The M8 sensor; a quantum leap?

[Goldie]

Advertisement (gone after registration)

In the light of the following observations, I wonder if the M8's sensor is going to be a significant step towards a high quality 35mm format sensor.

Firstly the M8's crop factor is 1.33, whilst the DMR's crop factor is 1.37. This in itself is but a marginal step, however allied with the following observation the sensor would seem to have moved forward significantly.

The M cameras have a flange to focal plane distance of 27.8mm, the R cameras have a much bigger distance of 47mm.

It is my understanding that the DMR's sensor size was determined by the ability of the sensor to handle angled light bundles at its outer zones; a problem resolved by angled micro-lenses in these sensor areas. With the M8's shorter back focal distance this problem becomes much worse, since the light bundles, in order to fill the same format size, must strike the image plane at a lower angle.

Since Leica, I am sure, will not allow the M8 to have a poorer performance than the DMR, it would appear that a significant advance has been made in this area of performance. Perhaps this might explain the mystery about the sensor in the M8, alluded to in other threads?

[masjah]

John

 

My understanding is that the constraint on the size of the DMR sensor was the requirement to fit within an existing film gate, hence the practical maximum of 1.37x. Angled rays were an issue to a degree, addressed by off centre microlenses. However, for historical reasons related to the original Leicaflex, the back focus distance of the R system is relatively large, meaning that R system lenses are "inherently more telecentric" than some.

 

The M8 sensor of course has no need to fit into a film gate. However as you point out, the angled-ray problem is very significant, because of the small back focal distance. Being able to get as far as a 1.33x factor is, I agree (FWIW) a very substantial achievment indeed, and doubtless the off centre microlens technology plays a crucial role. "Coded" lenses enabling in-camera-firmware correction of vignetting will doubtless also contribute - it will be interesting to see how much better coded lenses perform than uncoded ones.

[adan]

IMHO

 

Quantum leap - no.

 

Evolutionary improvement - yes. I suspect Leica would not have needed the lens coding, as well as the extra software to make use of it (in-camera or in-computer) if the sensor were really a quantum leap improvement.

 

A quantum leap will come when they can get M lenses to perform well on a 24 x 36mm sensor.

[Goldie]

Well I suppose it depends on how big a quantum leap is! But point taken Adan.

The lense coding raises some interesting questions. As soon as I read about it, I thought vignetting control, and certainly the 6-bit code is more than adequate to carry focal length and aperture data. It seems to me it can work in two ways.

Firstly to remove any sensor vignetting, which on reflection is probably lens independent.

Secondly to remove lens vignetting, this could be defeatable.

Since vignetting is aperture dependent, the system will have to have some method of setting the camera's sensor to the maximum aperture of the lense; perhaps opening up the lense and pressing a Set button.

The system would have to be based on a lense specific algorithm, creating a stepless, concentric increase in sensitivity towards the sensor's edges, so I hope it does not create noise problems at the image edges.

Perhaps an improvement in noise is coming our way; after all the M is historically an available light camera.

[Bob Ross]

Since vignetting is aperture dependent, the system will have to have some method of setting the camera's sensor to the maximum aperture of the lense; perhaps opening up the lense and pressing a Set button.

The system would have to be based on a lense specific algorithm, creating a stepless, concentric increase in sensitivity towards the sensor's edges, so I hope it does not create noise problems at the image edges.

Perhaps an improvement in noise is coming our way; after all the M is historically an available light camera.

Not all vignetting is aperture dependent. Mechanical vignetting from lens hoods, filter, etc. does reduce when the lens is closed down. Cosin4 light fall off vignetting is a minor component and varies by focal length, showing most in ultra wides, but doesn't vary much by aperture. Sensor fall off is dependent on the angle the light hits the sensor/micro lens and that is affected by the exit puple and diameter of the rear element and, of course the focal length, lens registration distance and micro lens off-set design. A lot of this came from long discussions about Oly's telecentric claims on various forums.

If you DIY in PP, you could take pictures through an Expo-Disk for each lens and create a mask. These taylored masks could the be used to correct exposure, color balances and noise reduction. All you have to do is remember which lens you used;) If Leica has all the coded lenses profiled like this and the dots even remember which lens you used, the cost for upgrading might seem more reasonable:) The other thing to wonder about is will Leica's RAW converter have this built in, when you or the dot coding provides the lens data.

Bob

[Goldie]

Bob

I think this lense coding is going to be really used by Leica to its fullest extent. That is to make lense performance as sensor neutral as possible and to retain that famous uniform "feel" across the focal length range.