Why Vengeance Trail is So Progressive: The Pain of Converting Analogue Filmmaking to Digital!

(Day 14 Counting Down to Memphis Premiere on April 23!)

Yesterday, I output Vengeance Trail at 24 fps (frames per second) to do a test screening. I played it from Vimeo on my 55" TV and it worked seamlessly. All I had to do was change the Sequence Settings from 25 fps to 24 fps.

YOU DIGITAL NATIVES HAVE IT SO EASY! This is not our origin story! Vengeance Trail was shot on Standard Def interlaced. Do you younguns even know what that means?

Instead of posting a quick update yesterday, I wanted to geek out with a teaching moment about how hard it was back in my day, LOL! Here's yesterday's belated blog...

Ian Max

TLDR: A primer on fields vs. frames. NERD OUT ALERT!

Films are edited at 24 fps and a frame is a frame. Film projectors actually double project each frame twice, to get the strobe into the persistence of vision range of humans. Think about the mechanics of advancing the film without tearing the sprocket holes, holding it locked while a shutter flashes twice, then advancing the frame again accurately. Now think about aligning two of these cameras and projectors for stereo 3D film projection. It's no wonder 3D needed digital capture and projection to work without giving people headaches. The vibrations!

CRT televisions required a raster to draw the frame twice, skipping lines, so a top (Odd) field and a bottom (Even) field. Who knows why they chose 30 fps back in the BW days? Oh, I remember, because NTSC is based on American AC electricity which comes into your house at 60Hz. So, cameras actually record at 60 fields per second and playback at 60 fields per second, but combining two fields into an interlaced frame becomes 30 fps. And don't get me started on changing the frame rate to 29.97 fps in order to squeeze in the color information. It took me years to realize that Drop Frame and Non Drop Frame were just different ways of counting the frames, the frame rate wasn't actually changing.

Nowadays, your digital monitor or television doesn't care if the video playback frame rate is 12, 15, 29.97, 30, 60, 120, 240, etc. It's so easy!

When we shot Vengeance Trail, DP Stephen McCurry chose a PAL prosumer camera, the Canon XL-1S because PAL is based on 50Hz AC electricity and its frame rate is 25 fps, which is pretty close to film's 24 fps and therefore once transferred to DVD would have a more filmic look.

How does 24 fps or 25 fps become 29.97 fps? I'm glad you asked. It's called The 3:2 Pulldown. 

So you see, in NTSC, frames 3 and 4 would have motion blur. 

Studio movies just repeat every 4th frame. 

Frame advance through a DVD and you'll see the pattern, every 4th frame is played twice, so 24 fps becomes 30 (ish - from now on I'm lumping 29.97 fps into 30 fps)). This method of repeating every 4th frame is a bit jittery. Since SD footage is actually captured at 60i

This time-saving factoid didn't apply to use, because we intended to burn NTSC DVDs! 

"PAL and SECAM television systems run at 25 frames per second. They are close enough to the film frame rate that 3:2 pulldown isn't required. One film frame is transferred to one video frame. The slight speed discrepancy makes a movie slightly shorter when it is transferred to video, but the speed variation is so slight it will not be noticed in viewing.

There is a method to convert 25 frames to 30 using a slightly different 3:2 pulldown to convert every 5 frames into 6 frames. In PAL, frames 4 and 5 (out of 6) would have motion blur. Unfortunately, Stephen converted all the footage before editing, and so locked in all sorts of motion blur. See a previous post about recapturing all the footage.

Don't forget about compression artifacts. If there's movement between two fields, it shows up as blur in a frame. And if you use JPEG compression on a frame with movement between each 60i field, the interlaced stripes become magnified.

Now you see why I sighed in relief when Topaz Labs added the de-interlace function to their Video Enhance A.I. blowup software? I don't know how they create good looking progressive frames, but I'm glad I don't have to ignore a field or blow up motion blur artifacts, and then reinterlace them to create a progressive frame.

You can definitely get into the nerd-weeds on JPEG compression.

And compression leads to generation loss, like when the first photocopiers would add dust to each copy and after a while a copy looked like it had black dandruff specs everywhere. Remember that? LOL. Remember when you recorded TV onto VHS tapes so you could watch a movie later? And if you copied a VHS, the edges would get blurry and colorful. Some weird people in Los Angeles have nostalgia for this. 

Again, you digital natives have no idea how good you have it. Even coding has become visual and intuitive; no more peeking and poking computer memory with machine code. Lucky! Ah, I've forgotten my roots. I think it's true, that I've forgotten more than my high schoolers have learned.

Out of curiosity, I exported the movie at 24 fps and 25 fps. Same timeline length of 01:23:17. Same exact file size. All I changed was the Sequence frame rate. But when I open both windows in QuickTime and hit play at the same time, there's no noticeable delay or difference over time. Then I frame advanced and 24 frames in the first equals 25 frames in the second. I don't need to find that dropped frame, but the sound track scaled perfectly. Now, to see if it matters to the theater when I deliver the DCP.

See you tomorrow with the post I started writing yesterday.

Ian Max

Western Authenticity in the Screenplay, Shooting and Editing of Vengeance Trail!

(Day 16 Counting Down to Memphis Premiere on April 23!)

AUTHENTICITY! Let's talk about seeking historical accuracy, and A.I. tools.

Vengeance Trail happens in the late 1800's in a fictitious west Texas town. Al Frisch and Keven Russell, the originators and keepers of the story, are Wild West aficionados. Al wrote the screenplay and played Mr. Robbins. Keven played Garvey and will attend the Premiere on April 23 in Memphis. 

At least one time on set, an actor thought a line sounded awkward and asked if he could say something more modern. Al would light into that person with a rant that could be heard back at base station, and Stephen the director would have to mediate and soften the blow. But the point was made: Al cared about the historical accuracy of his dialogue. Also, Al's wife Sue cared about the historical accuracy of the costumes. Attention to detail is evident all throughout the movie.

Two weeks ago, Keven watched the new Vengeance Trail, version 2.0 if you will. Here's what he said: 

The only issues I see, and don't know if they can be fixed now, is [some lines with hard to understand words] and in the scene when they are upstairs in town making a plan with Jim on the table and Jim says; "they stopped in front of the Sheriff's STATION," which is a modern term  and he should say OFFICE instead.

Normally, this line could be fixed with a recording studio session with the actor in a process called ADR or Automated Dialogue Replacement. Since the actor is no longer available, and would likely sound differently all these years later, I would collect thirty seconds of the character speaking in a similar tone from the same scene hopefully, as a voice sample, and upload it into a voice cloning software like PlayHT.

Then I would type in the new line of dialogue and do a comparison. The software gives multiple tries with the same line of dialogue, for different speaking rhythms, and now they've added emotions you can experiment with. I would likely then swap just the word "office" for "station," but to get the rhythm right I'd type in the whole sentence in order to get the one word right. I'd might have to swap the whole phrase for it all to sound right. Fortunately, in the edit there's no lip sync to worry about.

However, in this case, I went through all the takes and found that the actor used the right word once. Solution! Swapped it.

I know PlayHT works pretty well, because I already used it in Vengeance Trail. I'm waiting to see if the writer or producers notice, lol. An editor has his tricks and secrets, just like a magician. I did discover that I had to use a voice sample of the actor speaking with the same energy. If I mixed high-emotion dialogue with low-emotion, the output would be uneven and way off. But if I matched the same emotional energy in the sample, the output was actually quite usable. BTW, I think the lip sync will give it away more than anything, because I didn't get that quite right.

So, try out Play HT for yourself!

Ian Max

 

18 Days of Vengeance Trail - Countdown to Memphis Premiere on April 23!

Today's BTS insight: Key use of Video Enhance A.I. to blow up the original footage.

Before that, the gnarly task of recapturing all the old footage. Stephen and I couldn't find all the hard drives with footage and the EDL (edit decision list from the editing software). 

So, I borrowed an old DV deck and rescanned 70 Mini DV tapes. That’s 35 hours of footage. Then I cut those 70 clips into all the camera takes from set -- thousands of them -- and started matching them to the trailer. I couldn’t find all the footage, so I substituted shots.

Then I used Video Enhance A.I. software by Topaz Labs, on special for $150, to convert the footage from 480i to 1080p, a magnification of 250%. If you want more details to help for your own project, let me know. Check out this Vengeance Trail Up-res Comparison.

The last three years has been a labor of love, all for a library movie that's not proven to have found an audience yet. But everything in me says it's a decent movie with good production value, way better than our first DVD effort. And, I'm not tired of watching it. 

As Christian Toto just said: 

I am super excited to see Vengeance Trail on a big screen with a bunch of rowdy people, a shared experience the way cinema is supposed to be! Thanks for joining me on the journey.

Ian Max

P.S.  I'll be taking tomorrow off.