Upd: Sensor noise identical M8 and M9 - difference only in firmware

[offshore]

Advertisement (gone after registration)

I really just want to set the record straight, and then take a break from the forum, if I have the strength of will. It does seem that everything can be misinterpreted and often a negative spin put on even the most enthusiastic post.

I was simply excited by the post over on the other subsection, mostly because I'm eager for the M8 to have uncompressed DNGs - something which I've argued in favor of for a long time, and been told made no difference to image quality (often by the same people who are now saying it makes an enormous improvement to M9 files). My interest in improved noise characteristics is secondary.

 

1. I bought my M8 long after it was clear that the 'perpetual upgrade' program was dead - I have no beef about this whatsoever. I do think it is good customer relations not to suddenly abandon your loyal base, however - and therefore imho it would be to their own advantage if Leica supported the M8 in future.

 

2. I didn't consider this thread to be "shouting" about anything.

 

3. This was definitely not in any way intended to be any sort of attack or denigration of the M9. As I said, I was just trying to make the thread title snappy and short, and thereby the obvious misunderstandings.

 

Next time the thread will be called:

 

"Sensor noise in M8 (at the pixel level) rather similar (in the dark tones at least - and possibly not even there) to the M9 (which has other hardware advantages) but don't forget the bigger sensor of the M9 (which allows for much better noise characteristics in the final print - if the prints are the same size. If the M8 print is smaller and you stand closer, then it might look similar)."

Mods - could you change the thread title as above please? This will spare me a lot of grief and avoid misunderstandings.

Plasticman I agree with you regarding a break from the forum as I've taken one for a little over a week now and have checked in only to find the same things being chewed over. As far as the perpetual upgrade program for the M8 if Leica does abandon it that will definitely influence my decision to buy Leica cameras in the future. Others on this forum have called this perspective as naive to think that Leica would keep to that statement in today's electronic world. I disagree because I think that one of the reasons we purchase Leica products is that we have come to rely on Leica as a reliable company that produces reliable products. If they demonstrate that they are abandoning reliability for main stream marketing designed only for today then I would have serious doubts about them supporting any camera in the future. I'm betting I'm wrong and they will have further upgrades for the M8 but right now I'm taking a wait and see stance before I plunk $7K down for another Leica camera.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[ampguy]

What about the non-uniformity of the micro lenses across the M9's sensor area?

 

Not wanting to pour cold water on this one, but Hans's half stop figure as quoted on the the other thread I think would only be visible in deep blacks, not mid-tones.

 

I think that what the evidence is showing is that if you compare the M8 and M9:

 

1. On a pixel by pixel basis, the M9 has only a slight advantage over the M8. So if you were to compare an M8 image to an uncompressed M9 image cropped to the same number of pixels as the M8, 10MP image vs 10MP image, the images would be very similar, except perhaps in the shadows where the M9 has the advantage of not being compressed, and different black level processing.

 

2. However, if you compared a shot of the same scene, 18MP image vs. 10MP image both printed on the same size paper, the M9 would have the near 1 stop advantage that most people see, just because visually smaller pixels with the same amount of individual noise give a less noisy overall image.

 

Hope I'm being clear in comparing the two situations.

 

Sandy

[sandymc]

What about the non-uniformity of the micro lenses across the M9's sensor area?

 

Assuming you're using coding or using CornerFix, it would be (mostly) corrected for. If not, then certainly there would be difference in vignetting as well.

 

Sandy

[mhoersch]

Plasticman I agree with you regarding a break from the forum as I've taken one for a little over a week now and have checked in only to find the same things being chewed over. As far as the perpetual upgrade program for the M8 if Leica does abandon it that will definitely influence my decision to buy Leica cameras in the future. Others on this forum have called this perspective as naive to think that Leica would keep to that statement in today's electronic world. I disagree because I think that one of the reasons we purchase Leica products is that we have come to rely on Leica as a reliable company that produces reliable products. If they demonstrate that they are abandoning reliability for main stream marketing designed only for today then I would have serious doubts about them supporting any camera in the future. I'm betting I'm wrong and they will have further upgrades for the M8 but right now I'm taking a wait and see stance before I plunk $7K down for another Leica camera.

 

Very well put, in my opinion! Same here.

[scott kirkpatrick]

I keep seeing references to the idea that the M8 may not have enough memory to apply corrections to uncoded lenses... However, that can simply not be the case. At the moment every M8/M8.2 has the lens database in its firmware -- the six bit coding simply tells it which lens to choose. All that has happened in the M9 is that Leica has made this lens database accessible via a menu option on the back and not just via the six bit lens coding. (The lens coding contains no information of itself, merely a reference to be matched to a database, i.e. a way for the lens to tell the camera which lens it is.)

 

 

I'm very doubtful that table for manual identification of uncoded Leica lenses will make it into the M8. The table links to two different functions -- first it puts the desired lens focal length, maximum aperture and product code into the EXIF data of every shot. That part is easy to do, and we could probably have it in the M8. Second, it triggers vignetting corrections for every DNG file that is shot. These come in two flavors for the M8 (with and without IR filters), and only one flavor for the M9. The M8's corrections are contained in tables. I can read the M8 firmware in a text editor, and see the tables. They are HUGE. The M9 firmware is encoded, so at the moment no one has figured out if such tables are used there, or some completely different.method. We know from a comment in the Stefan Daniel online video at LL that the M9s vignetting corrections are done with a single curve for each lens, regardless of aperture. That's how the original Jenoptik vignetting corrections worked. The curves used when an IR filter is installed seemed to depend on estimated aperture. Anyway redoing that part of the firmware for each lens in the table seems a lot to ask. EXIF-only might be reasonable.

 

scott

[scott kirkpatrick]

What about the non-uniformity of the micro lenses across the M9's sensor area?

 

I think that trick is less magical than Kodak likes to make it seem. I read somewhere that the optimal shift of each microlens is known, something proportional to tan( theta) where theta is the angle that a light ray makes as it hits the sensor. I would guess that the middle 10 MPx of the M9's sensor has the same lens offsets as the M8's sensor, and the offsets get larger in the outer part of the M9, the new pixels. There's language in the press releases about using new glasses for the microlenses, so some coefficients may have been fine-tuned, but in this aspect, as with most others, the M9 is simply a bigger and better M8. And it is hard to see much wrong with that.

 

scott