Is the difference of DR on MFDB vs 35mm dslr discernible on print?

[ErikKaffehr]

Hi,

Just to add to the confusion I include another sunlit picture with clouds and shadows.

Here is original image (as DNG) and full size JPEG:

http://83.177.178.241/images/Jim2/index.html

[attachment=16693:20090604...32_small.jpg]

The reason I added the picture that this is a typical and not very demanding scene. It's not for artistic value. I made small fairly routine adjustments for the JPEG.

The picture is taken with a Sony Alpha 900

Best regards
Erik

I'm currently shooting landscape as a hobby, hoping that one day I can sell the ones I consider great. I'm currently using 1ds3 and have been wondering about whether to get an MFDB + view camera. The MFDB would give me a higher res and the view camera would give me better Dof on grand scheme type of shots.

- I'm wondering whether I should wait for 1ds4, hoping it would have a higher resolution + better DR (if that's even possible?) and use either the Canon TS lens or Cambo X2

Assuming (and please confirm?) that MFDB would have a much better DR than 1ds3 or future 1ds4 (speculative?), would the edge in DR show up in prints with the current technology? I'm using a Canon ipf6100 printer at the moment but the future might offer us better printing technology too.

I guess this is all technical and at the end of the day a nicely captured shots with 1ds3 would be quite nice too and I know a few professional landscape photographers who sell their photos for a living use 5d2.

Thanks for your input / comments / advice / suggestions / thoughts.

[Ray]

Interesting. If D3X is better than P65+ in DR, in your opinion is there any reason that you can think of, to use P65+ at all?

Of course there is. Having lots of pixels is wonderful when you want large prints. No-one can deny the resolution advantages of the larger and higher pixel-count sensor. You can downsample a large image to create a perception of reduced noise and even a slight increase in accutance, but you cannot create additional resolution by upsampling a smaller file.

I also like the idea of no AA filter, but I understand that many users of DSLRs shoot in jpeg mode and that aliasing artifacts, which the average user might not have either the skill or the inclination to remove, might be disturbing.

[ErikKaffehr]

Jim,

I'd suggest that DR is not really the issue that matters. On the other hand, it seem that observers can tell apart MFDB prints from DSLR prints even at reasonable sizes.

http://luminous-landscape.com/forum/index....st&p=310104

There are some questions tough, are we doing a apples to apples comparison? We normally use DSLRs and MFDBs differently. To begin width DSLRs are often used for free hand shooting and MFDBs are often used on tripod. DSLRs are often used with zooms and MFDBs are often used with fixed focals. Putting first class primes (like the Zeiss lenses) on your existing DSLR may improve your pictures? I'd say this is a consideration because the price of an MF-body pays for 2-3 Zeiss lenses not to mention the MFDB.

Check also this (especially that you go need to Sydney to find a dealer)
http://www.josephholmes.com/news-sharpmediumformat.html

Best regards
Erik

Interesting. If D3X is better than P65+ in DR, in your opinion is there any reason that you can think of, to use P65+ at all?

[Ray]

I'd suggest that DR is not really the issue that matters. On the other hand, it seem that observers can tell apart MFDB prints from DSLR prints even at reasonable sizes.

Now Erik, you know about Michael's famous G10/P45+ comparison. At a reasonable size of A3+, the prints from both cameras were indistinguishable, apart from the shallower DoF of the P45 image.

With a different subject, the DR differences would have been apparent, but not necessarily if the D3X had been substituted for the G10 in the comparison.

DR is often an issue for landscape photographers who are trying to capture natural scenes with a very high SBR. It's the reason why we have HDR programs.

[ErikKaffehr]

Ray,

I agree with what you say, the idea I'd like to conceive is like this:

1. DR is a resource for raw-processing, nothing that readily shows up in prints.
2. Reichmann and Doug Peterson et. al. clearly say that differences are visible even in small prints.
3. No one has really explained what the difference is
4. The difference is described in terms like "microcontrast" and "tonality" which I'd not consider well defined

I'm aware of MR's G10/P45+ comparison it's interesting but MR has since that upgraded to P65+ and M9, so I'd suggest that Michael doesn't think that G10 is good enough for all his photography but found that it's surprisingly good.

Going back to DR it is my opinion that DR is seldom a real problem with DSLRs. Any image really utilizing the full DR of anything more than a phone-cam will be flat in print (if no S-kurve is applied), because prints have small dynamic range, like 1:200 or about 7.2 stops (and I'd say that's optimistic). Making the print attractive takes some curve adjustment to get a steep gradient on midtones and have good separation in highlights and shadows. Now days some photographers do "virtual HDR", different developments from the same raw image, and combine in PS or some HDR tool.

All this processing  plays probably a larger role than small differences in DR.

Best regards
Erik

Now Erik, you know about Michael's famous G10/P45+ comparison. At a reasonable size of A3+, the prints from both cameras were indistinguishable, apart from the shallower DoF of the P45 image.

With a different subject, the DR differences would have been apparent, but not necessarily if the D3X had been substituted for the G10 in the comparison.

DR is often an issue for landscape photographers who are trying to capture natural scenes with a very high SBR. It's the reason why we have HDR programs.

[Jim2]

I think my conclusion is to get an MFDB if I can justify the spending from monetary point of view. I'm looking at getting the P40+ if not P65+. Thanks for all the info it has been an interesting thread to follow. I managed to learn a bit more on other topics also.

I have another question but probably best for dpreview - it is in regards to price

[Ray]

Any image really utilizing the full DR of anything more than a phone-cam will be flat in print, because prints have small dynamic range, like 1:200 or about 7.2 stops ...

Not at all, Erik. I think this is a grave misconception.

All you have to do to falsify this argument is take a series of shots of your living room, exposing for the view out of the window, using a P&S camera, an average DSLR, a D3X and a P65+. Process and print each image so both the scene out of the window and the scene inside your living room is as natural as possible.

You'll find it's not possible with the P&S. Your lounge room will look like crap. With a DSLR it will look better. With a D3X or P65+, both the view out of the window and your lounge room will likely look fairly natural on the print, with no blown highlights in the scene out of the window and no serious noise in your lounge room. This is because both the P65+ and the D3X have a high DR capability.

[ErikKaffehr]

Ray,

I'm not really sure. One issue with P&S cameras is that they shoot JPEGs. I'm quite certain that inherent DR of any PS is in excess of what can be reproduced on paper. Simple P&S cameras seldom have RAW format, so gamma correction and S-curve are already done on the JPEG. I certainly would expect any digital camera to have a sensor side DR in excess of 7.2 stops.

My statement "Any image really utilizing the full DR of anything more than a phone-cam will be flat in print" is a bit taken out of context and should have expressed it more clear. The issue is that an image containing the full dynamic range of a picture cannot be reproduced in print. To get a decent print we need to apply a compression on the shadows and highlights so we can have a steep gradient on the mid tones. In camera JPEGs do this automatically and so do raw converters.

Check this link: http://www.normankoren.com/Canon_PWPro_toneadj.jpg

The whole article is below: http://www.normankoren.com/digital_tonality.html


I won't really argue on this issue, I wanted to make a point clear, and certainly did exaggerate.

Best regards
Erik

Not at all, Erik. I think this is a grave misconception.

All you have to do to falsify this argument is take a series of shots of your living room, exposing for the view out of the window, using a P&S camera, an average DSLR, a D3X and a P65+. Process and print each image so both the scene out of the window and the scene inside your living room is as natural as possible.

You'll find it's not possible with the P&S. Your lounge room will look like crap. With a DSLR it will look better. With a D3X or P65+, both the view out of the window and your lounge room will likely look fairly natural on the print, with no blown highlights in the scene out of the window and no serious noise in your lounge room. This is because both the P65+ and the D3X have a high DR capability.

[bjanes]

bjanes, would you care to explain to me a bit more on this please?

The part that I don't understand is how does the 93 in gamma 2.2 becomes a pixel value of 28 in 8 bit?

I would assume that to convert 28 in 8 bit into the equivalent value in 14 bit would be: 28 / 2^8 * 2^14 =  1792... how did you get 1781?

I'm probably going to sound very dumb here: "very far above clipping" sounds weird to me because I thought clipping means values (i.e. luminance?) above certain points would be 'clipped' i.e. assigned the maximum value be it 255 for 8 bit, or 2^14-1 for 14 bit data. Don't you mean "very far below clipping" ?

I think I need to really sit and think about all this to understand the whole situation about clipping / raw processing etc...

blah all I want is to take beautiful photos to print and hang on the wall! My maths (discrete or calculus or whatever) have all evaporated off my brain many many years ago and I'm not even that old lol

Is there a page that explain this whole clipping thing that you demonstrated with photoshop? I'd like to read it to learn more.

Jim,

You have the right idea, but when doing gamma calculations, one first normalizes the pixel value. The 8 bit gamma 2.2 value is 93. Normalized, this becomes 93/255 = 0.364706. To convert between gamma of 1.0 to 2.2 you do the power calculation, 0.364706 ^ 2.2 = 0.108711. Then you convert back to a bit depth of 8: 0.108711 * 255 = 28. Clipping can be at either end of the spectrum--highlights or shadows. In the case under discussion, the highlights appeared OK, but Jack thoughts the darker portions appeared bad, so the shadows would be clipped in this case, but obviously they were not. If here were more analytical, he would look for another explanation.

The Threshold command in Photoshop is reached via the menu command Image, Adjustments, Threshold or from the new Adjustments panel is described briefly in the Photoshop help, "Create a two-valued black and white image".

[Ray]

Ray,
My statement "Any image really utilizing the full DR of anything more than a phone-cam will be flat in print" is a bit taken out of context and should have expressed it more clear. The issue is that an image containing the full dynamic range of a picture cannot be reproduced in print. To get a decent print we need to apply a compression on the shadows and highlights so we can have a steep gradient on the mid tones. In camera JPEGs do this automatically and so do raw converters.

Erik,
Your statement is very misleading and seems to ignore the entire 'raison d'etre' of editing programs like Photoshop. Any scene, however great the dynamic range captured in the image, whether in a single shot from a camera with a high DR capability, or in a series of bracketed exposures merged to HDR, can be represented on print in a realistic and convincing manner with appropriate processing.
 
But don't take offense   .

[Dick Roadnight]

I think my conclusion is to get an MFDB if I can justify the spending from monetary point of view. I'm looking at getting the P40+ if not P65+. Thanks for all the info it has been an interesting thread to follow. I managed to learn a bit more on other topics also.

I have another question but probably best for dpreview - it is in regards to price

Have you considered the Hasselblad H3D11-60 (or 50)?

The 60 will be available "in September" (which might mean Christmas), and "should" have all the advantages of a Hasselblad, and none of what I would expect from a Mamiya/Phase.

To someone who graduated from Mamiya (C330) to Hasselblad almost exactly 30 years ago, it seems incredible that anyone could consider them comparable... but Hasselblad is not Swedish anymore, and they do not use European lenses...

Apparently the DCU (digiback) is made in Europe, and I use my DCU on European (Sinar) cameras with European Lenses (Schneider).

One major factor you will need to consider is local dealer support for the two brands.

Here I have about half a dozen MFD retailers within two or three hours drive, but I gather that you do not.

DR depends on how many photons your pixels can gather... and from how many photons the associated analog components can store the signal (as a ratio to noise) so, for DR, larger pixels are better, so, ¿if DR is what you want, perhaps the Leica M9 might be the best choice? ... but I would not recomend it for landscapes!

[Jim2]

Have you considered the Hasselblad H3D11-60 (or 50)?

The 60 will be available "in September" (which might mean Christmas), and "should" have all the advantages of a Hasselblad, and none of what I would expect from a Mamiya/Phase.

To someone who graduated from Mamiya (C330) to Hasselblad almost exactly 30 years ago, it seems incredible that anyone could consider them comparable... but Hasselblad is not Swedish anymore, and they do not use European lenses...

Apparently the DCU (digiback) is made in Europe, and I use my DCU on European (Sinar) cameras with European Lenses (Schneider).

One major factor you will need to consider is local dealer support for the two brands.

Here I have about half a dozen MFD retailers within two or three hours drive, but I gather that you do not.

DR depends on how many photons your pixels can gather... and from how many photons the associated analog components can store the signal (as a ratio to noise) so, for DR, larger pixels are better, so, ¿if DR is what you want, perhaps the Leica M9 might be the best choice? ... but I would not recomend it for landscapes!

Thanks for the tip. I will also look into Hasselblad back. As for the 'body' I'm considering one of these three: arca swiss m-line 2, cambo wrs 1000, horseman ld pro. I have also seen that sinar has a technical body but I read that they're going out of business?

[ThierryH]

Very wrong reading and understanding: they (Sinar) are still there and very much alive.

If I would be you I would really have a look to the Sinar arTec camera and TRY it in real.

Here a link:

http://www.sinarcameras.com/site/index__ga...s-rand-846.html

Best regards,
Thierry

I have also seen that sinar has a technical body but I read that they're going out of business?

[ErikKaffehr]

Ray,

No offense, I actually mean it the other way round. To reformulate it:

- DR is helpful in capturing all brightness levels in a scene
- If the dynamic range of the scene does not fit print it must be compressed in a way which is pleasant to the eye/brain combination, otherwise the image will be flat

The important message is that the visual impact of the image will probably more effected by processing than by DR alone. But any way we don't need to waste bandwidth on this issue.

Best regards
Erik

Erik,
Your statement is very misleading and seems to ignore the entire 'raison d'etre' of editing programs like Photoshop. Any scene, however great the dynamic range captured in the image, whether in a single shot from a camera with a high DR capability, or in a series of bracketed exposures merged to HDR, can be represented on print in a realistic and convincing manner with appropriate processing.
 
But don't take offense   .

[bjanes]

DR depends on how many photons your pixels can gather... and from how many photons the associated analog components can store the signal (as a ratio to noise) so, for DR, larger pixels are better, so, ¿if DR is what you want, perhaps the Leica M9 might be the best choice? ... but I would not recomend it for landscapes!

The full well (number of photo electrons captured) is only part of the DR equation, since it determines the noise in the highlights. For the shadow noise, read noise (depending on where the noise floor is placed) is the most important factor. The engineering definition of DR is full well/read nose, both expressed in electrons. Read noise with big pixels with current cameras is relatively high at low ISO but decreases as the ISO increases. A consequence of this relationship is that a small pixel camera can have a good dynamic range as compared to a larger pixel camera when both operate at base ISO. However, at higher ISO, the large pixel camera will have a distinct advantage. See Figure 5a and the accompanying text in Roger Clark's article. Roger is using the engineering definition of DR, which is on a per pixel basis. The useful photographic dynamic range would place the noise floor higher, but the same principles apply.

For a more complete analysis, see Big Pixels vs Small Pixels by Emil Martinec, a photographer and professor of physics at the University of Chicago. To fully understand these principles, it helps to have a PhD in physics, but the rest of us can at least have an intuitive grasp of them. BTW, Roger Clark has a PhD from MIT in astrophysics and works with NASA digital imaging projects.

[Guillermo Luijk]

DR is often an issue for landscape photographers who are trying to capture natural scenes with a very high SBR. It's the reason why we have HDR programs.

I wouldn't forget that DR is more an issue on some synthetic scenes, such as interiors with artificial lighting or windows open to outdoor luminous areas. Both cases generate large luminance gaps.

In fact, unless the sun participates or is close to participate in the frame, landscapes aren't usually very demanding scenes in terms of DR.

BTW what is SBR?

Regards.

[Dick Roadnight]

Thanks for the tip. I will also look into Hasselblad back. As for the 'body' I'm considering one of these three: arca swiss m-line 2, cambo wrs 1000, horseman ld pro. I have also seen that sinar has a technical body but I read that they're going out of business?

Hasselblad more-or-less give away the "point-and shoot adaptors" with the MFD backs, (and they threw in the 50-110mm zoom in addition to a discount).

The "point-and shoot adaptors" have their uses for studio flash with models... you can even use them for some landscapes, on a tripod, with the mirror locked up!

[bjanes]

Higher DR on MFDB is a myth that serves well to MDFB sellers. DR depends on the overall quality of the sensor, not only on its size. Do some DR comparisons in DxO Mark and you will see the Nikon D3X's DR is higher than that found on any MDFB (Phase One, Hasselblad or Leaf).

Of course resolution, sharpness, being able to use great lenses,... is another story. But regarding DR, just look at the sensor your camera has.

Regards.

Many have criticized the DXO results, since they don't want to believe the results that contradict their own beliefs. Whether in politics or photography, this seems to be part of the human psyche. However, the DXO engineers have a thorough theoretical knowledge and use validated methods. They express dynamic range on a per pixel basis as would be seen when examining the image at 100% on the screen and on a normalized basis when the image would be viewed at a given overall size. They normalize to 8 MP, so when a 32 MP camera image is downsized to 8 MP, noise decreased when 4 pixels are averaged into 1.

Another statistic is tonal range, which has to do with the number of distinct levels that are present over the dynamic range of the camera. It is measured in the number of bits necessary to express that number of levels and good tonality will enable one to distinguish subtle differences in tonality. Jack Flesher mistakes dynamic range for tonality in his analysis of clouds.

Out of interest, I prepared a table showing some of these results:

[attachment=16704:DXO.GIF]

The DR of the Nikon D3x is better than that of the Phase 1, but the Phase 1 has a better tonal range. A better tonal range could be important in situations where fine gradations in luminance are important. The Nikon D3x and D3 have the same sensor size and the same number of photons would be falling on both sensors, but improvements in sensor design allow the D3x to best the D3 on both a per pixel and normalized (print) metric. The D3x could collect more photons via improved quantum efficiency or it could have better read noise or possibly both. The medium format backs have a much larger sensor area and should be able to collect many more photons, but this does not fully translate into the observed results.

How these differences in measurements correspond to subjective quality in the image are not clear to me, but the differences are not as great as the medium format people would like us to believe.

[BlasR]

Many have criticized the DXO results, since they don't want to believe the results that contradict their own beliefs. Whether in politics or photography, this seems to be part of the human psyche. However, the DXO engineers have a thorough theoretical knowledge and use validated methods. They express dynamic range on a per pixel basis as would be seen when examining the image at 100% on the screen and on a normalized basis when the image would be viewed at a given overall size. They normalize to 8 MP, so when a 32 MP camera image is downsized to 8 MP, noise decreased when 4 pixels are averaged into 1.

Another statistic is tonal range, which has to do with the number of distinct levels that are present over the dynamic range of the camera. It is measured in the number of bits necessary to express that number of levels and good tonality will enable one to distinguish subtle differences in tonality. Jack Flesher mistakes dynamic range for tonality in his analysis of clouds.

Out of interest, I prepared a table showing some of these results:

[attachment=16704:DXO.GIF]

The DR of the Nikon D3x is better than that of the Phase 1, but the Phase 1 has a better tonal range. A better tonal range could be important in situations where fine gradations in luminance are important. The Nikon D3x and D3 have the same sensor size and the same number of photons would be falling on both sensors, but improvements in sensor design allow the D3x to best the D3 on both a per pixel and normalized (print) metric. The D3x could collect more photons via improved quantum efficiency or it could have better read noise or possibly both. The medium format backs have a much larger sensor area and should be able to collect many more photons, but this does not fully translate into the observed results.

How these differences in measurements correspond to subjective quality in the image are not clear to me, but the differences are not as great as the medium format people would like us to believe.

 

Are you a science? wow, you good

I love you

I will sale my h 50 and I will get 300 dollars camera. (I can't wait to do that)

They look much better, they are better.

No only the I love you, I'm in love with you.(are you in San Francisco)?

Now I know why Jack, say it, sir, yes, sir.

Happy saturday

Blas

[Ray]

I wouldn't forget that DR is more an issue on some synthetic scenes, such as interiors with artificial lighting or windows open to outdoor luminous areas. Both cases generate large luminance gaps.

In fact, unless the sun participates or is close to participate in the frame, landscapes aren't usually very demanding scenes in terms of DR.

BTW what is SBR?

Regards.

Hi Guillermo,

I agree. A typical DR challenge would be to photograph an interior whilst simultaneously capturing a view out of the window without blowing the highlights. We should bear in mind that the eye adjusts its aperture very quickly as we shift our gaze from a bright area of a scene to a darker area, and vice versa, but we expect a camera to capture the whole scene at one aperture and exposure.

SBR stands for Subject Brightness Range which seems to be the terminology used to describe the range of EVs or stops in a scene.