CCD vs CMOS: Can you tell which is which?{merged}

[BerndReini]

Advertisement (gone after registration)

I am sure that this could be remedied in post-processing, but maybe users are simply guided in the wrong direction with these files.

 

I'll just repost this since as I said, neither Michael Reichmann nor Ming Thein has created anything that looked like their work from CCD cameras. Can they? Probably, but I guess they now like a different look for better or for worse. I encourage everyone to look at the photographs in the tests of the 645Z and tell me that they aren't processed more flat than any of the shots from the P25 in older reviews.

[Advertisement]

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

[Sharpdressed]

I'll just repost this since as I said, neither Michael Reichmann nor Ming Thein has created anything that looked like their work from CCD cameras. Can they? Probably, but I guess they now like a different look for better or for worse. I encourage everyone to look at the photographs in the tests of the 645Z and tell me that they aren't processed more flat than any of the shots from the P25 in older reviews.

 

Older ccd cameras had a narrrower dynamic range so the images looks a bit more punchy. Also probably kodak experience in colors had helped the development of a whole series of ccds which had a more "film like" color profile.

[Lenshacker]

A bigger file size may also be caused by more noise (e.g. pushing shadows even at low ISO).

 

The image was downsized to take noise out of the equation. I've been looking at data values in the uncompressed DNG files for the M8, M9, and M Monochrom. The latter is very uniform. I've got a routine for calculating some of the noise parameters, will look to put it into the code. My wife took an image processing course in the late 1980s, my "Honey Do List" was typically writing code for her.

 

The extra processing put into the M240 images should have increased edges, but it is hard to say. What would be interesting to see is an M240 picture with/without the IR cut filter. Camera on a tripod, lens hood, of a static scene under full sun. If the JPEG for the image with the IR cut filter requires more space to store, we have an answer.

 

I assume that David did not use DNG-8 for the M9. The compression artifacts with the coarse step size would introduce some sharp edges and this would require high-frequency to reproduce. I learned that almost 35 years ago with imagery from an early Digital IR sensor that we made. I wrote the code to calculate the "Confidence Integrals" for the sensor measurements.

[IWC Doppel]

The result is that there is no CCD look that can't be created with a CMOS camera. Most people just don't do it. As a big fan of the look of the M9 I will admit as much, and I expect people to respect that a lot of M9 users love the results they get from their M9 with minimal processing. I think this could be the conclusion of the argument and both camps can be content with it.

 

I differ in my views :

 

I couldn't get the M CMOS files to look as nice as CCD M9 files or look the same, I tried really hard but couldn't. The curves are also notably different. I am not sure this is possible and if it is how many can do it and how long it takes.Just look at how 'easy' it was to reduce skin redness and how consistent the results where This is people prepared to post their results on a forum not struggling at home with many sliders

[jaapv]

Now this makes me wonder. Images should be processed to the look that suits the subject in the eyes of the photographer. Nothing less, nothing more. I think that that goal can be achieved by files from many cameras, including the M9 and M240. It is futile to try and emulate one look over the other, with different filtering, processing engines, interpolations and postprocessing changes complicating matters.

David's approach of processing the images to his satisaction certainly appears quite rational.

[IWC Doppel]

I think I agree with most of what your saying, it's only that I look at a lot of pictures and do see a different look most of the time between M9 and M240

 

Perhaps I'm wrong, but I don't think so. I'm sticking with my view

[lct]

Advertisement (gone after registration)

David's approach of processing the images to his satisaction certainly appears quite rational.

For David and photogs sharing the same tastes i guess but telling people that they have to tweak files in PP to get such or such expected result is the best way to convince them that such files are basically different. It's the problem of comparos based on subjective preferences... Showing to same people that two cameras have a similar IQ when using the same raw converter with the same settings is another story but i suspect it would not work here, at least with LR, which would tend to prove that out-of-raw-converter as well as out-of-camera results are indeed different...

[jaapv]

Actually I am telling them to tweak their files to get the best/most pleasing to their taste results and not to get hung up on hardware considerations. You cannot print a camera, only its photographs

[dfarkas]

For David and photogs sharing the same tastes i guess but telling people that they have to tweak files in PP to get such or such expected result is the best way to convince them that such files are basically different. It's the problem of comparos based on subjective preferences... Showing to same people that two cameras have a similar IQ when using the same raw converter with the same settings is another story but i suspect it would not work here, at least with LR, which would tend to prove that out-of-raw-converter as well as out-of-camera results are indeed different...

 

I would of course hope that the results are different. The M240 features more advanced technology. If it didn't hold some performance advantage, it wouldn't be much of a successor to the previous model. In this case, the M240 has more dynamic range and lower noise. These advantages translate into more malleability in post processing, better low light performance, cleaner shadows, more highlight recoverability, etc. As I wrote in my conclusion, as a result of the broader dynamic range and the default profile in LR, images look flatter compared to those from the M9. And, as I also showed, just a few basic edits are all that are necessary to get you 90% of the way there. These can be saved as a preset and applied automatically upon import.

 

When I was first getting started in professional digital imaging 25 years ago, flat files from drum scanners were the norm. We called this a linear tone curve. Only when an S-curve was applied was the image anything but flat muck. But, that flat muck allowed for the most amount of correction and modification with the least amount of loss. Files destined for press also routinely had flat blacks to allow for dot gain. This is nothing new. The human eye prefers contrast as this is how we perceive our surroundings. Our eyes and brains are miraculous in their ability to simultaneously delivery both a wide dynamic range and contrast. Photographic tools, be they analog or digital, cannot....yet. So, we must either accept a clipped file out of camera that looks pleasing but has little ability to be modified, or a flatter image that can have contrast applied to it for output once other edits are performed. I prefer the latter, obviously.

 

I see very little value in comparing out-of-camera results. Even in the same software, different profiles are applied. If custom profiles were created for each camera under the same lighting situation (ex. daylight-balanced strobe) using color checkers and applied upon import, then perhaps we have something to compare. A user here posted test shots using this basic methodology and the results were awfully close. Of course, these profiles lose relevance with different light sources.

 

So, why didn't I do that? Well, besides the fact that I rarely shoot under controlled lighting for this type of walk-around photography, I truly believe that the only thing that truly matters is the final result. I will continue to edit my photos to my desired look. This look may change over the years as I'm exposed to various styles and influences.

 

I had a film professor in college who would end lively group critique sessions with one simple question: "But.... does it work?" No matter the comments for or against the technique, the style, the subject, etc. the only thing that mattered in the end was whether it "worked". This alone would determine your grade.

 

I can't think of a single photograph used for any commercial or fine art purpose that was ever used without any modification or enhancement. In the days of film, photographers were either accomplished printers themselves, or relied heavily on master printers they worked with for many years. In the digital world, professional photographers often work with retouchers and editors.

[dfarkas]

Oh, in case you guys missed it, I posted my Part 3 conclusion yesterday. I posted the raw voting data, revealed which images were which and provided my analysis. Check it out.

 

The Great Debate: CCD vs CMOS - Part 3

[CheshireCat]

What would be interesting to see is an M240 picture with/without the IR cut filter. Camera on a tripod, lens hood, of a static scene under full sun. If the JPEG for the image with the IR cut filter requires more space to store, we have an answer.

 

That would be interesting to see. However, the extra IR "ghost" could add more information in some cases, hence make the file bigger.

If possible, the same scene should be shot with two 50 lenses (an APO lens and a non-APO lens), as follow:

1a) APO + IR cut

1b) APO no filter

2a) Non-APO + IR cut

3b) Non-APO no filter

 

I assume that David did not use DNG-8 for the M9. The compression artifacts with the coarse step size would introduce some sharp edges and this would require high-frequency to reproduce. I learned that almost 35 years ago with imagery from an early Digital IR sensor that we made. I wrote the code to calculate the "Confidence Integrals" for the sensor measurements.

 

I am not sure the difference would be detectable after gamma compression. This has been Leica's argument in defense of DNG-8 for years.

To verify your hypothesis, it should be enough convert a linear DNG (no postprocessing) to DNG-8. Then compress both files to JPEG and compare their size.

[IWC Doppel]

I do see the point, in a way. For me after a glass of wine and an old iPad, I think nearly all M9 shots are image 1, there I've stuck my neck out

 

I did pick a couple of 2's and some were gut n guess

 

Hmmm..... Why do others find this so hard to see ?

[IWC Doppel]

I think all of these are the M240, there were a lot I did stop half way down. This could be different processing. Oops stuck my neck out again

 

There you go I'm still maintaining my position regarding CCD..... I for one CAN tell which is which

[IWC Doppel]

PS I didn't do part 3, no point

[dfarkas]

PS I didn't do part 3, no point

 

Part 3 is just a conclusion, not a test.

[CheshireCat]

Here are my original test results with reason why I picked that one:

 

[OK] Streetcar:1 (More vivid colors, especially orange)

[OK] Fishing Dock:1 (More vivid colors)

[OK] Life Preserver:1 (Blue shadows in bars)

[ERR] Fishing Boats: 2 (Water blue color)

[OK] Bay Boat Tour: 2 (More pop)

[ERR] Sailboats: 2 (Greens)

[OK] Magenta Tulips: 1 (Dynamic range)

[OK] Steps: 2 (Blue shadows, sky, and dynamic range)

[OK] Pier 39: 1 (More pop, less greenish cast)

[OK] Church window: 2 (Better colors, less greenish cast)

[ERR] Scarves: 2 (Bluer shadows, moire - but greenish cast ???)

[OK] Red Cards: 2 (More vivid colors, especially orange)

[OK] Slippers: 2 (more pop, less dynamic range... but may be wrong PP)

[OK] Skyscraper: 1 (better colors, less greenish)

[OK] Modern Building: 2 (more pop, less dynamic range)

[ERR] Gate: 1 (more pop)

[OK] Apartments: 1 (more pop, better colors)

[OK] Walkway: 2 (more pop, less greenish, much better magenta)

[OK] Lombard: 2 (more pop, blue shadows, less greenish, much better magenta)

 

I assure, no cheating.

I scored 15 over 19 (about 80% accuracy) in a test in which M240 images were manipulated in order to fake M9 images.

The conclusion is: YES, there is a difference between M9 and M240. At least for me.

David, thanks again. This has been a nice exercise.

[CheshireCat]

It is also important to point out that I have extensively used both cameras, unlike - I guess - most voters.

 

The test should have included an extra question: "Have you used both the M9 and the M240 ?", splitting voters in two groups. Results for the "YES" group might have been very different.

[Lenshacker]

 

 

I am not sure the difference would be detectable after gamma compression. This has been Leica's argument in defense of DNG-8 for years.

To verify your hypothesis, it should be enough convert a linear DNG (no postprocessing) to DNG-8. Then compress both files to JPEG and compare their size.

 

I am having a lot of fun with this. I was able to reconstruct most of the image from the residual error alone. I will alter the subroutine to write the truncated data to disk, rather than the residual error.

[Advertisement]

Hallo,

Look here: Leica M

[thighslapper]

Thanks Dave.

 

Your conclusions match mine exactly........

 

.... and as was noted very early on by those of us who took the plunge into the unknown with the M240 .... the files are much more malleable and can be subject to a huge degree of manipulation without degradation of image quality.

 

Give me that every time over the M9 output .......

 

My usual initial gambit with 'flat' images is to use Auto tone in LR (which hideously over-exposes in addition to the other adjustments) then roll off exposure by -0.5-1.0 ...... which invariably produces nicely saturated and more contrasty images that are then ideal for posting online...... quick and dirty .... but it often works fine.

[Jeff S]

I find it interesting that many Monochrom (CCD) admirers laud that camera's high dynamic range and resultant flat, but highly malleable, files….while many don't appreciate some similar qualities and PP flexibility from the M240 (CMOS) files that David has demonstrated.

 

Jeff