Have finished scanning 8mm and Super8 on Kodak Reels with the increased bitrate. While I was still slightly frustrated by the still-too-low bitrate, certainly a world of improvement over org bitrate.
Checking in with family to see if anyone else needs the modified-firmware-hardware before I try any of this hacking myself... still not eager to brick it due to any random incompetence on my part.
But I do want to share that I THINK i improved performance (assuming zooming-out completely is a good starting point for best-performance to avoid digital zoom) by blocking part of the sensor, so that it was capturing static white or black instead of motion. So that was the motion from adjacent frame, redundant info. I assumed that by doing so more bitrate could be dedicated to pertinent film-frame.
In theory I think bright-white would have been the best, but I couldn't insert anything uniformly bright enough between the sensor and the film. (Tried reflecting light with aluminum foil, tried diffusing the back-light with scotch tape.) Ultimately I think black worked best, same aluminum foil but without me shining a light on it.
And the light ended up being a $10 thing off Amazon, powered by USB. Some USB lights were annoyingly flickery, anything customizable in color is BAD. If you try using a USB LED light for something like this just get one that only shines bright-white.... the coloured ones might use rapid flickering in some way to achieve their colours.
If anyone has a program they can recommend to analyze MP4 frames where I did this, I'd really be curious if the black patch actually saved data, as compared to film capture. Sensor noise happens everywhere, but it looks like less if very-white or very-black. I expect a tool can tell us how much data is spend on every 16x16 grid of MP4, but my (admittedly casual) attempt to analyze was unsuccessful. Happy to send an MP4 if anyone's willing to try figure out is this is a smart move.
And also... Mac84 recommends zooming out 100%. Any second opinions on that? I put the footage thru Davinci Resolve and applied upscaling and all manner of trying to make it look good on YouTube. But if I didn't have the ability to re-frame the shot, then zooming out would have left it a harsh viewing experience... adjacent frames showing or pertinent frame not filling view. I agree avoiding cheap-hardware-digital-zoom make sense, but I can't say I ran tests to try confirm it.
And that's sort of the problem... my own try-cover-the-sensor might have been a fools errand. I can't see any difference, I'm just going by ideas that make sense to me but I can't actually see if it helps or does not. Obviously Mac82 increased bitrate helps. I can see that myself. But everything else is me squinting and second-guessing.
Playlist, for anyone's who's curious...
Share your videos with friends, family and the world
studio.youtube.com
...that's zoomed-out-completely, then using Davinci Resolve to upscale 2x with maximum noise reduction. (It does use >1 frame to calculate what the noise-reduced picture should be.) Really a lot of data and processing power thrown at Kodak Reels (Mac84 firmware) still tiny MP4... I'd get a 400MB MP4 capture and render it out as a 4GB 4K MP4... mostly so YouTube will treat it with respect. If anyone thinks their Kodak Reels capture looks better (on YouTube!) please direct me to it so I can see what is possible. And let me know what-ya-did.
