I skipped through the video and Jack was putting up some enjoyable images. I don't think that sensor will ever be something I want, but it can do some pretty good stuff in a short period of time.
I did a little more searching about the reason why the sensor has 9 micron pixels but still has only a 12-bit ADC. I didn't find an explicit answer but there is a hint at the answer:
At this link: https://www.framos.c...-global-shutter if you go about 1/4 of the way down you find a table with some specifications on the 1st, 2nd, and 3rd generation Pregius sensors.
To me the chart suggests (as does the prior verbiage) that the IMX432 may be doing a hardwired hardware binning (2x2) of some sort. I've also seen it claimed that the sensor is using a CCD architecture. So they may be using another sensor (maybe the IMX428?) with 4.5 micron pixels, binning the thing with effectively no other changes to the architecture. The IMX420 & IMX428 have the 12-Bit ADC so the IMX432 has the 12-bit ADC despite (assuming real hardware binning) having a well depth which may be four times that of the IMX420.
So you end up with an IMX432 well depth which is claimed to be (by Framos) at 100K instead of the 25K of the IMX420 and a gain of 86 dB to what I think may be 76 dB for the IMX420. But the ADC will not allow full exploitation of its binned 4.5 micron pixels.
So according to Framos the IMX432 doesn't really have 9 micron pixels, it has binned 4.5 micron pixels. I don't think it is dishonest to call it a 9 micron sensor because it can't function in any other mode, but it does sort of explain why the sensor has only a 12-bit ADC.
The base sensor appears to me to really be the IMX420. This sensor is apparently designed for high frame rate industrial use. Running at 7 MP with the system set for a 10-bit output it is supposed to be able to send 172 frames per second. . . We aren't going to handle that kind of data with the USB3 interface.
So they slowed down the frame rate for the IMX428 which will put out 51 frames per second which a USB 3 interface may be able to handle.
Then I'm guessing they decided that they could hardwire/hardware bin the output of the IMX428 and get a faster frame rate albeit with less resolution. There are likely some industrial applications which just don't need the 7MP resolution potential so you make the IMX432 which uses binning to reduce the number of pixels to just 1/4 that of the IMX420 and IMX428 and you get a much faster frame rather than the IMX428 over USB 3.
I'm betting that they decided not to mess with a 14-bit (or maybe better?) ADC for the IMX432 for two reasons:
1. They are trying for a high-speed frame rate over a USB 3 cable and 14-bits would make it a much slower frame rate sensor. So there isn't a very good niche for a 14-bit IMX432 in their intended market.
2. It'd cost them a lot of money and time in changing their systems from producing the IMX420 with a 12-bit ADC to making the IMX432 with a 14-bit ADC.
And it would not surprise me at all if the IMX420, IMX428, and IMX432 have exactly the same architecture and silicon. They just change a few settings in the firmware and it is considered a different sensor.
Anyway, I'm betting that is how this is working. It's disappointing because I think we'd all really, really like to see what a 14-bit IMX432 could do.
But it might be worth remembering that if the IMX432 can do this, the IMX428 may be able to do what most of us want for OAP (sub-exposures of a few seconds rather than dozens of subs per second) and do it at 7 MP. And since there is a chance the IMX432 is using the same kind of clunky/noisy binning used with the IMX071 sensor (if you mistakenly turn it on), the IMX428 may actually give you a better SNR than does the IMX432!
So it is possible that you could bin the IMX428 in software and get a better SNR than you get with the IMX432 and you still have the same dynamic range. I hope Mallincam will look at using the ToupTek IMX428. Since they've already got the IMX432 working with their modifications I'd think it would be pretty easy for them to do the same with the IMX428 camera and they'd likely have a camera which would interest a lot of us who aren't going to want the IMX432 camera. I'm thinking that a modified ToupTek IMX428 camera would be a great replacement for the Mallincam Universe in their line-up.
I also came up with an answer to my wondering just where the DS432 might fit for a future use which I would appreciate. The thing is actually able to do some interesting work and since it actually doesn't put out a whole lot of data when run at several second exposures, you could potentially use a larger sensor than something like the IMX224 with what is probably a pretty good SNR and you might be able to actually do your processing on a faster smartphone or tablet. So you could match a larger image circle with a rather long focal length and be using a light footprint processing system for an experience which might look a little (but only a little) like what I think they are doing with the Stellina. But someone will have to develop the software for that purpose.
Edited by OleCuss, 14 August 2019 - 08:49 PM.