David Farkas;190 wrote: Okay, here is a big one I left off that I’ve written about before…
Dr. Zimmer, head of digital R&D, explained that noise could be reduced by either dropping the sensor readout from 24Mhz to 12Mhz or by using a dual readout vs. a quad readout from the sensor.
That’s surprising: why would dual vs. quad readout make so much difference? My first thought would have been to add more readout channels and clock them slower, but he seems to imply that loses in the real world.
It might be worth asking if the hardware can do single-channel readout. If anyone has ever detected banding a single-channel readout would eliminate it. But maybe banding is already a solved problem and no longer an issue.
Either of these options would cut the continous frame rate in half and double the time from capture to review, but also would deliver a one stop improvement in noise performance.
It may not be necessary to do readout in strict row order. Perhaps they could readout every tenth row first and display the preview quickly while doing the rest of the readout to write to the card(s).
Compared to pixel binning options like Phase’s Sensor+, this solution would offer full resolution not 1/4 res.
binning may increase dynamic range considerably and improve chroma resolution depending on how the sensor works.