"Retro Film" - Testing Topaz Starlight AI Enhancer with Super-8mm film

 

I have been testing AI Enhancers for a year now and until recently, March 2025, I have seen poor results with black and white, and with film grain. A common result was no change, with or without an error message. But now, surprise! a dramatic transformation from Topaz Starlight, a "research preview ... first-ever diffusion AI model for video enhancement".  At the time of writing, 13 Apr 2025, it is open for free testing with limited amounts of footage. That comes with a condition that they can use uploaded footage in their research. Therefore the performance is from ugly old me and a much better looking cat. IMO Topaz Starlight is a little "over the top" in its changes and if the release version has a control for something like degree of enhancement then I would be into "dialing it down a bit". 

I can see 2 uses:

Old films.
When we were filming with Super-8mm in the 1980s it was because it was accessible and affordable for telling our stories. IMO it is valid with some of these fictional drama films to go for enhancement. I am thinking minimal or no enhancement for our documentary material.

New Retro Filming.
We are having more of a debate with this! Is there a point to analog filming then having it change to look digital - or does it? Is there a case for saying that this hybrid of Super-8mm and Digital Enhancement gives a distinctive new look with its own new artistic validity? Is it good artistic revenge on filmmakers who fake the film look with digital tech, to fake the digital look with retro film? Our actors, documentary subjects and crew are enjoying working with vintage cameras, and they may enjoy it more with enhancement.

See the Youtube description and additional clips for more about camera, stock, chemistry and cat.

"Retro Film" - Processing colour film as black and white

In learning "retro film" we do tests with stills cameras. It is a quick way to test different films and processes. So these still photos really do relate to film-making.

I can sometimes buy 16mm colour shortends, the unexposed remaining parts of big rolls of film. The colour process is a big challenge and I am focussing on black and white. How does processing colour film as b/w work out?

I got my Kodak Retina 1B out of storage. This is a classic camera from about 1954. It had a partly exposed roll of Kodak Gold 200 in it. I shot some more then processed as black and white in Ilford LC29 diluted 1+19 for 13 min at 20 degC. A usual time for LC29 is 8 min. I read somewhere to give more time to colour film and that seems to have worked. 

Now on the TODO list, film something creative with those shortends! As in short comedy sketches which act as team-building test runs for our mad idea of competing in the 48 Hours film-making competition using retro film.

"Retro Film" - Double-8mm with "Double Bohemia"

 

Getting to know Double-8mm by testing with Double-8mm film made by Foma in Bohemia. So of course we went 10km to film in Puhoi, NZ settled by immigrants from Bohemia. Hence the title "Double Bohemia". AND we are testing a Meopta Admira 8E camera made about 1949 in Prerov which is near Bohemia. So does that qualify as Triple-Bohemia?!

I bought the Admira for its removable lens which I assumed was a “D-Mount”. And the threading was correct for D-Mount. It accepted a C-Mount to D-Mount adapter and I tested C-Mount lenses from 16mm cameras, and also a C-Mount CCTV lens. On developing the film – shock! Horror! There was only about 20 seconds of movie and some single frame stills – all taken with the original lens. All the interesting other lenses were extremely out of focus. After some detective work I found a strange sad answer. The D-Mount flange-to-film distance standard is 12.29 mm. On the Admira it is much further, 17.3 mm. Why?! I can only guess that the “communist bloc” worked to different standards but a different thread would have been nice. In the “wise after the event” department, the lens is unusual for having threaded structure going a long way, about 9mm, from the flange into the camera.

The film is Fomapan 100 R designed for reversal processing. I processed as negative because I needed a quick result for planning my next “retro project” move.

I used Ilford Ilfotec LC29 at 1 + 19 dilution for 5 minutes at 20 deg C.

Here in NZ the reversal chemicals are difficult to get, especially during the Christmas/New Year southern hemisphere long summer holiday period. And we have no reversal-capable cine film labs here – the nearest is in the US. Therefore this is not a fair test of Foma film. What info we do get is:

* Processing Foma R as negative is possible.

* The anti-halation layer acts like a 4.5 stop neutral density filter. My homemade digitiser was able to cope with that because it works by taking a photo of each frame with a digital camera. That camera normally runs with a shutter speed of 1/60 sec. In this case it was setting speeds like 1/2 sec. Question: can commercial digitisers handle that big a variation?

* The old Admira spring motor means that I do not know the shutter speed for the bracketing single frame shots. I did light metering assuming 1/20 sec and ASA 50 indicating f/2.8 as the best setting. The result is very different! My digitiser camera is automatically adjusting for negative density which is why the results for f/5.6 to f/16 look much the same. The fact that they look much the same indicates good exposure latitude.

About the Meopta Admira 8E

* In my opinion this has a non-standard design that works against the interesting option of using a variety of D-Mount lenses and other lenses with C-Mount to D-Mount adapters.

* In this test we have random scratches on about half of the film

* The gate and transport look like high quality engineering. However the gate wiggle effect is no better than seen in Super-8. Especially look at the opening shot with the Bohemian Museum road sign. I had the Admira on a tripod and operated with a cable release for max camera steadiness to test for gate wiggle. I am guessing that this particular Admira has had a hard life. A lot can happen in 75 years.

The Admira lens does not have a focussing ring. It relies on Double-8mm having a long depth of field. I am sceptical of this and I am going to suggest that a focussing ring is a must-have. The distant shot of the roundabout appears to have better sharpness than the test chart which was at a distance of 6 feet (1.8 m). I have also been testing wide angle C-Mount lenses on a mirrorless camera and it is showing precise focus changes with only small movements of the focussing ring.