Sorry for being late to the party!
The Kinograph is 35/16 only. Good. Best to stick with dual-format, attempting to do everything in a single machine would prove too difficult. The 35/16 Kinograph needs to support up to 2,000ft reels whereas a S8+R8+9.5mm Kinograph would only need to support up to 400ft, ideally with a short film path.
Optical module. My opinion is that you should buy the cameras wholesale pair them with a lens, mount them on a track and sell it as a part. I reckon only the 5K and 2K cameras should be supported, at least initially, pick the manufacturer you want (Flir? Emergent Vision Tech?) and stick with Pregius S for now. Each time you add support for a new camera - who is going to be the one who provides the support for it? The end-user? That’s why it’s better to define the optical module and control it IMO.
The track it gets mounted on for focusing should probably be a machined part.
Film Cleaning. Ideally the design should facilitate an optional in-line Kelmar Christie film cleaner. Isopar-G will dry on take-up.
The light is being re-engineered. Excellent. You ideally want someone who will manufacture the light as a part. Letting them work out the diffusion and cooling for it will mean you don’t have to worry about it. In answer to your question - in the Bayer scanners they all manufacture their own lights and they are Red + Green + Blue + White full spectrum, and flash discreetly on each capture. They have liquid cooling because they make them very bright. Getting someone to engineer it as a part will provide good value without breaking the bank on the cost of the device.
Colour inversion is done using their own in-house developed LUTs that are trade-secret and embedded in their closed-source software. Same with debayering.
Software. The machine vision cameras are not video cameras, you have to debayer yourself unless you want nearest-neighbour debayering! That should be a priority to get sorted out in the workflow software wise. There are free solutions (eg this) but you need a way to convert raw camera to Prores that does an acceptable debayering job that the average Kinograph operator can use.
Worry about that first and then colour inversion later.
There should be clear written goals for the host software capability, and it will take time to realise those goals.
Split-reels. You’ll need retaining platters as a part. Maybe not right away, but you will need them even if the machine is tilted back.
Film gates/skid plates. You want a warped-film clamp/gate as well, but that can come after the initial design.
The host computer. Once you get going with film capture software, you’ll find that the hardware requirements for the host computer may go up depending on how much you’re asking the computer to do. The old-hat scanners of the past like the Spirits used internal logic boards to do the image processing etc not the host computer. The preference with the modern scanners is to use the host computer instead. Power requirements go down, the engineering is simpler, and technical specifications for the host PC go up.
The Kinograph doesn’t have logic boards (other than the bare minimum to run the motors etc), so anything and everything that isn’t done in the machine itself has to be done either by the host computer or manually by the user. Film tensioning? Focusing the camera? Frame registration?
The 8mm machine needs to be cheap. People complain that the Pictor is too expensive! Film-Digital sell existing projector telecine kits dirt-cheap.The aim should not be to undercut Filmfabriek on an 8mm device.
It should not be the priority IMO as the Pictor already exists, it is affordable, and any 8mm Kinograph device would be far more limited.
Non-profit vs for-profit. While I support the idea, I’m lost on the details. The companies that make the cheapest devices (Blackmagic, Filmfabriek, Ventura Images etc) price their machines with support included for free. So what does product support look like, and how is that factored into the business plan? Filmfabriek is basically run by volunteers, but it’s not a registered non-profit organisation.
Modules. Keep it to a minimum. You have a modular design - but I would suggest developing just one light, perhaps two optical modules, and one film transport module (at least initially). The more options you give, the more complicated the device becomes, and the more expensive it will be to provide technical support to the user.
I hope I don’t sound too critical or anything, my concern is that your goal seems to be that the user builds the device to spec, and that any changes/modifications they make they need to work out on their own. That’s perfectly fine, but the Kinograph itself needs to be a singular well-defined product/device with clearly defined parameters. Without that, the host software (yet to be written) will never work.
