Modding the Kodak Reels 8mm Film Digitizer (Firmware Hack)

[0dan0]

Funny story... It seems I've 99.9% recovered my Type-A. I soldered on the new SD Card connector, but discovered the missing pad is the write protect switch signal. I tried connecting to some of the nearby traces, with no luck, unit would report SD was write protected. However, as the unit could now read the SD, I could update the firmware. So I wrote firmware to ignore the write protect switch, so it now records too.

Some lesson learned, "SD" stands for secure digital, turns out it is not very secure, as the write protect switch is completely ignorable if the driver wishes. When I enabled the write protect switch on the SD card, my Type-A Reelz can happily write to it. The switch does nothing, other than provide false hope that your data is "secure".

 

[0dan0]

Here's the video reposted, hopefully it works for you @0dan0. I'm not sure that I'm mechanically inclined enough to solve that problem right now. I think I found some of the code related to the motor function, at least potentially where it was choosing between the super 8 and 8mm switch. But that turning mechanism is a little above me. I can look at it and understand what it's doing, but I have no clue how to change it to increase the distance it goes. I'm guessing I'd have to pull that plate out completely to figure it out.

I would be interested in any code you have found regarding the motor functions.

I didn't see any sprocket movement difference between Super8 and 8mm, even though the there is ~8.5% difference in the sprocket gap (~3.78mm - 8, ~4.1mm - S8). So it is possible that just a cam change for 16mm would work.

 

[fishgee]

Here's the video reposted, hopefully it works for you @0dan0. I'm not sure that I'm mechanically inclined enough to solve that problem right now. I think I found some of the code related to the motor function, at least potentially where it was choosing between the super 8 and 8mm switch. But that turning mechanism is a little above me. I can look at it and understand what it's doing, but I have no clue how to change it to increase the distance it goes. I'm guessing I'd have to pull that plate out completely to figure it out.

Hard to be sure from looking at the vids, but it doesn't look like you have the extra room in the shuttle mechanism to put a larger cam in there. You might be able to grind away some metal on the left of the shuttle to give you the extra room you need. It looks like they are only using the spring tension to pull the arm back to the left after the cam pushes it to the right.

 

[0dan0]

Hard to be sure from looking at the vids, but it doesn't look like you have the extra room in the shuttle mechanism to put a larger cam in there. You might be able to grind away some metal on the left of the shuttle to give you the extra room you need. It looks like they are only using the spring tension to pull the arm back to the left after the cam pushes it to the right.

The cam is not a circle, so the 1.5mm would on the lump side only. You could also only increase the lump only 0.75mm, if you decrease the opposite side 0.75mm, it is the difference in the displacement at different angles that creates the motion path.

 

[DocMorris]

Wow it´s incredible what has been done here! I´ve got a C model and I'm scanning with 0dan0´s Version 5.7 at the moment. I absolutely love the enhancement!
I have been scanning almost 4h of film now, so the machine ran about 20h in total and lately it started to make some loud noises. It seems that the friction of the film transport parts in the middle increased. Shortly after, it was the same loudness as before and sounded healthy again, but I´m afraid of it getting worse again.
Has anybody experienced something similar and got tips on what to do? I would rather not open it up and thought about spraying some WD40 inside where the transporting needle comes up, but wasn´t sure if that's a good idea.

P.S. I really like the mostly manual exposure and got some decent results with it! But in some darker clips it was jumping some times. It would be fantastic to have some kind of scene cut detection so it only measures and sets the exposure once per shot.

 

[Brionco]

I would not recommend WD40. It probably won't react with the plastic like a petroleum lubricant, but it also might adhere to the film. It is hard to say how it might react with the acetate over time.
At work we scan 35mm sound opticals, nitrate, acetate, everything. The best we have found is to clean the contact area thoroughly with denatured alcohol (if it available where you are), or as pure as you can get.
We use Tri-Flo (https://www.amazon.com/Tri-Flow-TF0021060-Superior-Lubricant-Bottle/dp/B0002ITTK2/), but any teflon, non-petroleum, lubricant will work. I cannot overstate how little you need. A drop or two on a cotton swab. With 8mm it is going to hard to keep it only to contact area, but I don't think getting on the entire gate is going to be an issue.

CAVEAT: All of our machines have stainless steal rollers. The film never touches plastic or nylon. It would be a good idea to do a test on the side of the scanner (assuming it is the same plastic everywhere). See if a cotton swap comes away with any black after a day.

 

[0dan0]

Wow it´s incredible what has been done here! I´ve got a C model and I'm scanning with 0dan0´s Version 5.7 at the moment. I absolutely love the enhancement!
I have been scanning almost 4h of film now, so the machine ran about 20h in total and lately it started to make some loud noises. It seems that the friction of the film transport parts in the middle increased. Shortly after, it was the same loudness as before and sounded healthy again, but I´m afraid of it getting worse again.
Has anybody experienced something similar and got tips on what to do? I would rather not open it up and thought about spraying some WD40 inside where the transporting needle comes up, but wasn´t sure if that's a good idea.

P.S. I really like the mostly manual exposure and got some decent results with it! But in some darker clips it was jumping some times. It would be fantastic to have some kind of scene cut detection so it only measures and sets the exposure once per shot.

Thank you for the feedback. You should switch to V5.8 as the also fixes the missing full range flag in the video header (more dynamic range without color correction, or it saves you the ffmpeg step.)

The crude manual-ish exposure control is still a work in progress. There will be more tweaks coming, but a good scene change detector is unlikely, and doesn't solve for all scenarios (like fades from black.) I'm find I need different exposure tuning for home movie vs commercial content.

No idea on addressing mechanical noise issues.

 

[0dan0]

I might need help to find the trigger for the motor step function (and/or the frame grab function.) Even on the stock lens and firmware, you will occasionally see some instability at the bottom edge of the frame. This is a rolling shutter artifact, as the frame is being grabbed, at nearly the same time the film is starting to be moved. You can see this as unwanted motion on the bottom of the frame. See video example. Why the timing is setup this way is bizarre, more sloppy engineering. Unfortunately with a lens swap this happens more often, as the large sensor readout window takes longer, the full frame readout could be up to 40ms (25fps is the video mode used). With the stock lens the only the middle ~500 scanlines are used (compared with my 1200), so it readout will complete up to 16ms earlier.

In other news: Busy week, progress on Type-B is slow, have not started on Type-A yet.

 

[ThePhage]

I might need help to find the trigger for the motor step function (and/or the frame grab function.) Even on the stock lens and firmware, you will occasionally see some instability at the bottom edge of the frame. This is a rolling shutter artifact, as the frame is being grabbed, at nearly the same time the film is starting to be moved. You can see this as unwanted motion on the bottom of the frame. See video example. Why the timing is setup this way is bizarre, more sloppy engineering. Unfortunately with a lens swap this happens more often, as the large sensor readout window takes longer, the full frame readout could be up to 40ms (25fps is the video mode used). With the stock lens the only the middle ~500 scanlines are used (compared with my 1200), so it readout will complete up to 16ms earlier.

In other news: Busy week, progress on Type-B is slow, have not started on Type-A yet.

That's interesting about the rolling shutter artifact on the bottom of the frame, when I've only experienced on the top of the frame (and only occasionally). Your explanation about why it's more pronounced (now after a lens swap) makes perfect sense, but I still can't figure out the reason for the variability of this issue in general. Why it happens sometimes but not others, is simply beyond me. And unfortunately I wouldn't have a clue about the motor step or frame grab function. That being said, if the community here is able to access and adjust that, I would imagine it could open up a whole host of new possibilities that could further increase the quality from this machine.

And thank you for persevering on the Type B and Type A unit mods!

 

[0dan0]

I too very occasionally see the error on the top of the frame (~2% of captures.) That is more baffling, as it might suggest there is no good timing between the motor control and the frame grab.

The socket feed is in motion a lot, there is very little time when nothing is happening.

a) 100ms - the pin is moving the film forward
b) 67ms - the pin is stationary at the end of the path
c) 67ms - pin is motion downward
d) 67ms - pin is stationary retracted
e) 132ms - the pin is moving to the start position
f) 67ms - the pin is stationary at the beginning of the path

I just an idea. I move the Reels camera to film its own socket mechanism.

I found a large degree of timing variance between captures, although it doesn't seem to drift within a capture. The unit tested has stock firmware. The frame grab is generally upon the start from pin reset (stage e). All these captures are after the pin was retracted from the film, but the timing very sloppy, approximate 100ms of error observed.

 

[0dan0]

More things sloppy

Where Reelz or Reels, both claim 8.08 Megapixels, yet they both using an Aptina AR0330 sensor, which is only 3.15 Megapixels. The video mode used (likely) on the Aptina sensor is limited to 2304x1296 (ever wondered where the 1296p came from?) So have the marketing is from the Aptina sensor , but it is not 8MP. Not that an 8MP sensor would perform better, is just so sloppy.