Is Lightoom mangling M9 images? Or is there a compression issue?

[chris_tribble]

Advertisement (gone after registration)

Scott / Jaimie and others - I was able to grab ten minutes and took the shot below at ISOs 160 through to 2500. I didn't have a tripod at this house, but had a monopod - hence it's stable but not an exact replica frame. I used the 50 lux asph at f5.6 throughout. Speeds ranged from 1/60th through to 1/1000th. Images were processed in LR using my default settings which I posted elsewhere (http://www.l-camera-forum.com/leica-forum/leica-m9-forum/97845-m9-real-world-experience-3.html#post1032226)

 

The uncompressed raws can be downloaded from the links below. Remember, you have to register with DropBox to get access - takes a minute.

 

Very interested to hear what results you get. For what it's worth - I'm very happy with the M9 at 1250 and 1600 - and 2500 could be usable at a pinch.

 

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000438.DNG

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000439.DNG

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000440.DNG

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000441.DNG

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000442.DNG

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000443.DNG

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000444.DNG

http://dl.getdropbox.com/u/862415/Leica_ISO/L1000445.DNG

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Advertisement]

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

[chris_tribble]

CORRECTION - the 100% images I sent had been upsized slightly - so weren't a fair representation. The attached crops are as is.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Shootist]

CORRECTION - the 100% images I sent had been upsized slightly - so weren't a fair representation. The attached crops are as is.

 

Thank you Chris, and especially for the updated images. I started to wonder about the first set, not to impressive. Second set much better. Now if I only had 7 grand to spare. I'll still enjoy using my M8.

So you are Ok with the output from LR/ACR and don't see the need to go to C1.

 

I hear so much about C1 but everytime I try using it I don't see what the fuss over it is. I still like the output from ACR.

[chris_tribble]

I'm hoping for an improved profile for ACR / LR - I loathe the workflow with C1 because I never got round to learning it! Interested to hear what the forum's RAW processing gurus have to say!

 

[erl]

A bit off topic, but related. I can't get C1 to recognize DNG's from the M9. Compressed or uncompressed.

 

Am I missing something obvious?

 

I have no desire to learn Lightroom at present as I am well versed in C1. Learning the new camera is enough for me at present.

 

Any quick advice greatly appreciated. Then I can start supplying files as required by willing testers.

[rvalarcher]

try C1 4.8.3 !

[Shootist]

Advertisement (gone after registration)

A bit off topic, but related. I can't get C1 to recognize DNG's from the M9. Compressed or uncompressed.

 

Am I missing something obvious?

 

I have no desire to learn Lightroom at present as I am well versed in C1. Learning the new camera is enough for me at present.

 

Any quick advice greatly appreciated. Then I can start supplying files as required by willing testers.

 

You need to download and install the latest version of C1. As stated above 4.8.3.

[erl]

rvalarcher & shootist. Thank you both.

 

My next pic will be of my red face.

[t024484]

I'm hoping for an improved profile for ACR / LR - I loathe the workflow with C1 because I never got round to learning it! Interested to hear what the forum's RAW processing gurus have to say!

 

 

Chris,

 

Since you have acces to a M9, could you please provide me with 6 DNG's as described in posting 18 and 19.

This will enable me to do the investigation.

 

Hans

[chris_tribble]

Hans - sorry I'd not taken on board the request. I've now done this and uploaded them as a Zipped file to my FTP site. You can download from:

 

http://dl.getdropbox.com/u/862415/ISO_TEST_M9.zip

 

I've just started to upload this (it will take some time) - hope you're able to use it - and very interested to see the results.

 

V Best

[t024484]

Hans - sorry I'd not taken on board the request. I've now done this and uploaded them as a Zipped file to my FTP site. You can download from:

 

http://dl.getdropbox.com/u/862415/ISO_TEST_M9.zip

 

I've just started to upload this (it will take some time) - hope you're able to use it - and very interested to see the results.

 

V Best

Chris,

 

Thank you for making these files available.

 

I have analysed the files and came to the following results:

 

1) a DC offset voltage has been applied before digitizing. The numerical level of this voltage starts at 44 for ISO 160, 54 for ISO 200, 68 for ISO 250 and so on, all the way and almost fully linear up to 624 at ISO 2500.

This DC offset voltage has probably been applied to have a more accurate way of calculating the black point. The white point in all cases is 16.383.

 

2) All ISO steps are done in the analog domain, to keep the noise as low as possible.

 

3) The Read Noise has an incredibly low figure of 1.05 ADU at ISO 160. while the A/D noise is also very very low at 1.15 ADU.( ADU meaning the digital number coming from the A/D converter, figures only for uncompressed DNG).

At ISO 200 this means a combined noise of 1.75 ADU, compared to a value of 2.75 for the NiKON D3 !

From a hardware point of view, the physical limits are more or less reached.

 

Noise will be dominated by shot noise, apart from the very darkest areas, below ca. 2 ADU at ISO 160 and below 10 ADU at ISO 2500, as from where the combined Read plus A/D noise will become more dominant.

This is a big improvement over the M8, where this crossing point lies at ca. 16 ADU for ISO 160 and at 17 ADU at ISO 2500.

 

Hans

[chris_tribble]

Hans- many thanks for the precise technical account. From a lay perspective are you able to summarise any of the real world implication of this analysis? eg

 

• what will the likely impact of different light sources (tungsten etc) on high ISO performance
  
• what tolerance is their likely to be to under / over exposure - I assume shooting to the right will still hold strongly...
  
• what looks like the best possible choice for a top ISO to work with - 1250, 1600, 2000, or 2500
  
• etc...
  

 

Thanks for any thoughts...

[t024484]

Hans- many thanks for the precise technical account. From a lay perspective are you able to summarise any of the real world implication of this analysis? eg

 

• what will the likely impact of different light sources (tungsten etc) on high ISO performance
  
• what tolerance is their likely to be to under / over exposure - I assume shooting to the right will still hold strongly...
  
• what looks like the best possible choice for a top ISO to work with - 1250, 1600, 2000, or 2500
  
• etc...
  

 

Thanks for any thoughts...

Chris,

 

1) To answer this question, a noise analysis against exposure value has to be made for the M9 for each individual color. Imatest is one of the tools that can do the job.

The impact of various light sources on noise will directly depend on the color content of this light divided in red, green and blue causing color noise for each of this three colors within the M9.

As a first guess I would say, the more green in the light, the less noise, and blue generating the most noise, so the higher the kelvin temp of the light source, the more noise, after having corrected the white balance.

2) Shooting to the right is always to be preferred, because shot noise is proportional to the SQRT of the sinal. So the more signal, the better the Signal to Noise ratio.

With the M9 however, you can recover more from the dark as was possible with the M8.

3) I cannot say anything on the best possible high ISO value to work with. It depends on so many factors as taste, type of pictures, color or B&W, preferences etc.

 

Hans

[chris_tribble]

Hans - again, many thanks. I do wish colleagues would read this as it strikes me as being the most cogent account of critical differences between the M8 and M9 image sensor / processing capacity.

 

A thought - is it worth repeating the process with COMPRESSED DNGs - for the moment I'm avoiding using these as I'm not confident that Leica's got the compression algorithms as right as they could be.

 

Very best - and thanks again.

[t024484]

Hans - again, many thanks. I do wish colleagues would read this as it strikes me as being the most cogent account of critical differences between the M8 and M9 image sensor / processing capacity.

 

A thought - is it worth repeating the process with COMPRESSED DNGs - for the moment I'm avoiding using these as I'm not confident that Leica's got the compression algorithms as right as they could be.

 

Very best - and thanks again.

 

Chris,

Thanks for your friendly words.

You are right about not using compression. This adds half a stop of noise over almost the entire exposure range, 8 bits are not enough for lossless compression.

 

You triggered however something that kept me thinking of all the time, being the effect of the color temp of the Light source on the pictures noise content.

That is because in various examples, we have seen the results of M9 pictures at High ISO values, compared to other camera brands being used at the same ISO setting.

What strikes me is that there is a large difference between the various tests.

In some tests the M9 is almost as good as the 5DII at ISO 2500, in other tests there is a large difference to be seen.

Hoping that all pictures were made shifted to the right as much as possible, the thing that is definitely different in all tests, is the color temp of the light.

Some test have been performed indoors with artificial light, and others did their tests in the open air.

 

So my question goes to one of my my favourite testers, Sean Reid,

Sean I hope you are reading this thread, is it an idea to perform the noise test you have done between the M8.2, M9 an Canon 5DII with a light source with a much higher color temp and see what the result are noise wise.

 

Hans

[ennjott]

When comparing against a 5D2, it should first be tested if the ISO speed settings are identical, because newer Canon models are 1/3 of a stop less sensitive at the same ISO rating than older ones (30D and lower).

Also note that the 5D2 only has ISO 100 to 3200 in full stops, everything else is firmware tricks.

 

I think the M9's ISO2500 mode looks very good in this example. Try reducing chroma noise in the raw converter (no luma NR!). The result should be interesting.