M8 sensor source?

[albertwang]

Advertisement (gone after registration)

It says that the M8 sensor is a Kodak version on the spec sheet. But I thought that Jenoptik was handling it this time around??

 

Unless there was a secret collab that I didn't know about. (Note that Kodak handles Olympus equipment so the Oly-Leica share the same matrix).

[Bob Ross]

It says that the M8 sensor is a Kodak version on the spec sheet. But I thought that Jenoptik was handling it this time around??

 

Unless there was a secret collab that I didn't know about. (Note that Kodak handles Olympus equipment so the Oly-Leica share the same matrix).

Alfie,

Kodak does the sensor in the DMR. The KAF-FFT sensor architecture is also found in MF backs and has some qualities that makes it a useful choice for the offset microlens approach. Jenoptik does the processing software. Different matrix;)

So, who does the sensor in the E-400?

Bob

[marknorton]

Is the Jenoptik connection confirmed? I thought I saw a Danish company (not Imacon) were doing it. Maybe Jenoptik makes the electronics?

[albertwang]

So, who does the sensor in the E-400?

Bob

 

I would be floored if it were the same sensor as the DMR/M8... If this be the case, the Oly needs to come up with good processing software.

[adan]

Kodak makes the DMR sensor and the M8 sensor.

 

Imacon makes the 'digital package' for the DMR - combining the Kodak sensor on a circuit board (or boards) with all kinds of other chips from other manufacturers - some off-the-shelf, some likely purposed-designed for the DMR. Everything from the analog/digital converter to the LCD screen.

 

Someone (not yet revealed) makes the 'digital package' for the M8, using the Kodak sensor, but also a whole lot of out-sourced stuff from other makers.

 

This is how stuff gets made these days. Dell laptop, assembled in Indonesia with a CPU from Intel, software from Microsoft, I/O electronics from Taiwan, memory boards from Malaysia (built with memory chips from Thailand), fan from China, keyboard from any of the above, LCD from Korea, etc. etc.

 

Since Leica chose Capture One LE as the included RAW software - perhaps Phase One had a hand in the in-camera processing as well.

 

Kodak makes a lot of very different sensors for a variety of different purposes. Big, small, monochrome, RGB. Two cameras using Kodak sensors could be as different as Tri-X and Kodachrome 64.

[Bob Ross]

I would be floored if it were the same sensor as the DMR/M8... If this be the case, the Oly needs to come up with good processing software.

The 4/3rds sensor is one half the 1.33 sensor size, so it wouldn't be the same one. Kodak made the KAF-8300 for the E-500 and maybe they did something to that one that caused a pixel population explosion:eek:

If you want to get sensative about sensors start at KODAK: Topic Menu) They haven't got any of the new ones listed yet.

Bob

[billh]

Advertisement (gone after registration)

The 4/3rds sensor is one half the 1.33 sensor size, so it wouldn't be the same one. Kodak made the KAF-8300 for the E-500 and maybe they did something to that one that caused a pixel population explosion:eek:

If you want to get sensative about sensors start at KODAK: Topic Menu) They haven't got any of the new ones listed yet.

Bob

 

 

http://www.dpreview.com/news/0609/06091501kodakccdleicam8.asp

[Bob Ross]

Thanks Bill, I just came from reading that. There are some interesting changes from the KAF-10010 in the DMR. I allerted Phil of a likely typo for the red channel in quantum Efficiency. It should probably read 37% instead of 17%, to be in line with other KAF sensors.

Bob

[billh]

Thanks Bill, I just came from reading that. There are some interesting changes from the KAF-10010 in the DMR. I allerted Phil of a likely typo for the red channel in quantum Efficiency. It should probably read 37% instead of 17%, to be in line with other KAF sensors.

Bob

 

Hi Bob,

 

Since you understand these things, would you mind sharing some of that knowledge with us - but distilled into simple language we can all understand? For example, would what you read give you a sense of the resolution and noise we can expect?

 

Thanks!

[Bob Ross]

Hi Bob,

 

Since you understand these things, would you mind sharing some of that knowledge with us - but distilled into simple language we can all understand? For example, would what you read give you a sense of the resolution and noise we can expect?

 

Thanks!

Hi Bill,

The noise picture will have to wait until we see how the images are processed with the final firmware. The resoulution we can figure from the pixel size of 6.8µm, which yields 147 pixel per mm. A line pair is two pixels wide (white line & black line) so 147 p/mm is 74 lp/mm. That would be the maximum resolution from which everything gets subtracted like any filters in front of the sensor and, of course, the lens (which probably out performs 74lpm). The actual test resolution on the review tests will tell what gets lost in the processing. 74lpm =1880 lines pr inch, so the test rez charts will show less that that.

There are are other ways to look at digital resolution, such as print size. Some of the talk about resolution doesn't take into consideration that maximum resolution of detail only shows where there is the finest detail in a picture to show it. The picture quality has more to do with the rendering of tonalities and the Kodak (KAF-FFT) sensors do that very well.

Bob

[billh]

Hi Bill,

The noise picture will have to wait until we see how the images are processed with the final firmware. The resoulution we can figure from the pixel size of 6.8µm, which yields 147 pixel per mm. A line pair is two pixels wide (white line & black line) so 147 p/mm is 74 lp/mm. That would be the maximum resolution from which everything gets subtracted like any filters in front of the sensor and, of course, the lens (which probably out performs 74lpm). The actual test resolution on the review tests will tell what gets lost in the processing. 74lpm =1880 lines pr inch, so the test rez charts will show less that that.

There are are other ways to look at digital resolution, such as print size. Some of the talk about resolution doesn't take into consideration that maximum resolution of detail only shows where there is the finest detail in a picture to show it. The picture quality has more to do with the rendering of tonalities and the Kodak (KAF-FFT) sensors do that very well.

Bob

 

Thank you. Now, can you tell me what the resolution limit is on my 1Ds2?

 

I can easily see the difference in an 8 x 11 print from it, the 8 MP 1D2, and the 6MP Epson RD-1. When I bought the 1D2 (before the release of the 1Ds2), I often missed the extra resolution I could see from my 11 MP 1Ds. I also remember someone here recently saying this difference was also obvious between the 1Ds2 and the 39 MP back he uses (in small prints). Of course a decent image form any of these cameras usually looks just fine, so it is really when you compare them together that the difference becomes obvious. I am assuming the M8 will be at least as good as my old 1Ds (now worn out and laid to rest). There also appears to be a difference between the CCD and CMOS - can you comment on this too?

 

For 1Ds2 specs:

http://www.dpreview.com/reviews/canoneos1dsmkii/

 

Thanks again!

 

Bill

[Bob Ross]

Hi Bill,

I used to have some notes on the 1DsMII around here, I'll have to look for them. There are some comparisons between the 1DsMkII and the DMR on line out of Spain (I don't have a link). Leica doesn't use an AA filter in the DMR or the M8 and while this can encourage moire, it does remove the blurring effect of the filter, so detail is rendered finer.

I think the increased megapixels and pixel size in the 1DsMkII's sensor do give it an advantage, which then may be cut back by the AA filter and the CMOS + processing. We'll have to see what the differences are on monitor and in prints. Printers and the printing process does tend to level the playing field in normal sized prints and exaggerate it in jumbo prints.

Bob

[rosuna]

This is the link:

 

Rubén Osuna Guerrero

 

R.

[rosuna]

A 24x36 frame needs a x8 enlargement for a 8x12 (aprox. A4 print).

Whether the sensor and lens resolve 40lp/mm we will get 40/8 = 5lp/mm of real detail in the print. Just the amount of detail the human eye can resolve at optimum distance.

For a A3 print (12x16, aprox) and 5 lp/mm of real detail the sensor (film) need to resolve 80 lp/mm. There is no full frame digital camera with this resolution power (I think a 22MP full frame camera would be necessary). Only medium format cameras, mostly due to the smaller enlargement factor, can reach the necessary resolution for those prints.

 

These requirements are increased for a cropped sensor. The M8 sensor has a crop factor of 1,33, and therefore 40x1,33 = 53 lp/mm and 80x1,33 = 106 lp/mm should be resolved in order to have 5lp/mm of real detail in those prints.

 

The 74 lp/mm of maximun theoretical resolution, and less than that number of real maximun resolution, means that the M8 is capable of gorgeous A4 prints, but other cameras actually in the market will provide visibly better prints at A3 sizes or bigger (I am thinking on the Canon 1Ds Mark II and Canon 5D, due to the bigger picture size -number of pixels- and smaller enlargement factor).

[rosuna]

Here we have all the information about the sensor:

 

Kodak Sensor in Leica M8: Digital Photography Review

 

R.

[waterlenz]

Hi Bill,

I used to have some notes on the 1DsMII around here, I'll have to look for them. There are some comparisons between the 1DsMkII and the DMR on line out of Spain (I don't have a link). Leica doesn't use an AA filter in the DMR or the M8 and while this can encourage moire, it does remove the blurring effect of the filter, so detail is rendered finer.

I think the increased megapixels and pixel size in the 1DsMkII's sensor do give it an advantage, which then may be cut back by the AA filter and the CMOS + processing. We'll have to see what the differences are on monitor and in prints. Printers and the printing process does tend to level the playing field in normal sized prints and exaggerate it in jumbo prints.

Bob

 

I am wondering if the lack of an AA filter will enable significant upres'ing?? Maybe 2x linearly, which would lead to 40+MP!!

Tom

[Bob Ross]

Hi Ruben,

Thanks for joining in and giving the links and a very nice explanation. The 1DsMkII sensor gives 69 lp/mm, if I did the math right. So, for apples and oranges we have 47 lp/mm with no AA filter, CCD, 1.33 crop compared to 69 lp/mm, CMOS, 1.00 crop.

What are your comments on Tom's question about no AA filter and up rezzing?

One thing that I noticed in the sensor specs is that the M8's sensor have a saturation signal of 60k electrons and the DMR's sensor is 40k electrons. This extra signal capacity may result in lower noise (higher signal:noise ratio) and even better tonality for the M8.

(BTW, I liked your write up on the Mamiya ZD over on LL)

Bob

[rosuna]

Thanks Bob,

 

the saturation value points to a wider dynamic range.

 

Whether we avoid the AA filter we gat two effects: higher "per pixel" resolution and higher contrast (the own MTF of the sensor changes).

 

The problem of the actual digital images is its inefficiency. We have more pixels (theoretical resolution) than real detail. The number of pixels determine the size and resolution (density) in the print, but it does not determine the real detail in the print. AA filters, Bayer patterns (and interpolation) and the combined power of sensor and lenses (lower than the separated power of the sensor or the lens), explains why we have "detail" and "smoke" into the pixel matrix. More detail in the same picture size makes easy to interpolate for obtaining bigger images (we add just more "smoke" by interpolating).

 

I think Erwin Puts' comments are fair, but I expect no much differences in prints between images from the M8 and the Canon 5D (real detail on paper).

[rosuna]

Thanks Bob,

 

the saturation value points to a wider dynamic range.

 

Whether we avoid the AA filter we gat two effects: higher "per pixel" resolution and higher contrast (the own MTF of the sensor changes).

 

The problem of the actual digital images is its inefficiency. We have more pixels (theoretical resolution) than real detail. The number of pixels determine the size and resolution (density) in the print, but it does not determine the real detail in the print. AA filters, Bayer patterns (and interpolation) and the combined power of sensor and lenses (lower than the separated power of the sensor or the lens), explains why we have "detail" and "smoke" into the pixel matrix. More detail in the same picture size makes easy to interpolate for obtaining bigger images (we add just more "smoke" by interpolating).

 

I think Erwin Puts' comments are fair, but I expect no much differences in prints between images from the M8 and the Canon 5D (real detail on paper).

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[john_f]

According to the specs for the Kodak KAF-10500 sensor that were kindly provided by dpreview (see http://www.dpreview.com/news/0609/06091501kodakccdleicam8.asp), the saturation signal is given as 60,000 e- and the "read noise" is given as 15 e-

 

Therefore, the linear dynamic range of this NEW Kodak sensor (not at all the same sensor as in the DMR) can be calculated as: 60,000 / 15 = 4,000

And the corresponding number of bits as: log2 (4000) = 11.9658 or almost 12-bit

 

This indicates that the sensor itself is capable of a maximum 12-bit resolution, which is even better performance than the DMR. However, please note that this has nothing to do with the AD/C (analog to digital converter) used to create the final image (which Leica claims as 16-bit). It is a very common practice for camera manufacturers to "stretch" the performance of the sensor by a few bits ... (to 12, 14, or even 16-bit). And this is where the real magic takes place: in the AD/C and "imaging engine" (e.g. DiGIC, VENUS, etc...). Luckily for Leica, this time it has teamed up with Phase One (NOT Imacon or Jenoptik - our predictions were wrong...) to design the imaging engine of the M8, along with dedicated post processing software uniquely tailored to the Leica M8.

 

And if you want to calculate the linear dynamic range in dB of the KAF-10500 sensor, the calculation is simply: 20 x log10 (2 ^ 11.9658) = 20 x log10 (4000) = 72.04 dB (which is slightly higher than the 71.5 dB quoted by Kodak).

 

All in all, what this means is that Kodak, PhaseOne and Leica have worked together to develop a sensor+firmware+software solution that will yield a level of image quality quite possibly never before seen in a compact digital camera. Once again, Leica is proving that quality (contrast, brilliance, colors) is more important than quantity (resolution). The M8 will be quite a camera.

 

Cheers,

John F.