Full Frame vs Cropped Sensor Color Depth Comparison

[payamghafoori]

Hi Everyone,

 

I have heard about Color Depth difference between full frame and cropped sensor DSLRs (please refer

tohttp://www.snapsort.com ). As I am a Canon user and planning to upgrade my gears, I’d be grateful if some of

you pals share her/his field*experience regarding the Color Depth Difference between say 5D mk II and 7D or some

comparisons like that. Actually I want to know whether such claimed Color Depth Difference could noticeably

influence my work (mostly portraits and fashion shots and almost always post processed in Photoshop).

 

Thank you in advance for your kind replies.

[danielleetaylor]

<p>* Snapsort is merely a front end to data fed from DxOMark.</p>

<p>* DxOMark is a joke.</p>

<p>* Whatever accurate testing (as opposed to DxOMark testing) might reveal about the differences in color output between high end crop and FF bodies, I can pretty much guarantee you they would be too subtle for anyone to see.</p>

<p>* Any such differences are also about 2-3 orders of magnitude smaller than the differences that result from camera/RAW converter settings, post processing, printers, and paper choices.</p>

<p>Buy based on something other than meaningless and inaccurate color depth tests.</p>

[ilkka_nissila]

<p><em>* Whatever accurate testing (as opposed to DxOMark testing) might reveal about the differences in color output between high end crop and FF bodies, I can pretty much guarantee you they would be too subtle for anyone to see.</em></p>

<p>Actually the differences are very obvious even in web size images. People have noticed this on Wednesday threads over the Nikon forum, FX images are richer in colour and tone. Having recently acquired a late-model DX camera (D7000) I can tell that it's extremely finicky with regards to exposure accuracy and the tonal depth is very much in favour of my 3-year old FX stuff (D3)... at least when shooting above ISO 100 (which the FX camera doesn't support). But of course there will always be people who are able to destroy the fine tonality in post-processing or don't have an eye for such things. (Although my comments are based on Nikon cameras, I have seen output from Canon cameras of my friends' (7D and 5DII) and they show exactly the same behaviour.)</p>

<p>I agree with you on the DXO lens tests, they're seriously lacking in perspective but their sensor tests are very good and illuminating.</p>

 

[g dan mitchell]

Of all the things that might have significance in a camera purchase decision this is among the most trivial, and is more

likekly completely insignificant.

[ilkka_nissila]

<p>First of all, which raw converter you use is irrelevant; you choose whichever raw converter you like. Differences in cameras are seen by comparing the information in the file, before raw conversion, and that's what DXO is trying to measure. Differences in information content are then transmitted through the raw conversion process and you can use it how you like. If you make a crappy conversion the image will be poor but if you use the same conversion for images from both cameras, the difference will transfer through. </p>

<p>Now, a lot of people online without any understanding of imaging technology and information science will say "okay,one camera records 10 bits and the other 11. What difference could that possibly make?" On the perceptible quality of an individual pixel, none. There will be sufficient noise that you can't tell the difference. However, if you take a larger area, such as the sky, the tonality of the sky will be perceived by averaging it spatially to some extent, and 1 bit difference in one pixel will suddenly translate to a million bits in the patch of sky (corresponding to 1MP of the image) that you're comparing to another patch of sky, and this way you actually see a huge difference in tonality. And that's what the difference between medium format film and 35mm film, and MF digital vs. FX vs. DX is all about, to someone who doesn't make huge prints. Even small prints show this. And the more post-processing you do on the file, the more the difference in tonality will be obvious.</p>

<p>Those extra photons captured by the full-frame sensor do not disappear anywhere, you know. They are the foundation of the information content of the image.</p>

[dom1]

<p>Payam, I own and use both cameras. A very simple answer is yes, the 5D produces better color. I find it most noticeable in flowers. But if you think you may need 8 frames/ second, the 7D is the way to go.</p>

[markonestudios]

<p>I agree with GDM on this one. Especially as you say you post process all the images later.</p>

[bobatkins]

<p>I don't think it's a factor. It might be measurable I suppose, but in practice most people probably won't see it, even if they look for it.</p>

<p>Proper exposure and skillful post exposure processing (including RAW conversion) will have most influence over the image.</p>

<p>Even if it is there, it 's not really a full frame vs. crop effect. It's a pixel size (or more accurately, signal to noise ratio) and bit-depth issue. It's also the case that most current DSLRs have very similar color depths.</p>

<p>So in short I'd base my decision on what camera to buy on other factors and leave color depth down the list only to be used as a tie breaker if everything else comes out exactly equal.</p>

 

[matthijs]

<p>To me the jump from 400D/XTi to 50D was probably bigger than from 50D to 5D-ii.</p>

<p>(In color depth that is. In other aspects there's a whole new thread... Let's just say the 50D is infinitely more snappy in handling than the 5D-ii but the latter creates subtly better looking pictures.)</p>

[robertjewett]

<p><img src="http://farm6.static.flickr.com/5288/5322800392_1e770c53c6_b.jpg" alt="" width="1024" height="683" /></p>

[robertjewett]

<p>All of these were only converted to sRGB using CS5 and a profile for each camera created with the x-Rite Passport (and software).<br /> No crops. All were done with the same lens and I moved in or out to compensate.<br /> When I posted this without the answers on multiple photography forums (in which people were arguing about this), very very few (2 I think) were able to guess which was which.</p>

[KenPapai]

<p><em>Of all the things that might have significance in a camera purchase decision this is among the most trivial, and is more likekly completely insignificant.</em></p>

<p>Agreed -- I'd LOVE one day for these hyper bit peepers to design CMOS sensors. It's like this: in 2011 we pretty much have *perfection* in DSLR tech. It will get better though -- can you imagine EOS DSLR tech in 2021? (none of us can) We really have it all and some want more! ;-)</p>

<p>Monday morning QBing... that's all this stuff is ... Over and out.</p>

<p><em>Proper exposure and skillful post exposure processing (including RAW conversion) will have most influence over the image.</em></p>

<p>100% true.</p>

<p><em>Proper exposure and skillful post exposure processing (including RAW conversion) will have most influence over the image.</em></p>

 

It's worth repeating.

 

[scott_ferris]

<p>Robert are you sure the 7D is post profile? It is far and away the boldest in the reds, the chair and the collar. My passport gets me different cameras, like the other three, much closer together.</p>

<p>It is a shame I noticed the camera labels while looking, I was enjoying trying to separate them :-).</p>

[danielleetaylor]

<p><em>Actually the differences are very obvious even in web size images.</em></p>

<p>Good to know! Then you should have <strong>no problem</strong> telling me which crop came from which camera, right?<br>

<a href="http://www.taylor-design.com/photos/color_test.jpg">http://www.taylor-design.com/photos/color_test.jpg</a></p>

<p>You couldn't tell us which crop came from which format, much less make and model. You couldn't even tell us if they all come from DSLRs or if there's a P&S in that list.</p>

<p><em>But of course there will always be people who are able to destroy the fine tonality in post-processing or don't have an eye for such things.</em></p>

<p>And there will always be people who interpret simple psychological biases and improper testing methodologies as evidence of their superiority.</p>

<p>If you want to know just what things you have an eye for when it comes to color, feel free to take this test and report your score:<br>

http://www.xrite.com/custom_page.aspx?PageID=77</p>

<p>I can repeatedly get a perfect score on that test and while I've seen subtle color differences between photo printers and even RAW converters, I've yet to see a difference between DSLRs that couldn't be attributed to settings and/or wiped away with minor processing or profile tweaks, assuming good exposure. Just about any change in a RAW converter or in PS would destroy such differences any way. <strong>Photographs with great color do not come from sensors, they come from the artists who make them. </strong>And by "make them" I don't necessarily mean PS. Light and subject are the first and most important choices when it comes to color.</p>

[scott_ferris]

<p>Got to be honest, from a colour point of view I haven't seen a difference when calibrated, with a simple and automatic X-Rite Passport profile, between crop and full frame digital cameras. I haven't done colour ramp test comparisons though, actually taking pitures gets in the way of that stuff!</p>

<p>I think this means I basically agree with Daniel on this one :-)</p>

[philip_wilson]

<p>I would not base my choice between the 5DII and 7D on a web site test of colour. I own both and they are great cameras but they have very different strengths and weaknesses. The 7D is a much better action shot body due to its better AF and fast fps. The 5DII is a better landscape and portrait body as it has better wide angle lens options, better high ISO performance and even at low ISO I find the images smoother than those from my 7D (although the 7D images are still very good)</p>

[drpath]

<p> I don't feel entitled to disagree with the several commenters who obviously have more technical knowledge of DSLR's than I -- especially Bob Atkins. Yet I have to ask: why then do my 5D images so often <em>look</em> as if they are have richer color than those of my 500D/T1i and the various other crop bodies I have used? The difference seems most apparent in scenes with high dynamic ranges. I always ascribed this to the (relatively) gigantic 8.4 micron pixels of the 5D, and their lower S/N ratio.</p>

<p> If I am wrong about this, what am I doing wrong with my T1i images? I shoot RAW with both bodies and process in DPP for best print or monitor appearance. At their best, 5D images look to me like images I shot with one of my MF cameras. T1i images are more like high quality 35mm. </p>

[scott_ferris]

<p>Dr Path,</p>

<p> The colour response characteristics of the camera that we see are a product of the sensor, the cameras processing and the interpretation the RAW converter you use put on that information. Different camera models look different in the same way that different film emulsions look, well, different. But if you adjust the processing you can get them to all look the same, with digital getting all cameras to look the same is very easy.</p>

<p>But this is only relevant to colour, I agree with you that crop camera images can "look" like 135 film images and good 135 digital images can "have the look of" medium format film, but that is very personal and the size of prints that any one person is happy with from whatever format is up to them. The "look" that you observe is more than colour it is a combination of the total of the images characteristics, sharpness, contrast etc and isn't the real subject of the thread.</p>

[DickArnold]

All I have to say being 78 yrs. old with poor vision (correctable to 20-20 in a 70-200 2.8L EF lens) that I do well on the x-rite test so that doesn't prove much in my mind. I am actually a pretty good printer. I think this whole argument equates to a discussion about how many angels can dance on the head of a pin. Who in the practical world can tell the difference? I do own an XTi and a 5D. If my vision were better I would say the color is a little more intense from the 5D but I am certainly no expert. To repeat something I have said earlier I have, in the past, managed to make money with Bronicas, at least eight Canon bodies, I have no idea how many lenses, and both full frame and crop Canon bodies. They have all done a better than an adequate job for me. I choose not to take sides in these sickeningly long arguments involving film vs. digital, crop vs. full frame, etc. I believe, if I were still in business, I could use any decent quality kit and make money with it and satisfy my level of customer. I have enlargements from all of the above media hanging in my house. No one I ask can identify from which of the above media each picture was made. There are even some good ones from an old 6MP D60 that have won awards. IMO we are approaching that region of the DSLR progress curve where it becomes increasingly difficult to see in an actual enlargement or image visible differences in so called improvements in technology. As technology matures the rate of change slows and becomes more costly and it occurs in smaller quantifiable increments. I just bought a G12 for a friend that is extremely good in comparison to earlier dslrs. Everything produced today, IMO, is unbelievably good. So was the post WWII Minolta rangefinder one that I owned over 50 years ago.

[lalon_karim]

<p>I think people are misinterprating the meaning of color depth/ color sensitivity that is measured in dxomark. From my limited understanding, color sensitivity is nothing but the opposite of chroma noise. More color sensitivity = less chroma noise, less color sensitivity = more chroma noise. It has no relationship with how pleasing the colors will look in output (until the chroma noise is noticable). A P&S camera can have much pleasing color than a full frame DSLR if the manufacturer of the P&S do more research on what colors look more pleasing to most people.</p>

[scott_ferris]

<p>Lalon,</p>

<p>I think we are all really thinking about colour sensitivity, not their definition of colour depth. But I don't think it as simple as "More color sensitivity = less chroma noise, less color sensitivity = more chroma noise." I think the point is how many colours are distinguishable in gradations equal to the amount of colour noise, so a less noisy sensor will have an advantage because the gradations are smaller. But the distance from the top to the bottom is a measure of depth. So a sensor with more depth but more noise might, or might not, beat a sensor with smaller depth but less noise.</p>

[lalon_karim]

<p>I think more color sensitivity = less color noise and less color sensitivity = more color noise. If a sensor is less color sensitive, it has to apply greater degree of color processing (through color matrix) to match the output of any standard color space (ie sRGB, Adobe RGB etc) which results in more color noise. See at dxomark, Canon sensors are less color sensitive than Nikons' and Canons trully have more chroma noise in my experience. Among the currently available Canon and Nikon DSLRs, Canon 500D has a very low score in color sensitivity and it truly has much chroma noise.</p>

[bobatkins]

<p>Color depth simply measures the number of individually discernable shades of color that can be recorded. The higher the noise (expecially chroma noise), the lower you'd expect the color depth to be.</p>

<p>Most DSLRs seem to be in the 22-24 bit range, which corresponds to from 4 million to 16 million different colors. There's a prize for anyone who can name the first million colors....</p>

<p>Whether 16 million is better than 4 million also depends on the color depth of your output device.</p>

<p>Color depth has nothing to do with color accuracy and nothing to do with which colors can be displayed (that's gamut and is limited by the color space you work in - typically RGB). In general it's not even how "rich" the colors look (which is more to do with saturation).</p>

<p>Is greater color depth better? Yes, in principle it is. Whether you can actually see it in images is another matter. Images may well look different and image "A" may look better than image "B", but you can't just assume that difference is due to color depth, especially not when the images are displayed on a monitor and probably not even when printed.</p>

<p>Color depth is also ISO sensitive. You have to shoot at the lowest ISO setting (typically ISO 100, not an expanded lower ISO) to maximize color depth. That correlates with noise of course.</p>

[ilkka_nissila]

<p><em>you can't just assume that difference is due to color depth, especially not when the images are displayed on a monitor and probably not even when printed.</em></p>

<p>If the raw conversion algorithm is identical apart from the normalization (calibration) of the colour response of the sensor (a procedure designed to match the output in colour and tone), the final RGB images can be compared between camera types without so much of an issue. Saturation and color depth won't be confused since the algorithm matches the former. Whether you can only see noise in general or specifically in colour is not obvious to the naked eye at this stage. However, a camera with better colour depth can be used to make better quality high saturation images; this is one way that one can take advantage of the additional information. If you use the flattest tone curve of course there will be no visual difference but the image will be hopelessly lifeless. Let's imagine an image shot of a person in a wintery scene with overcast sky, it's almost like having a subject in an integrating sphere, very little contrast between highlight and shadow. Now, to bring life to the subject you need to increase contrast and saturation quite a bit. After you do this the tonal quality differences (whether they are caused by noise in luminosity or colour) will be very obvious even at intermediate (i.e. 800) ISO. I'm not saying that I can tell by naked eye if breaking of the colours in this case is more severe than the breaking of luminosity ... that's not the central point. As the user you can alter the trade-off between luminosity and colour noise by varying the parameters of the raw conversion / noise reduction algorithm. If you want to characterize a camera, however, both types of noise need to be measured. It's ridiculous to claim that the difference in tonal quality between formats isn't of real world significance.</p>

<p>The monitor quality ... if you use a monitor designed for graphics art work with side light shields and if it is designed not to be too sensitive to viewing angle it is probably good enough.</p>

<p><em>You have to shoot at the lowest ISO setting (typically ISO 100, not an expanded lower ISO) to maximize color depth.</em></p>

<p>Of course, but in real-world situations you often have to use something other than base ISO to capture the image.</p>

 

[ilkka_nissila]

<p>There is one factor that may explain the visual difference in some cases, and that's the fact that you can't use the same lens (at the same focal length) to make the same picture on two different formats. If you use the same lens at a different zoom setting you may get different flare behavior which can reduce the contrast of the image taken with the smaller sensor. And to recover the contrast in a way you need to apply a steeper contrast curve and that increases noise. This could be a part of why I see a considerable overall difference in the editability of images from the two formats, but the signal-to-noise obviously plays a part as well. Again, it's not about giving a distinct name to millions of colours but how much you can warp the file until you see breakup of tonal continuity.</p>