Film Research & Preservation

HEXLAB.tv
Hidden Experiment Lab

Hexlab is a nascent project dedicated to independent film research and preservation, focused on rare and overlooked media and related ephemera.

Features covered include:

High-quality film scans (35mm, 16mm, 8mm and Super 8) from a personally developed DIY telecine process
Rare VHS and Laserdisc recordings
Archival case studies on promotional film material
Creative experiments in media capture

More to come. Sample project below.

Capture Equipment

Custom-built Film Digitisation Rig

3D Printed 35mm, 16mm and 8mm Reels
ToneCarrier 35mm Film Holder
Cinestill CS-Lite (Light Source)
SmallRig Super Clamp with 360 Degree Ball Head Mount
Leofoto MPU-100 Universal Camera L Bracket
Hanimex E310 8mm/Super 8 Viewer (Repurposed Reel Arms)
Timber Board
Galvanised Steel Pipe + Flange

Sony FX2 - Camera
Canon 550D- Camera
Sony FE 90mm F/2.8 Macro G OSS - Lens
Tamron SP 90mm f/2.8 Di AF 1:1 EF Macro - Lens
Denon LA-3100 - Player (Laserdisc)
LG V271 VCR - Player (VHS)
I-O Data GV-USB2 - Capture Device
7Artisans EF to Sony E - Adapter
Canon LiDE400 Scanner
Samsung 4TB T7 SSD
Samsung 2TB T7 SSD

Software (used in preservation workflow)

Adobe After Effects
Adobe Premiere
Adobe Photoshop
Adobe Lightroom
Adobe Audition
Adobe Media Encoder
Adobe Bridge
AEO Light
VirtualDub2
Hybrid
Handbrake
OBS
MakeMKV
MKVToolNix

Supermarket Woman

Coming Soon

35mm Film Trailer Scan
Supermarket Woman (スーパーの女, 1996), Jūzō Itami.

Final 35mm film scan of Supermarket Woman trailer

Japanese auteur Juzo Itami, perhaps best known for his 'ramen western' Tampopo (タンポポ, 1985), had a tradition of appearing in short teaser trailers for each of his films — news-flash-style shorts known as tokuho (特報), in which he'd break the fourth wall and address the audience directly, like a reporter broadcasting live from the world of his own film. These teasers were cool little bits of promotional ephemera in their own right, essentially self-contained shorts.

I stumbled upon a 35mm release print of one of these trailers for his film Supermarket Woman (スーパーの女, 1996) in an online auction, and decided to snatch it up. All Itami's tokuho trailers included on their respective Japanese Blu-ray releases seem to only exist in standard definition (with the full films appearing in 1080p). Which got me thinking about whether I could personally create a 4K scan of this one. Sending the film to a professional telecine service, as awesome as they are, seemed impersonal and expensive. So a DIY option seemed right to explore — a worthy act of film preservation if done well, or a fun experiment otherwise.

In this particular trailer, Itami is seen loitering in a supermarket in an incognito mode style hat and trenchcoat. He immediately gets close to the camera and asks, "Have you ever shopped in a supermarket like this?" — then launches into a rapid-fire diatribe describing the shady cost-cutting practices supermarkets engage in to the detriment of customers. The film is as anti-capitalist as mainstream 1990s Japan could be. Which is to say: just kinda. Definitely critical, with a socially harmonious "consumer activism" bent.

After looking into the world of DIY 35mm cinema film scanning, I felt incredibly inspired by Rob McLaine’s beautiful preservation projects, and encoured to try it myself after watching Al Warner's scanning and post-processing/audio extraction methods. The idea of using an old, dusty lump of physical polyester to assemble each frame into high-quality, smoothly flowing footage was incredibly exciting to me - and extracting the audio from the optical soundtrack visibly embedded in the film seemed completely magical. But I also noticed modern film scanners' ultra-impressive setups seemed gargantuan — seemingly occupying an entire room of their house; messy, with cables, wires, machinery, spools, and computer chips strewn about everywhere; and very expensive.

From my experience in image and video editing though, the theory behind how everything was captured and processed made complete sense to me, and seemed vaguely achievable without the hassle or expense. I decided to try building my own smaller, cheaper DIY film digitisation rig (more like the kind still photographers use), and see if I could use it while applying my knowledge of motion picture film. Essentially, the setup I built is a chunk of timber with a galvanised steel pole sticking out of it (to see a complete list of all the pieces involved, check out the 'Capture equipment and software' list above.

Then came time to test it out. I started with an old Canon 550D camera and Tamron 90mm macro lens I had lying around, using an iPad as the light source — but this quickly proved unworkable. The iPad's individual pixels were visible once photographed, and after cropping, the camera could only achieve 2K. I wondered whether the iPad could still work as a light source if I defocused it by suspending the film farther away, but space constraints ruled that out, so I changed direction. Still, it was a useful first test and a chance to capture a few frames to run through the whole process end to end before committing to a proper setup.

Early 35mm film scanning + optical soundtrack extraction tests

I decided to start from scratch, switching to a Sony FX2 camera with a Sony 90mm macro lens and using a Cinestill CS-Lite as the lightsource. I captured each frame individually with the new gear, using an air blower to remove debris as I went. Despite this, a lot of imperfections wound up in the captures. I used my photography retouching skills to remove a substantial amount of fluff, scratches, and other gunk — particularly when they seemed extra distracting (i.e. thick curly strands of hair sitting directly over someone's face). This retouching step was surprisingly relaxing, and made me feel like I was faithfully restoring an old chair or conserving a historic painting in an art gallery. I decided to deliberately leave a lot of the imperfections in though, as I enjoyed the 'warts and all' unvarnished tangibility it gave the image quality.

I then stitched the frames together in Adobe After Effects at 24fps to create the footage. Because each frame was manually wound into place for photography, the positioning was inconsistent and imprecise, resulting in the footage jittering around the frame. To stabilise this, motion tracking was applied using the perforation holes as an anchor point. The film had faded to a crimson hue over the years, so each shot was then colour-corrected using the standard-definition Blu-ray transfer as a reference.

Finally, with the moving footage constructed, I brought the optical soundtrack, now resembling a moving audio waveform, into AEO Light to extract it and turn it into audible sound. AEO Light was developed by a group of mathematicians, film preservationists, computer scientists and coders from The University of South Carolina, and made freely available. Pretty cool. The idea of this stage had initially seemed so magical to me, but in practice I found it to be the biggest pain in the arse.

I started by cropping in on the soundtrack and adjusting the colour saturation, brightness and contrast as I'd read this can help the program interpret the audio more accurately. Despite my precautionary efforts, the resulting audio had a consistent raspy, fluttering distortion throughout it. After many hours of troubleshooting (e.g. tweaking various capture settings, adjusting the exposure/contrast/colour levels, using a TIFF sequence rather than an uncompressed MOV, correcting a slight rotation in the waveform, etc.), it turned out to be an issue with where I was setting the pitch start and end points. Despite the pitch sounding roughly correct in the problematic outputs, tweaking the values ever so slightly eventually produced a much cleaner-sounding file free of raspy distortion. I had assumed problems with pitch would just manifest as the voice sounding either 'demonic' or 'chipmunk-like' – but evidently it can be more complicated than that. Once fixed, the audio was then applied to the footage, and voilà.

You can see the final result in the accompanying footage. Please note that this is not the final high-quality 4K crop (stay tuned for this in the future iteration of this page), but rather a version including the film perforations and optical soundtrack — which I thought would be interesting to present — and only 6.5mbps to keep loading times manageable.