Pages: 1 [2] 3   Go Down

Author Topic: M9 Review  (Read 18092 times)

tho_mas

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1799
M9 Review
« Reply #20 on: October 01, 2009, 03:27:17 pm »

Quote from: method
Users waiting for a Lightroom Profile can, of course, create their own with the DNG Profile Editor.
regarding the profile the comparision is completey useless as the "compare.jpg"-file has no profile  
Logged

jervisart

  • Newbie
  • *
  • Offline Offline
  • Posts: 2
M9 Review
« Reply #21 on: October 01, 2009, 03:42:43 pm »

I too noticed the camera shake on the canon shot, whats up with that. Also I don't see how the M9 could be perceived as better then the Sony but that is only one example shown as I suspect many more were done.  I would also like to point out to all the people who say the M9 provides the user a unique way of taking photographs.  All DSLR's allow you to turn off all the dodads and shoot completely manual. As far as quality is concerned Michael has proved the point that in no way is the M9 4 to 5000 dollars better then what you can get else where.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
M9 Review
« Reply #22 on: October 01, 2009, 03:47:38 pm »

Sorry Michael, good question!

Sharpening is a complex issue. Is the PSF (Point Spread Function) known, the original image can be reconstructed. This was pretty well demonstrated in the original images from the "Hubble Space Teleskop" which had a small error that could be reproduced. Now, the AA-filter works as a beam-splitter with well defined properties, so it's affects on the image may be reduced by appropriate processing.

On the other hand, passing image detail above the Nyquist limit will cause artifacts. The problem is that if this artifacts show up in color we call them "moiré" and react, if they are monochrome no one complains but they are still there. Aliasing may even improve perceived sharpness.

A very good example is here: http://a.img-dpreview.com/reviews/SigmaDP2...res_ACR-003.JPG

As you can see, there is a lot of false information in the image, but we don't react because we don't see colorful artifacts.

So, my point is that the issue is quite complex. Processing needs to be optimized for the image. Making objective comparisons is an illusion.

I may also point to this page that contains links to some excellent info from Zeiss, no easy read but worth the effort:

http://83.177.178.241/ekr/index.php/photoa...-and-perception


Best regards
Erik


Quote from: michael
Care to elaborate?

Michael
Logged
Erik Kaffehr
 

cmi

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 492
M9 Review
« Reply #23 on: October 01, 2009, 03:48:31 pm »

"polycarbonate wonders"



...ahahahaha priceless! Very good one! Just started reading but had to say this!

Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
M9 Review
« Reply #24 on: October 01, 2009, 03:54:13 pm »

Hi,

AFAIK there is a lot of magnesium, aluminium and stainless steel in those polycarbonate wonders...

Best regards
Erik


Quote from: Christian Miersch
"polycarbonate wonders"



...ahahahaha priceless! Very good one! Just started reading but had to say this!
Logged
Erik Kaffehr
 

zlatko-b

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 52
    • http://www.zlatkobatistich.com
M9 Review
« Reply #25 on: October 01, 2009, 03:54:29 pm »

A very enjoyable and thoughtful review!  Thank you!

To nitpick just a little,  my first impression of the Canon shot was also of shake, as if there was some vertical camera motion.

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
M9 Review
« Reply #26 on: October 01, 2009, 04:08:00 pm »

Quote from: ErikKaffehr
Sorry Michael, good question!

Sharpening is a complex issue. Is the PSF (Point Spread Function) known, the original image can be reconstructed. This was pretty well demonstrated in the original images from the "Hubble Space Teleskop" which had a small error that could be reproduced. Now, the AA-filter works as a beam-splitter with well defined properties, so it's affects on the image may be reduced by appropriate processing.

On the other hand, passing image detail above the Nyquist limit will cause artifacts. The problem is that if this artifacts show up in color we call them "moiré" and react, if they are monochrome no one complains but they are still there. Aliasing may even improve perceived sharpness.

A very good example is here: http://a.img-dpreview.com/reviews/SigmaDP2...res_ACR-003.JPG

As you can see, there is a lot of false information in the image, but we don't react because we don't see colorful artifacts.

So, my point is that the issue is quite complex. Processing needs to be optimized for the image. Making objective comparisons is an illusion.

I may also point to this page that contains links to some excellent info from Zeiss, no easy read but worth the effort:

http://83.177.178.241/ekr/index.php/photoa...-and-perception


Best regards
Erik

Erik, those articles are about the performance of lenses. I think in the context of the discussion, Michael's remark about taking care of the distinction between acutance versus resolution still makes sense. BUT I would add a thought: IF it's correct to say that a primary impact of the AA filter is to reduce acutance, then it would seem reasonable for one purpose of comparison to deploy a certain amount of acutance-restoring sharpening on the AA'd image, recognizing of course it is an extra step repairing what the AA filter damaged, not needed for the non-AA instrument. But if I were doing the comparison work between these cameras, I would still think it interesting and useful to see the difference of the initial starting point between the two sensor technologies before introducing any software fixes, and this is my sense of what Michael was trying to do. Different comparisons for different purposes are normal and sensible.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
M9 Review
« Reply #27 on: October 01, 2009, 04:18:17 pm »

Quote from: jervisart
As far as quality is concerned Michael has proved the point that in no way is the M9 4 to 5000 dollars better then what you can get else where.

I didn't read this into the article at all. As well, I hope you understand there is no particular reason why there should be a lock-step relationship between the prices of two cameras and the image quality they deliver. All kinds of other factors relating to production cost, build and component quality, features, and the market determine prices.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

jervisart

  • Newbie
  • *
  • Offline Offline
  • Posts: 2
M9 Review
« Reply #28 on: October 01, 2009, 05:08:07 pm »

Quote from: MarkDS
I didn't read this into the article at all. As well, I hope you understand there is no particular reason why there should be a lock-step relationship between the prices of two cameras and the image quality they deliver. All kinds of other factors relating to production cost, build and component quality, features, and the market determine prices.

You are right all those factors do go into making up a price point.  I suppose I have simply attached my own personal judgments to it.  As a professional money has to factor into all your decisions.  If you can get a tool that will, on output look the same to your client from a $2000 camera as it does from a $7000 one then you have just made $5000 more this year.  If a salesman comes to me and says listen if you spend this much money on a camera your clients will immediately  notice the difference and spend 3 times as much money on the same images you were doing before.  So really this is where I am coming from, you might be different, but I still think a camera manufacturer has to be able to justify there price.  And the most important justification is what that cameras final output is.  As the image quality is the same then the build component quality and features need to be really really good.

On another note about the sample comparisons. I thought about it a bit more and  it is actually a good test for IQ as the Leica has no mirror to shake the camera on exposure and the Sony has in body image stabilization for all lenses, you get a cleaner image.  The Canon has no stabilization on the 50mm and a big ol' mirror, hence camera shake. Clearly a shown benefit when applied to acquiring the most detail.
Logged

michael

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5084
M9 Review
« Reply #29 on: October 01, 2009, 05:22:00 pm »

Quote from: MarkDS
BUT I would add a thought: IF it's correct to say that a primary impact of the AA filter is to reduce acutance, then it would seem reasonable for one purpose of comparison to deploy a certain amount of acutance-restoring sharpening on the AA'd image...

Mark,

An AA filter's purpose is not to reduce acutance, it's to reduce resolution, which it does by blurring detail.

Michael
Logged

sojournerphoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 473
M9 Review
« Reply #30 on: October 01, 2009, 05:36:13 pm »

Quote from: jervisart
You are right all those factors do go into making up a price point.  I suppose I have simply attached my own personal judgments to it.  As a professional money has to factor into all your decisions.  If you can get a tool that will, on output look the same to your client from a $2000 camera as it does from a $7000 one then you have just made $5000 more this year.  If a salesman comes to me and says listen if you spend this much money on a camera your clients will immediately  notice the difference and spend 3 times as much money on the same images you were doing before.  So really this is where I am coming from, you might be different, but I still think a camera manufacturer has to be able to justify there price.  And the most important justification is what that cameras final output is.  As the image quality is the same then the build component quality and features need to be really really good.

On another note about the sample comparisons. I thought about it a bit more and  it is actually a good test for IQ as the Leica has no mirror to shake the camera on exposure and the Sony has in body image stabilization for all lenses, you get a cleaner image.  The Canon has no stabilization on the 50mm and a big ol' mirror, hence camera shake. Clearly a shown benefit when applied to acquiring the most detail.

Except that Michael said he used mirror lock up and a big tripod. The 1Ds3 image does look shaken to me - coud be a passing truck? -  or some lens aberation, but I presume the crop was from the centre field and it doesn't look like mine.

OTOH, I would still like an M9 because I like using rangefinders.

Mike
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
M9 Review
« Reply #31 on: October 01, 2009, 05:48:24 pm »

Quote from: michael
Mark,

An AA filter's purpose is not to reduce acutance, it's to reduce resolution, which it does by blurring detail.

Michael

Ya, but I didn't say the *purpose* was to reduce acutance - I was speculating that this is one of its *impacts*. I know the purpose is to mitigate moire etc. Anyhow, it turns my stomach every time I think about something blurring detail - unless of course it's the kind of picture which wants it's detail blurred - and even then I want to be in control of it. So kudos to Leica for taking the bold step of eliminating this piece of glass from the sensor. I wish Canon had done likewise with the 1Ds3, or at least offered photographers the choice - sure more engineering and cost, but I would have sprung for the no-AA model even if it cost a bit more.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
M9 Review
« Reply #32 on: October 01, 2009, 06:11:44 pm »

Quote from: pom
Isn't comparing unsharpened files to those from an AA filterless camera misleading given that by the time you have sharpened the AA'd files you will probably have more resolution whereas you cannot regain more resolution from a camera that has no AA filter? (honest question) Anyone else notice the moire on the M9 ruler shot?  

Infact following MR's last article, should it not be a comparison based on the final image that be used rather than unprocessed files?

One thing I did notice is the report that the highlight clipping now matches that in LR. As the LR histo is based on the enormous ProPhoto colour space that seems to me that the camera is telling you you have information where non will exist the moment you print or display, mapping is not a 'save all' solution relative to capturing the information in the first place based on the output medium. My own opinion though. It's actually why a lot of wedding shooters hate the ProPhoto space that LR uses, their output is never in such a wide colour space and they need to work with hundred of images fast, playing with mapping is not a solution unless you have the time or inclination to fiddle with each image seperately.

Nice review, I like the street pics too. The M9 is definitely an interesting camera, but I am unfortunately not in a position to look at it seriously now.

As far as detail/accutance goes, the best way would be to compare optimally processed files for each camera, and a print would in the end be even better. It is well known that some files take sharpening much better than others.

The followed image is a correctly sharpened handheld shot of a moving helicopter. It seems sharper to me than the sample shots of Michael (even after I tried sharpening them by the way).

http://www.flickr.com/photos/bernardlangui...647680/sizes/o/

There is indeed moire in the M9 files.

Now, the real question I gess is, what % of street shots taken with a M9 will be focused well enough that the lack of AA filter really brings value? I confess never having used an M series camera. Tthat will be fixed in a few days thanks to a colleague owning a M7 and M8, he is telling me that few of his M8 images appear sharp when looked at 100% on screen.

Cheers,
Bernard
« Last Edit: October 01, 2009, 06:54:21 pm by BernardLanguillier »
Logged

BurtonRandol

  • Newbie
  • *
  • Offline Offline
  • Posts: 1
M9 Review
« Reply #33 on: October 01, 2009, 10:34:35 pm »

I am not an expert on optical blur filters, but it is my impression
that their effect can be modeled as a convolution of the pre-filter
image by a suitable filter-dependent function. If this is correct, the
pre-filter image is in principle completely recoverable from the
post-filter data by deconvolution, which should therefore be performed
when a blur filter is present, if one wishes to study the resolution of
such a system. This technique does not affect the moire' reduction
properties of the blur filter.

Note that deconvolution is not the same thing as the usual sharpening
algorithms. The latter selectively boost high-frequency content in the
image, and so enhance zones of rapid transition.

Deconvolvers are available to photographers. For example, the program
Focus Magic appears to be one.


Best regards,

Burton Randol
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
M9 Review
« Reply #34 on: October 01, 2009, 11:06:07 pm »

Quote from: BurtonRandol
I am not an expert on optical blur filters, but it is my impression
that their effect can be modeled as a convolution of the pre-filter
image by a suitable filter-dependent function. If this is correct, the
pre-filter image is in principle completely recoverable from the
post-filter data by deconvolution, which should therefore be performed
when a blur filter is present, if one wishes to study the resolution of
such a system. This technique does not affect the moire' reduction
properties of the blur filter.

Note that deconvolution is not the same thing as the usual sharpening
algorithms. The latter selectively boost high-frequency content in the
image, and so enhance zones of rapid transition.

Deconvolvers are available to photographers. For example, the program
Focus Magic appears to be one.


Best regards,

Burton Randol

This is a very interesting idea if true, because the implication is that we could eat our cake and have it in terms of the AA filter. But I'm skeptical. Even if it were possible to model the function one would have to know the function in order to model it, and as for Focus Magic, yes, it is presented as a deconvolution program, but it's not the most easily controllable piece of software for such subtle effects.

Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

dseelig

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 596
M9 Review
« Reply #35 on: October 01, 2009, 11:32:39 pm »

the thing of the m9 that Michael has got very right is it size lenses and way of working. To worry about who has the absolute best image quality well go buy a medium format. The one thing Michael does not say is when shooting strangers people tend to stay much more relaxed in front of a leica then some huge dslr. They stay themselves.  Anyone want to run around with a canon or nikon d700 or 5d with a 35 or 28 1..4 , 50 1.4 85 1.4 and not have a shoulder ache at the end of the day . I can do that with my m9 . The dslrs for me are assignment cameras the leica for my soul and by the way I have never made more then 50 grand in a year. And no trust fund. I have leicas because I have to have them. David
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
M9 Review
« Reply #36 on: October 02, 2009, 12:12:03 am »

Hi,

The AA filter works as a beam splitter. It's characteristics should be carefully matched to the sensor and balanced to the microlenses and the fill factor. Increasing the fill factor reduces the amount of blurring needed to avoid aliasing.

Technically speaking Michael is right about not being possible to increase resolution, except that resolution is related to a certain level of contrast. Increasing contrast on very small detail may increase resolution. On the other hand resolution is actually determined by the Nyquist limit, we cannot have real resolution above that limit. According to the old version Bruce Fraser's book "Image Sharpening with Adobe Photoshop CS2" pretty aggressive sharpening is needed with a small radius (for the Canon 300D he mentions 500% at 0.6 radius), a camera from Kodak having similar resolution but no AA-filter needed only 210%.

Increasing the fill factor acts in the same direction as an AA filter, so if we increase fill factor we may have less need for AA-filter. Increasing the fill factor also reduces resolution.

To my best knowledge, it is impossible to remove aliasing in post processing, what can be done is to remove the colorful artifacts caused by interaction between aliasing and the "Bayer" filter array. On the other hand it seems that non color artifacts matter little to observers.

Best regards
Erik


Quote from: MarkDS
Erik, those articles are about the performance of lenses. I think in the context of the discussion, Michael's remark about taking care of the distinction between acutance versus resolution still makes sense. BUT I would add a thought: IF it's correct to say that a primary impact of the AA filter is to reduce acutance, then it would seem reasonable for one purpose of comparison to deploy a certain amount of acutance-restoring sharpening on the AA'd image, recognizing of course it is an extra step repairing what the AA filter damaged, not needed for the non-AA instrument. But if I were doing the comparison work between these cameras, I would still think it interesting and useful to see the difference of the initial starting point between the two sensor technologies before introducing any software fixes, and this is my sense of what Michael was trying to do. Different comparisons for different purposes are normal and sensible.
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
M9 Review
« Reply #37 on: October 02, 2009, 04:08:18 pm »

Hi,

The third installment of Erwin Puts's article on the M9 is released and available here: http://www.imx.nl/photo/leica/camera/page157/page157.html

He compares the M9 with M8 and Nikon D3X. His findings are:

The Nikon is slightly sharper than the M9
The M9 has issues with Moiré while the Nikon has almost none
The M9 is a major improvement over the M8

This may or may not contradict other authors experience. Mr. Puts is a well known expert on Leica cameras and certainly has no bias against Leica.

Best regards
Erik


Quote from: ErikKaffehr
Hi,

The AA filter works as a beam splitter. It's characteristics should be carefully matched to the sensor and balanced to the microlenses and the fill factor. Increasing the fill factor reduces the amount of blurring needed to avoid aliasing.

Technically speaking Michael is right about not being possible to increase resolution, except that resolution is related to a certain level of contrast. Increasing contrast on very small detail may increase resolution. On the other hand resolution is actually determined by the Nyquist limit, we cannot have real resolution above that limit. According to the old version Bruce Fraser's book "Image Sharpening with Adobe Photoshop CS2" pretty aggressive sharpening is needed with a small radius (for the Canon 300D he mentions 500% at 0.6 radius), a camera from Kodak having similar resolution but no AA-filter needed only 210%.

Increasing the fill factor acts in the same direction as an AA filter, so if we increase fill factor we may have less need for AA-filter. Increasing the fill factor also reduces resolution.

To my best knowledge, it is impossible to remove aliasing in post processing, what can be done is to remove the colorful artifacts caused by interaction between aliasing and the "Bayer" filter array. On the other hand it seems that non color artifacts matter little to observers.

Best regards
Erik
Logged
Erik Kaffehr
 

madmanchan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2115
    • Web
M9 Review
« Reply #38 on: October 02, 2009, 04:52:14 pm »

Burton's thoughts are spot on, but the key is knowing the behavior of the blur. Unfortunately it can be very complex.
Logged
Eric Chan

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
M9 Review
« Reply #39 on: October 02, 2009, 05:37:20 pm »

Quote from: madmanchan
Burton's thoughts are spot on, but the key is knowing the behavior of the blur. Unfortunately it can be very complex.

As I was thinking more about this idea, I'm wondering whether any of camera manufacturers would be prepared to divulge enough technical information about the nature of the blur so that a third-party, or if they won't divulge anything - they themselves could then market a "deconvolver" for it. This way the AA approach stays intact but we have a tool for undoing its damage. Too good to be true?
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."
Pages: 1 [2] 3   Go Up