequate. So THANK YOU VERY MUCH! And I hope to be able to return the favor some day. Your comprehensive sharing of this information is incredibly helpful. I think I will capture it to a PDF so I don't loose track of it. Until you mentioned it I hadn't yet thought about the interlace issue. I'll have to check the project properties! I haven't done anything with Vegas since I migrated to HD at home. Been in the dark ages so to speak.

The transfer was to a Sony SD Mini-DV camera. I assume the video has to be interlace and will have to check. Actually there would be no reason the format on the tape would have to be interlace and I don't remember noticing that in some high-speed pans. Hmmm.

Transfer was thru a Buhl Aerial Image unit which I aligned and cleaned carefully. Projector had a true telecine shutter, so no flicker at all and the light source had the translucent white "ruby glass" diffuser so there is pretty even lighting. Anyway, if I get this proofed out I may do a higher-end transfer to HD or opt for the "Spirit" transfer at $.66 per foot. But that's way pricey for my budget here.

For the transfer I made sure I got all the frame edges which I will crop out using Motion control settings. For my need the slight loss of resolution is better than something cut off in transfer.

I will read your detailed info later, but I guess the thing about events, when indeed the video is actually a single event, is to put in splits so I can apply CC (or timing correction fm the film days :-) to individual scenes.

I wasn't sure the projector was "perfect" in terms of speed so I did a manual count of how long 5 seconds according to my verified film Academy Leader ran. It showed that applying a playback factor of 1.004 to the event would make it perfect. I did disable resampling. That is a small enough error and a I can account for it when I recreate the music track that is the sole sound track. I may just reset that to 1.000 to avoid adding any other artifacts at this test point.

Thanks one more time for this valuable reference for the whole community!

Wayne
WayneM wrote on 5/11/2011, 1:27 PM
Wow, there is some interesting/amazing info on the Doom9 thread. I've printed it so I can highlight things. Is this specifically associated with software that is purchased?

My transfer is OK for this first test, which was seeing if I could recreate the edit of the sound track. My film was two reels, one picture and the other one-channel of sound on 16mm perfed mag film.

I did have a couple questions. I just looked at my project and the video transfer is 720x480x24 at 29.970 NTSC. So I'm guessing that has to be interlaced. In my project properties/Video what would you recommend as the Template to choose? Should I choose one of the HDV templates? I have never before ventured into this territory.

The other question is whether once I have split and event, is there any way to select the area and rejoin?
johnmeyer wrote on 5/11/2011, 4:29 PM
Is this specifically associated with software that is purchased?Most everything discussed over at doom9.org is about open source, free software.


I just looked at my project and the video transfer is 720x480x24 at 29.970 NTSC. So I'm guessing that has to be interlaced. In my project properties/Video what would you recommend as the Template to choose? Should I choose one of the HDV templates? Yes, that is interlaced footage. However, in my long post above you will see that I change everything to "lie" to Vegas to tell it that this video that is being reported as interlaced is in fact really progressive. I must do this because when film is transferred by a frame-accurate method (using a Rank Cintel or Workprinter, for instance), each frame of film is transferred to one frame of video. Put another way: there are no frames of video which contain "blends" from two different frames of film, something that WILL happen if you just point the video camera at the screen while projecting a film. Since both fields of video come from the same frame of film, both fields are recording the same exact instant in time, namely the instant at which that frame of film was exposed. Therefore, even though the video camera is recording the top frame at a different moment in time than the bottom frame, since the thing being photographed (the frame of film) hasn't moved, the result is exactly the same as if the video was progressive: there is no temporal difference between the two fields. Thus, you need to tell Vegas that this is really progressive, and this needs to be done both for the media (in the media or event properties) and for the Vegas project properties.

There is no preset template for this, although you can create your own.

I don't know why, however, you would consider using an HDV template since your source video is 720x480 29.97 (i.e., standard definition NTSC video). Instead, start with the NTSC 4:3 template, and then make the change to progressive, but only make this change IF your film was transferred with a frame-accurate system. If it was not, do NOT change to progressive.

The other question is whether once I have split and event, is there any way to select the area and rejoin?I don't think so. The usual advice on this request (which many other people have made) is that you delete one event, and then grab the edge of the other event and lengthen it to fit the hole. This is a few more steps than just hitting a "join" button, but it accomplishes the same thing.
WayneM wrote on 5/11/2011, 7:20 PM
Thanks much John. I was assuming (and we know what that does) that NTSC implied both the color standard and interlace. I'll be experiementing with that tomorrow after I finish highlighting the printed version of your detailed post.

I also decided to bite the bullet and, with the special deal this week, upgrade from Vegas Pro 8.0C to 10.0D, although I may not install it until after i finish the first couple passes on this project.

I am trying going with splitting the video and using event effects for the correction. That does seem to be the better plan. Over 100 events once broken up. But I did a sort of ready, fire, aim, and I guess I will now have to tound each of the events to change the property to progressive.

Again, thank you.

Wayne
johnmeyer wrote on 5/11/2011, 8:30 PM
Over 100 events once broken up. But I did a sort of ready, fire, aim, and I guess I will now have to touch each of the events to change the property to progressive.You can change this in the media pool, and it will propagate to all the events.

If you know how to use scripts and can find a copy of an old script called "Flicker Buddy," it will let you change all the resample properties with one command, although I now use Excalibur for this (and many other) functions.
farss wrote on 5/11/2011, 8:53 PM
Just my wo bob's worth.

If it was me i'd just "find" the money to get the film scanned right in the first place. Given that it's 16mm it was probably reasonably well shot so it's probably worth doing right from the get go. The cost of film transfers on good gear seems to have come down a lot recently. I've been offered a couple of 16mm jobs over the past 18 months and I sent them off to a local post house with a Cintel to be done at standby rates.

Getting a 2K transfer gives you a lot more image data to work with compared to DV.

Bob.
bsuratt wrote on 5/12/2011, 8:43 AM
I second Bob's suggestion that you get a pristeen transfer via Rank-Cintel first. You may not need much, if any, correction beyond that.

16mm .14ft at:
http://www.mymovietransfer.com/scanning_process.html

I have used them for 8mm and the quality is second to none!