Restoring the Hollywood 10: Part Two

Blake Jones continues his story detailing the rescue and restoration of an archive of 16mm footage for an upcoming documentary about McCarthy-era Hollywood.

In my last instalment we documented how we found and recovered the film material from the 1981 reunion of actors, directors and writers who were blacklisted in Hollywood in the 1950s. We take up the tale again by preparing each roll for scanning. This process entailed manual cleaning using a film cleaning solution and ensuring that there was enough leader on the head and tail of each roll to aid in the scanning process. Once this process was completed, we were ready for scanning.

For the scanning process I used the Blackmagic Cintel G3 film scanner that did a fantastic job. I also needed fast and reliable storage for the data files. For this I used a direct attached SSD Thunderbolt storage connected to my MacBook Pro. I then dove straight in and started the scanning process. Since I pre-prepared all the rolls, this ensured that a smooth scanning process was followed. The rolls that were in decent condition were easy to scan.

I then came onto the rolls that were in questionable condition. I used the stabilizer in the scanner and when this was not enough I opted to fix the stability afterwards using the Resolve Studio application. Occasionally the scanner would stop when the film was too warped and I would then restart the process and re-assemble the scene later using Resolve. The entire scanning process was painless thanks to Resolve working together with the Blackmagic Cintel film scanner. 

Starting to scan the rolls

Each night I backed up all the scans in the hotel using a large capacity storage. Since we had a large amount of data, I let this run through the night while I was sleeping. I also made low resolution copies that I stored in the cloud. These were done mainly so that colleagues could see how the project was progressing. 

Scanning some of the warped film was a particularly difficult task as it was winding so loosely on the scanner. I had to carefully keep an eye on the film to make sure it didn’t fly off the machine. I decided to hold off on doing any audio syncing until I finished with the image restoration. I opted to use the original ¼” Nagra recordings for my sound source as the full coat magnetic sound tracks were not in optimal condition. 

Scanning warped film

Beginning restoration

Once back in our office in Europe, I started with the restoration process. I have always found that I handle image restoration differently with each project. This is mainly due to the fact that each project tends to have different problems that need to be addressed. 

In this situation I tried several different approaches and the one that seemed to give the best results was to first apply a basic color correction in the first node. In the second node I applied the Neat Video Open FX plugin. In this plugin I applied the dirt and scratch filter in addition to their contour correction. In the next node I applied the Resolve Revival Auto Dirt and Dust Open FX plugin to catch anything Neat may have missed. I found that the order in which these nodes were applied was critical to achieving the best results. 

Post and pre-restoration

I also had several rolls with an extreme blue stain caused by the Vinegar Syndrome bleaching. The problem with this was that it variably applied itself throughout the image. I tried various remedies and the best result was to simply isolate the blue with a qualifier. The blue stain could be seen in the scopes as changing mainly in luminance although there was also some lift change in this area. I found that by correcting this region using mainly the Gamma control and applying changes in Saturation gave a very good result.

Film passing through scanner gate

I next had to fix the image stabilization that could not be overcome by using the internal stabilizer within the film scanner. I found that using the standard GPU accelerated stabilization in Resolve did a very good job. However, in the cases where the film was extremely warped this was not enough. In this case I used the Classic Stabilizer together with the Point Tracker in Resolve. I tried applying track markers on the edges of the frame but this did not work out so well. I then noticed that the subject did not move during the interview. I then tried applying track markers to the shoulders of the subject and this worked out perfectly.

Moving onto the audio

Once I had all the images restored, I could then move onto the sound. My first step was to create Wav files for each of the audio tapes recorded during the shoot. My solution here was simple. I used a Focusrite Scarlett unit connected to a Nagra recorder. I connected the Line Out of the Nagra to the Scarlett input. I then connected the USB connection from the Focusrite to my MacBook Pro. The next step was to then use the Quicktime application to record the input of the Scarlett.

Creating audio files from each audio tape roll

This was followed by the laborious process of digitizing 104 audio rolls. Once finished with this I needed to start with syncing the sound and picture. This proved to be far more difficult than you could imagine. As there were attempts to edit the material following the shoot, the work print rolls had sections missing. This meant that the rolls no longer matched the original audio track as syncing the picture and sound using the manual clap, clap board or beeper would not work. My first solution was to find and mark the splice point in the picture. I knew this was where the footage was extracted. I then searched through the remainder of the take to make sure no other splices existed. If no other splices were there, I could try tail syncing. I searched for the last phrase spoken during the end of the take and tried to resync at this point. When successful, it should go all the way back to the splice point.

The next problem was that some rolls were excerpts and had no sync data at all. This meant that the picture had no clap boards, beepers or any other sync reference. So I had to formulate how to find out which sound roll matched and where the dialog actually was. I had the idea to use AI Lip Reading technology. I found a fantastic website called ReadTheirLips.com. It’s a paid service but certainly works well and definitely saved the day. 

I put all my audio files on DropBox which has a great transcription feature. I then loaded around ten seconds of video to the ReadTheirLips website. I created a small region of interest around the lips of the subject and started the process. Once completed, the text is then available. I then selected the audio file for the person being interviewed. By using the transcriptions created by DropBox, I was able to search for the words that the lip reading application has found. Once found, it showed me the timecode where it appeared. I placed this section over this portion of the video and just had to slide the sync a couple of frames. This process helped me through many problematic syncing issues which would have been otherwise impossible. Once all the audio syncing was completed, the footage can now be integrated into the rest of the documentary. 

Thoughts and conclusions...

What this project demonstrates is that you need to step back before starting the project and think carefully about the tools required to do the job. In certain situations you may find yourself in need of specific tools and a simple search can solve this. As a colorist with over forty years of experience, I have learned that you handle each project differently.

tl;dr

• The restoration process for the 16mm footage involved manual cleaning and scanning with a Blackmagic Cintel G3 film scanner, paired with SSD Thunderbolt storage for efficient data management.
• The scanning of warped film required careful monitoring and stabilization techniques using Resolve, with any challenging sections reassembled post-scanning to ensure quality.
• In image restoration, a systematic approach with color correction and specialized plugins like Neat Video and Resolve Revival was adopted to address issues such as dirt, scratches, and staining caused by Vinegar Syndrome.
• The audio restoration involved creating WAV files from Nagra tape recordings using a Focusrite Scarlett unit, followed by the meticulous digitization of 104 audio rolls to ensure fidelity in the final documentary.