Read Sean's ISO results and quite disappointed.

[sean_reid]

Advertisement (gone after registration)

If I were to make one comment it is that unless you know for a fact that Leica is doing in camera smoothing on ISO 1250 and 2500, why write that you think that they are?

 

One can see it in the samples. You saw them, right? A little observation and a bit of detective work on the RAW file timings and I'm pretty confident about that conclusion. But the smoothing isn't very strong.

 

Cheers,

 

Sean

[robertwright]

One can see it in the samples. You saw them, right? A little observation and a bit of detective work on the RAW file timings and I'm pretty confident about that conclusion. But the smoothing isn't very strong.

 

Cheers,

 

Sean

 

But is this not how internet "legends" get started?

 

I know you think you know and it makes sense but, there might be more to it than that.

 

And just what is "smoothing" anyway. We are talking digital signal processing. All of this is a sort of quasi-scientific tech talk. Yes the files I saw looked different, but as to why? Maybe the timing is different because of how the signal is amplified not how it is altered. I don't even know what I am talking about.

 

Y'all have great relationships with the sources, it seems worth a comment on the record and you'd be done with it.

 

You see I think you can only report what you see, you can offer "opinion" but that might not in the long run be as worthwhile. For example, the whole IR flap, IF-MR has simply reported he had seen an IR problem with a preproduction camera, which was a fact, he is not required to understand why he is seeing it, to do his readers a service he simply has to report it. Later, when the camera comes out either a: it is fixed and was an artifact of being half-baked, or b: it's not fixed and indeed, there is a problem. He then neatly avoids either a: starting a rumour, or b: being called a shill.

 

You can only report what you see and have proof of or someone on the record. The rest is opinion, which is important, but can only be built upon the rock of solid reporting. That way, your opinion is more respected.

 

anyway, just my opinion.

[Fotomiguel]

When shooting, one stop of better iso make a big difference! If you add to this one stop more of dynamic range in high lights, the M9 is really a great improvement. The small differences in a test become big differences when shooting. I'm longing to try the "soft release" as well! I think that this new feature could be very important when shooting in low light. There are many others small improvements that make the M9 more confortable and will make a range finder photographer really happy.

[sean_reid]

But is this not how internet "legends" get started?

 

I know you think you know and it makes sense but, there might be more to it than that.

 

And just what is "smoothing" anyway. We are talking digital signal processing. All of this is a sort of quasi-scientific tech talk. Yes the files I saw looked different, but as to why? Maybe the timing is different because of how the signal is amplified not how it is altered. I don't even know what I am talking about.

 

Y'all have great relationships with the sources, it seems worth a comment on the record and you'd be done with it.

 

You see I think you can only report what you see, you can offer "opinion" but that might not in the long run be as worthwhile. For example, the whole IR flap, IF-MR has simply reported he had seen an IR problem with a preproduction camera, which was a fact, he is not required to understand why he is seeing it, to do his readers a service he simply has to report it. Later, when the camera comes out either a: it is fixed and was an artifact of being half-baked, or b: it's not fixed and indeed, there is a problem. He then neatly avoids either a: starting a rumour, or b: being called a shill.

 

You can only report what you see and have proof of or someone on the record. The rest is opinion, which is important, but can only be built upon the rock of solid reporting. That way, your opinion is more respected.

 

anyway, just my opinion.

 

So basically you want a confirmation from Leica that the camera is applying smoothing to the high ISO files? Ok, we'll see how that goes but understand that they may not care to phrase things in quite that way. Meanwhile, the actual pictures examples are already there to seem and smoothing by any other name....will look like we see in those pics.

[sean_reid]

But is this not how internet "legends" get started?

 

No, those don't usually have supporting pics.

 

<G>

 

Cheers,

 

Sean

[robertwright]

So basically you want a confirmation from Leica that the camera is applying smoothing to the high ISO files? Ok, we'll see how that goes but understand that they may not care to phrase things in quite that way. Meanwhile, the actual pictures examples are already there to seem and smoothing by any other name....will look like we see in those pics.

 

I'm not trying take away from the work you are doing or the value you are adding to the community.

 

it just seems unknown is unknown.

[matts]

Advertisement (gone after registration)

Excuse me if this comment sounds a little amateurish, but the way I see it:

 

After shooting a 5D, it's remarkably nice to be able to go ISO1600 with a clean image, and while it's kind of inexcusable that an M9 (a camera that is easily 2-3x the price) can't handle noise as well, does it really matter?

 

I mean, did ASA3200 color film ever give super clear images anyways? People just worked around those limitations back then.

 

Again, while it can be said that such a camera should be modern enough to feature a low noise sensor, I don't think it's that big of a deal.

[robertwright]

No, those don't usually have supporting pics.

 

<G>

 

Cheers,

 

Sean

 

 

just call me mr. dragnet. the facts please mam:D

 

mr. dragnet is off duty now.

[Samir Jahjah]

Sean,

 

You end your ISO tests by saying, and I quote, 'In a section coming up soon, I will look at ISO noise in the corners (an important topic for the M9)'.

 

What is it to be expected here? This comment along with another comment you made on the CV 15mm seem to imply they are some performance issues in the corners of the M9. Could you give us a bit more information?

 

Thanks

 

Samir

[Jamie Roberts]

Excuse me if this comment sounds a little amateurish, but the way I see it:

 

After shooting a 5D, it's remarkably nice to be able to go ISO1600 with a clean image, and while it's kind of inexcusable that an M9 (a camera that is easily 2-3x the price) can't handle noise as well, does it really matter?

 

I mean, did ASA3200 color film ever give super clear images anyways? People just worked around those limitations back then.

 

Again, while it can be said that such a camera should be modern enough to feature a low noise sensor, I don't think it's that big of a deal.

 

You're right, but...

 

If history is anything to go by, before long the M9 will have a perfectly printable ISO 1600 (if it doesn't now). Wait and see: these are early days and DSPs and firmware are yet to be optimized, I believe.

 

Now, I don't know that, of course. But it was true for the DMR, true for the M8 and true for almost every digicam I've bought right out of the gate: it takes a little bit of time and extra testing (that is customer use) to squeeze all the IQ out of a system.

[giordano]

I'm not trying take away from the work you are doing or the value you are adding to the community.

 

Glad to hear it. Your words gave the impression that you were.

 

Re the science of sensors and noise reduction: some is trade secrets, but internet searches will find you a lot.

[sandymc]

You need to very careful about assuming that any blotchiness or whatever that you are seeing off of a compressed DNG is smoothing rather than compression artifacts. I think there's a problem with the M9 compression, and I would think its worse at high ISO. See the discussion here: http://www.l-camera-forum.com/leica-forum/leica-m9-forum/98212-lightoom-mangling-m9-images-there-compression.html

 

For the purposes of discussing high ISO noise (or any IQ issue for that matter) I think you really need to be looking at uncompressed DNGs, and its not clear to me thats what's being discussed here.

 

Sandy

[yanidel]

For those like me whose favorite FOV is 35mm, the M9 represents potentially about a 2-3 stops improvement over the M8. I use the 24mm Elmarit at F2.8 and will know use the 35mm Lux Asph 1.4 so two stops are already gained here (one for those that use the 28mm Cron). Add the one stop in prints gain + maybe a little bit more from the soft release option and you get IMO a big boost in low light capability at 35mm.

Obvioulsy one could opt for 24 Lux + M8.2 instead but as the Lux is basically the same price as the M9, it makes IMO more sense to go for the M9 and all the great improvements it brings beside higher ISO. Also need to take into account the significant difference insize and weight of the 24 Lux vs 35 Lux or even 35 Nokton 1.4.

For those whose favorite focal is 50mm and above, the practical gain of stops might be less as fast lenses at decent pricing were available.

[chris_tribble]

Sean - your experience tallies with mine - 1 stop difference - and for me this is actually a BIG deal. However, I'd be interested to know if you have any thoughts yet on the impact of using compressed DNG. Jamie and Sandymc have come up with what feel like important reflections on this. http://www.l-camera-forum.com/leica-forum/leica-m9-forum/98212-lightoom-mangling-m9-images-there-compression.html#post1033764

 

Grateful for your work thus far () and looking forward to any further pearls of wisdom!

[nugat]

Sean Reid delivers what he is universally revered for: facts, facts and only facts.

To me the conclusion is that M9=M8 in all aspects of picture quality, but M9 can print twice as big as far as high ISO noise levels.

 

 

-M8 high ISO performance = M9's (pixel for pixel)

-"cleaner" M9 look at 2500 ISO comes from processing that can well be done in post*

-S.R. throws a lifeline to Leica's "1 stop gain" claim by presenting the tautology: "Same sensor but doubled in size will deliver twice size prints while holding the noise level" (or "M8 prints on A3 will show twice the noise of M9 A3 prints ).

 

Although the actual system (camera+lens) resolution was not tested yet, one can assume with high level of probability it will be similar (+/- 10%) on both cameras. (Puts et al).

 

What this leaves me with is that M9 offers 'true" FOV for my 15, 21/2.8, 28/1.9, 35/1.4mm lenses.

M9 is 5,500 euro. Taking all of the above into account the better startegy for me is to buy a second-hand summilux 21/1.4 now and a used M9 in 2 years time when my M8 will be up for a change. I will spend the same 5,500 euro in total. For the interim 2 years 21/1.4 will nicely do the job of "true" 28mm lens, the widest one used by me on a daily basis with full frame.

In two years, time M9 will join my Leica stable.

 

* For the current generations of sensors it is better to stay at 1600ISO and underexpose raws 1-2 stops than go to 3200 or 6400.

[marcusperkins]

Len at R.G. Lewis let me have a quick play with an M9 the other day. Shot full DNG.

 

I thought the 160 ISO files viewed at 100% looked smoother than I was accustomed to from my old M8 at 100%. In short, they didn't have the bite I was expecting from a CCD without an AA filter. Not a complaint, I liked how they looked. So it does not surprise me that the higher ISOs would also look 'smoother'.

 

To my eyes at least they looked more Canon like (1DsM2).

[Ecaton]

One stop better coupled with faster lens options per focal length equivalent doesn't sound to bad.

Not considering the lux 21 and 24:

21mm f2.8 instead of CV 15 at f4.5, 24mm 2.8 instead of 18mm at 3.8, 28mm 2.0 instead of 21mm at 2.8, CV 35mm f 1.2 instead of 28mm at 2.0 etc

[Steve Ash]

-M8 high ISO performance = M9's (pixel for pixel)

 

This quote does not make sense to me. For what I see from the images available so far there is a valuable improvement in the high iso behavior.

 

To summarize my understanding from what I have read and seen so far is the following

1) there is a gain of minimum one stop in high iso

2) Leica decided to go no further with in-camera-noise-treatment with the advantage of preserving the details in high iso images.

3) This is contrary to the approach of their competitors who prefer higher levels of in-camera-noise-treatment at the expense of image details.

4) Leica's approach has the advantage that the user is free to add further noise treatment if desired or required but the disadvantage is that this requires some effort and knowledge.

5) The competitors approach has the advantage that no further noise treatment is required but the disadvantage that the details are lost for ever.

6) last but not least there is no other full frame digital rangefinder to compare with!

 

Some clarification from somebody who can present knowledge contrary to opinions would be appreciated.

 

Sean,

may be you can find a way to clarify without releasing all the details presented on your valuable review site.

 

Regards

Steve

[dalippe]

Okay, then maybe Leica should offer the M9 with either a CCD sensor or a CMOS? No problem discussing, but it really seems a damned if you do, damned if you don't situation for Leica.

 

No, that's not what I'm talking about. I guess you didn't read my original post in the thread in which I explain my understanding that having fewer pixels in the first place would have helped more with noise than does subsequent down-sampling. If I'm correct, this is part of how our D3s achieve what they do. And I haven't damned Leica for anything, which I think is also clear from my original post.

[Advertisement]

Hallo,

Look here: Leica M

[sean_reid]

Sean,

 

You end your ISO tests by saying, and I quote, 'In a section coming up soon, I will look at ISO noise in the corners (an important topic for the M9)'.

 

What is it to be expected here? This comment along with another comment you made on the CV 15mm seem to imply they are some performance issues in the corners of the M9. Could you give us a bit more information?

 

Thanks

 

Samir

 

It's not a performance issue but simply relates to what happens when corners are given a levels boost to offset luminance vignetting. That will increase the visibility of noise. Give me some time (I need to focus on client work this weekend as well as an S2 article) and I'll get into it. Theoretically, I'm also going to spend at least a bit of time with my family this weekend as well.

 

There's nothing about the corners noise levels that should be a deal breaker for anyone but it may be something Leica will want to revisit in future firmware (turning that correction down even further at high ISO levels).

 

Cheers,

 

Sean