Quick Dynamic range comparison - Leaf AFi-ii 12, CFii-39MS, Phase P20, Canon 5D2

[pschefz]

first: obviously the mf backs have more resolution and dr then the 5d...there is no question...can i get the same color from them all...probably...
i really don't believe in such a thing as correct color....i see what i like and hopefully my clients do too....
there is a color, texture (or lack of) combination that works out really well with the canons....i know from experience that there is nothing worse then shooting a celebrity and the files you are showing (back of the camera, screen, ipad,...) don't look flattering...
i hate the 3:2 ratio....always have, always will.....i really hope that the next generation of DSLRs finally break from the 35mm mold...there is no reason to stick with it (other then lenses) but there are other formats that could work with existing lenses....
regardless, the next model is always better and i am pretty sure we will see something in the 30-40mpix range (i am hoping more for 30) with DR better then the d3x across manufacturers....and with canon hitting the under 3000 pricepoint that hard with the 5dII, it has to be around that...unless you prefer sigma of course.....

[marcmccalmont]

Well the lens has an image circle so any aspect ratio could be accommodated?
Personally I like 3x2 for landscape orientation as it is closer to the angle of view that 2 eyes give you, more natural to me.
But... I prefer 4x5 for portrait orientation probably because of a lifetime viewing pictures in books, magazines and prints with this aspect ratio
Marc

[ejmartin]

last time i said that on this forum i was told different and given a technical explanation as to why it isn't 
i still use 160/320/640 as to my eyes the images are less noisy and pattern noise in the blacks is greatly reduced.

ISO 160 on the 5D2 is obtained by using the hardware ISO 200, digitizing the file, and then multiplying all the raw values by 0.8 (similarly for 320/640 etc; for 125/250/500 etc, the next lowest ISO is used and the result multiplied by 1.25).  Therefore the results are indistinguishable from using ISO 200 and applying EC in the converter (since that again is accomplished by multiplying the raw values by the appropriate factor, just done in the conversion software rather than in the camera -- perhaps more accurately if the converter uses floating point rather than integer math which the camera uses).  There is an illusion that 160 has less noise because the camera meters for ISO 160 even though it shoots at ISO 200 (ie it is secretly doing 1/3 stop ETTR).

The 5D2 is about the worst DSLR to use for this sort of comparison.  The camera has horrible pattern noise which renders the deep shadows pretty much unusable, and substantially cuts down on the useful DR.  The issue turns out to be compounded by the color filters, the transform to a standard color space such as sRGB/Adobe/Prophoto amplifies this noise.  As people have said, cameras using the new Sony Exmor sensors (such as the K5, D3x and D7000) do much, much better in this regard.

[torger]

And I'm not familiar with Rawnalize software so can you explain why in the chart you included the green histogram is ahead of the red and blue - what is that showing? Differences in sensitivity or

Did not see this interesting question answered (sorry if I missed it). The color filters of the sensor is fixed, that is regardless of white balance settings (just a software setting) the color filters on the silicon remain the same. This means that when Red Green and Blue channels will saturate depends on the color of the incoming light. And then there's sensitivity, ideally the sensor would capture all incoming photons and RGB would be equally sensitive, but in practice they are not. Red is usually (always?) the least sensitive.

A trick that some perfectionists use is to put a coloured filter on the camera to even out the sensitivities and match it for the given lighting condition (that is usually reduce sensitivity of the green channel). That way you can in some circumstances gain ~1 stop of extra dynamic range (not 100% sure about the 1 stop, but I think it is around that). You need to correct the color cast in post-production of course.

[TMARK]

Eric,

I love 5D2 skin in C1 and in DPP.  I shoot a Macbeth chart, balance off of that.  Strong light gets a nice file.  The only camera I like better, at least out of the box, is the M8.

The 5D2 is remarkable.  I have lots of cameras, ds3, two 5D2s, two backs, RZs, Sinar 4x5 to 8x10, an Arca F line, too many Leica film cameras, an M8, and M9.  I don't shoot stills much these days, but when I do, and it has to be digital, I toss two 5D2s, a 35, two 50's, two 85s and a 135 into a bag, and get on with it.  No BS, I just get to making photographs and forget all the science, magicical thinking, consumer psychology and angst involved with cameras and just make photographs that inspire my client, and hopefully me as well.  While the 5d2s thmselves don't produce rapture and trancendence, it just works, all of the time, whenever asked, and you can use it for B roll as well.   Oh yeah, thy are $2200 bucks, slightly more than the Leica 28 ASPH.  All the talk about pattern noise and false ISO, read noise, blah blah blah is fine, but look at the photos coming from Libya, Afghanistan, Yemen.  Beautiful images, inspiring. Mainly shot with 5D2s. 

Carry on, I'm done.

[fredjeang]

I don't like the 5D2 as a camera, I mean by that the way it feels and balance in my hand is not "right" for me. I don't like it's plastic feeling either compared to the 1D (wich also fits much better and more naturally in hand for me).

But I have to say that TMARK is right IMO.

The incredible thing of this camera is that it just works. Skin tones are great also in the sense that they are good right-out-the-box wich means less hassle.

I can't think of a more 4x4 camera in the world.

I find though that the 1D gives a more neutral point wich makes it IMO a better tool for using it in heavy post-prod, but the image right-out-the-box from the 5d2 is IMO superior.

[bcooter]

Eric,

I love 5D2 skin in C1 and in DPP..........snip

Carry on, I'm done.

+1

I don't care about numbers, or get flipped out over shadow noise.  I mean every image goes through post and it takes about 5 minutes in photoshop to selectively add noise reduction in different tonal areas.  We do this in medium format and the smaller formats, always have.

I'm with T-mark, just throw it in a bag and use it.  Good gosh, we carry thousands of pounds of equipment and can use anything we want, but the Canons work, you don't think about them, you don't beat yourself up trying to make them work.

This image (shot with a 5d2 and one kobald bron hmi) and 18 other images are going into an advertising show next week of 11x17 and larger prints.



I doubt seriously if anyone viewing these images which are of mixed formats, really care or notice if it's got 1/2 more stop of dr, or 10% more or less noise and the people viewing are buyers of photography which is my core market.

IMO

BC

P.S.  The single light fixture on this shot cost $9,000, the camera $2,000.

[torger]

I doubt seriously if anyone viewing these images which are of mixed formats, really care or notice if it's got 1/2 more stop of dr, or 10% more or less noise and the people viewing are buyers of photography which is my core market.

Dynamic range is more important in say landscape photography in natural high dynamic range light, so you in post-processing often need to compress the range to fit nicely in a print. In a studio condition with planned lighting there is wrong with your lighting or exposure if you need to raise shadows in post-processing.

When you don't need to raise the shadows 5Dmk2 works nicely. However, when you do need that, the line noise found in 5Dmk2 can be a very real problem. This is however not a generic DSLR problem, but a Canon problem which I truly hope they fix in their next model. Nikon D3x is currently the DSLR to compare with concerning technical image quality on 35mm. The only dynamic range advantage for medium format that should exist thanks to the format is the fact that the sensor area is 2.5 times larger, which means 2/3 stop less photon shot noise. The rest is due to differences in quality in the electronics, where many MFDBs seems to have a considerable advantage over say 5Dmk2.

It is commonly claimed that it is claimed that MFDBs are *several stops* better than DSLRs but I'm a bit skeptical about that. I'd like to see a comparison with properly expose-to-the-right D3x for example. To make it really tough competition you could boost the D3x with a color filter on the lens to balance the green channel :-). Eric's result of 3 stops between 5Dmk2 and the best MFDB seems very plausible though, due to the line noise problem.

Concerning fine differences in quality the buyers generally care much less than the experts, that is the photographers themselves. Picture content is always most important of course, but it does give an extra dimension of satisfaction if I know that also the technical quality is top notch. But people are different in this regard, and I guess it is probably a lot healthier not worrying so much about the technical quality :-).

[EricWHiss]

James,
These are good points, and agreed, its great that digital tech has gotten to the stage where we can put our energy into thinking about lighting or somewhere else entirely. For me that means I spend a lot less time trying to fix images in PS.     I can't really tell much about skin tone from your shot, not any more than the three images I've attached.  

One from each camera.....canon 5d2, p20, CF528c

[amsp]

All the talk about pattern noise and false ISO, read noise, blah blah blah is fine, but look at the photos coming from Libya, Afghanistan, Yemen.  Beautiful images, inspiring. Mainly shot with 5D2s. 

Amen to that.