The Character of the Digital Image

[lawrence]

I sometimes take a look at the dpreview.com site and notice that there's a never ending quest for more megapixels but personally I just don't need them, the 12Mp-16Mp range being adequate for the size I print.

The question I have is that, post processing aside, is there anything much that separates one digital camera image from another? I mean, are there any digital cameras that produce images that can be immediately distinguished from those from other digital cameras when printed at a size that is appropriate to the camera's resolution? I'm not trying to start some kind of film v digital debate, I'd really like to know.

 

[Dogman]

About the only thing I've seen that separates one digital image from another is the photographer involved.

Most cameras are excellent. Most photographers using them are mediocre at best.

 

[Ste_S]

The manufacturer's jpg settings can give different outputs when comparing jpgs. If you're shooting RAW though it doesn't make a difference.
ISO performance in low light is also a thing, with the Canon sensors usually regarded as inferior to Nikon/Sony in that regard.

 

[Godfrey]

The question is very broad ... A lot of factors can come into play. Some first order considerations:

• Sensor type and configuration
• Format and Pixel density
• Lens qualities
• JPEG or raw capture
• Comparison criteria

I list sensor type because there are those who swear they can see a difference between photos made with a CMOS sensor as opposed to a CCD sensor. Personally, I've not seen much that supports this distinction, although I will say that my Olympus E-1 (Kodak CCD) has a unique look and feel to its images compared to other FourThirds and Micro-FourThirds cameras (CMOS sensors) that take the same lenses. I attribute most of that difference to the fact that the E-1 has a much stronger antialiasing filter than any of the others. So, to me, sensor configuration is a far bigger factor than sensor type.

The size of the format is critical because it interacts with both the focal lengths used and the pixel density, together giving the pixel resolution achieved, and these factors influence sensitivity and noise. For example, my Leica CL has a 16x24mm (APS-C) format with 24Mpixel resolution. A camera I've wanted for some time is the Hasselblad X1D, which has a 33x44mm sensor with 50Mpixel resolution. That means that the X1D has four times the area but only twice the number of pixels ... each photo site is larger, allowing for less noise, more sensitivity, and more differentiable tonal qualities (presuming equal technologies in sensor, A->D conversion, etc etc etc).

Of course, just like with any film cameras, different lenses—even of the same brand and model!—produce different results because lenses are individuals. What lenses you use and how you use them is one of, if not the strongest factor in how an image looks from the capture side of the equation. All 50mm lenses are not the same... 🙂

Then we come to the far more difficult issues of how the image is rendered. Are you looking to see whether there are distinctions built into the in-camera processing system that renders out RGB JPEG images? Or are you considering what comes out of the camera as a raw file to be processed in a huge diversity of raw conversion and rendering by image processing software? This is such a huge area of distinctions and variations it is very very difficult to pick out distinctions that say "I can get this result with this camera and NOT with that camera."

And lastly, looking at all of the above, how do you specify what to compare when trying to decide what constitutes a particular camera's "look and feel" compared to another's?

As I said, the question is very very broad and difficult to answer succinctly. My take on it, after the many years and hundreds of thousands of photos I've made with digital capture tools, is that I concentrate first on what lenses any given camera has/can work with, and after that on format+pixel density, and after that on what software I use to render raw files into finished photographs/my skills in using same.

I do see specific character to various digitally captured photographs across my archive of thousands of photographs, most of it reflecting the lenses and my skills/notions of rendering at the time rather than specific camera attributes.

G

 

[lawrence]

I do see specific character to various digitally captured photographs across my archive of thousands of photographs, most of it reflecting the lenses and my skills/notions of rendering at the time rather than specific camera attributes.
G

That's pretty much my own finding, beyond a certain point the output of one digicam is visually pretty much the same as that of any another. Sure, more sensitivity and higher resolution can be good in some circumstances but generally they all look very similar.

 

[Russell W. Barnes]

I suppose your question must only consider shooting in RAW - and when is RAW actually RAW? Doesn't the camera image engine have some say-so on this as well?

In other modes the camera engine manufacturer will put their own stamp on the algorithms and PP software used to define colours. I think this was more noticeable, say, fifteen years ago, mind.

My little Canon Ixus 750 has a CCD sensor and the images look smoother* than on a contemprary camera with a CMOS sensor. But I couldn't look at both images blind and say definitely that one was shot on CCD, the other on CMOS.

* Shooting in different light makes a difference between sensors too. I found CCD 'warmer'. But I'm talking JPG here, not RAW.

 

[Alentejo]

Interesting question, and I think (we all can) I will answer it for myself asap.
I will take my Canon M6 with 22mm prime and my Fujifilm X100f 23mm prime and take the same picture from a tripot.
Will there be a noticable difference?

 

[joe bosak]

If you take post processing and resolution out of the equation, and seemingly sensor size size too, and assume sensor tech and internal camera tech or filters makes no difference (whether ccd vs cmos, rgb vs whatver else, bayer vs xtrans vs foveon, and any improvements over time - so all that talk of improved dynamic range or iso sensitivity also doesnt matter; also ignore jpeg engines), take lenses out of the picture (which effectively neuters any mention of compacts) then you are getting close to saying that all pixels are equal, and accordingly that one recording of an incomplete range of light wavelengths as a pattern of dots probably looks like another. Er, yeah. All photons are equal!

 

[lawrence]

If you take post processing and resolution out of the equation, and seemingly sensor size size too, and assume sensor tech and internal camera tech or filters makes no difference (whether ccd vs cmos, rgb vs whatver else, bayer vs xtrans vs foveon, and any improvements over time - so all that talk of improved dynamic range or iso sensitivity also doesnt matter; also ignore jpeg engines), take lenses out of the picture (which effectively neuters any mention of compacts) then you are getting close to saying that all pixels are equal, and accordingly that one recording of an incomplete range of light wavelengths as a pattern of dots probably looks like another. Er, yeah. All photons are equal!

That is not what I was asking, rather whether types of sensor etc. make a discernible difference or not.

 

[olifaunt]

Foveon sensors seem to have a somewhat unique and very pleasant color and tone rendering, from what I have seen. I assume those images are obtained with certain default camera or bundled software settings. (I have also seen a lot of post-processed Foveon images that are not to my taste, so YMMV.)

A lot of people swear by CCD sensors especially for black and white images. For example, the Ricoh GR Digital (sometimes called the GRD1) has a cult following for its camera jpeg monochrome rendering (it’s color isn’t great from what I’ve seen). Later GRD models are supposedly less magical. This is a camera that people use specifically for its camera jpegs monochrome mode, so it is hard to say whether it is really the CCD or simply the in-camera jpeg rendering algorithm.

Fuji has spent a lot of effort making camera jpegs look good. The Fuji X100 series, for example, has film simulation modes that are very popular and it is difficult to reproduce these looks by post-processing raw files from other cameras. Possible in principle to reverse engineer these looks, but life is too short.

Fuji X-trans sensors are controversial. Some consider them superior, others inferior. But they do make a difference to the image at pixel peeping level.

 

[Russell W. Barnes]

Modern digital cameras seem to have greater IR filtering. I used to use an older digital camera for testing TV remote handsets but modern digital cameras don't show up the IR illumination so readily.

There must be a reason for increased filtering holding back part of the non-visible light spectrum on the final image but I'm not sure what it will be.

 

[joe bosak]

The Character of the Digital Image

That is not what I was asking, rather whether types of sensor etc. make a discernible difference or not.

Ok, sure, my apologies. Hope this is a bit more helpful 😉

The foveon sensors are structured differently and render differently - they are the only digital cameras whose outputs make me go "wow", for what that's worth. This seems to be because they record each colour at each pixel (layered sensor) so there is less interpolation ... for the merrills, whereas the quattros do have some interpolation so i think are regarded as a little bit more like bayers in rendition.

The bayer and xtrans seem to be variations on the same thing as each other, same underlying sensor technology but different colour filter arrays and algorithms.

Opinions seem to vary about ccd and whether it's a source of fundamental imaging differemces or whether those are due to the associated processing, but since you can't disentangle the two in practical terms, not sure that matters.

The digilux 1 and lumix lc5 have some different kind of filtering which apparently gives them more sensitivity than corresponding rgb filters, but I'm not sure that makes a difference in practice. The low resolution and old sensor technology is a bit overwhelming from today's perspective.

( ... and yes there were improvements in IR filtering, because the underlying sensors would otherwise record IR hitting them as well, which could result in the image not coinciding with what human vision sees. Early digital cameras were quite good for IR work due to relatively poor IR filtering. The aforementioned Digilux 1/LC5 for example are not bad.)

 

[Contarama]

The old Fuji s3 and s5 do highlight roll off really really good...more like film. Perhaps still better at that than anything out there. However snaps from those cameras look very "digital". Very unique sensor.

 

[Ted Striker]

The old Fuji s3 and s5 do highlight roll off really really good...more like film. Perhaps still better at that than anything out there. However snaps from those cameras look very "digital". Very unique sensor.

I was going to chime in with the same answer. Images from the Fuji S3 & S5 Pro cameras have a very unique look. Nothing else like it. That was one amazing sensor. One wonders what it would be like today with more development.

 

[J enea]

..The foveon sensors are structured differently and render differently - they are the only digital cameras whose outputs make me go "wow", for what that's worth. This seems to be because they record each colour at each pixel (layered sensor) so there is less interpolation ... for the merrills, whereas the quattros do have some interpolation so i think are regarded as a little bit more like bayers in rendition.


...

I agree with this statement. I shoot 99% film. but I used a friends merrill DP1 once for fun. it had a different look to the images, almost a film like with some other wow factor to it (dont know how else to describe it). yeh the camera itself is not the greatest. reminds me a lot of my contax G1. the camera has its quirks and its not a great camera for those who dont want to learn to use it with its quirks, but the results speak for themselves. The Merrill is the same way. The Fovean sensor I find does in fact give me different rendering of the files compared to my older nikon camera. it very much is a light box with a great lens attached to it. will it make me shoot less film? actually, yes it will. instead of shooting color 35m film now i can take this camera on scouting trips and still shoot a 35mm B&W camera. I did hear that the merrill does have great B&W potential, but haven't tried it out yet.


I did buy a DP1 merrill a month ago and couldn't be happier. now thinking hard about a DP2 also



john