Crop Factor - Is It Really Giving Better Reach?

[Ed Cordes]

A few years ago a 1.6 or 1.3 crop factor sensor was touted to give a "focal length multiplier" so the angle of view of a given lens was the equivalent of a longer focal length lens and, therefore, is better for long lens use. While I understand the concept, I wonder if, in today's world, this is still true?

My thinking is as follows - a 1.3 crop sensor will appear to make a 500 mm lens a 650 mm equivalent. However, if you crop a 21 MP FF 5616 X 3744 image (5D2) in PS to the equivalent image size of a 1.3 crop 3888 x 2592 image (1DMK3) I think there are more pixels on the subject with the 5D2 or 1DsMK3.

So, while the subject may appear to be bigger in the view finder or when first opened in PS, the detail of an image with the subject occupying the equivalent % of the framing shown is actually better in the FF sensor. Therefore the only benefit of a crop camera today is less cost and faster frame rate along with smaller files for our computers to handle.

So this means that crop factor sensors are a dying breed as they are being eclipsed by the newer FF sensors even when considering long lens reach.

Do I have it right or have I completely missed the boat??

[Axel Hildebrandt]

I think you are right at the moment, but then who knows the amount of MP of the 1D4.

[Ed Cordes]

I think you are right at the moment, but then who knows the amount of MP of the 1D4.

Thanks for responding Axel. My thinking then would say that Canon and Nikon and the rest could save production $$ on their pro level bodies and only produce pro bodies as FF models. They could concentrate on better pixels, save money in production and marketing of crop cameras which will reduce the cost of the FF bodies.

Of course, the prosumer and consumer bodies in crop models will offer a lower cost alternative for those not needing a pro level body.

[Axel Hildebrandt]

In theory you are probably right but I think the price and lower frame rate would not be easy to overcome.

[Alfred Forns]

Ed just imagine the higher MP with a crop factor camera :)

[Ed Cordes]

Ed just imagine the higher MP with a crop factor camera :)

Yes, as long as the noise is controlled and the color depth is maintained.

[Don Lacy]

A full frame sensor would have to be around 39 mega pixels to match the 50D.

[Paul Granone]

Don,

Yes, the 50D does put the most pixels on the bird. The detail on the 50D is impressive.

I have had good luck reducing noise in 50D images with Neat Image v6.

[Jim Neiger]

Not all pixels are created equal. 14 bit pixels are better than 12 bit pixels for example. The image quality of the 1D3 with 14 bit pixel depth and the extra reach of the 1.3 crop factor is a combination that is difficult to beat. Add in the 10 per second frame rate and 30 image buffer size and you can see why the 1D3 is a great choice for action photography such as BIF.

[Ed Cordes]

Not all pixels are created equal. 14 bit pixels are better than 12 bit pixels for example. The image quality of the 1D3 with 14 bit pixel depth and the extra reach of the 1.3 crop factor is a combination that is difficult to beat. Add in the 10 per second frame rate and 30 image buffer size and you can see why the 1D3 is a great choice for action photography such as BIF.

Thanks for the response Jim. I understand pixel quality makes a difference along with pitch and gaps etc. I have been trying to think about the image size on a sensor and the idea that a 1.3 crop sensor gives longer reach than a FF sensor. For instance if an elk's head is the subject and is shot with a 500 mm lens and the 500 mm lens produces a 20 mm elk head in-focus image at the distance of the focal plane it should be the same on a FF or 1.3 crop. The difference should only be the amount of image around the head depending on the angle of view recorded. On a FF sensor, with a larger angle of view, more additional imagery is recorded than on the smaller sensor with a smaller apparent angle of view. However, the elk head is still the same size; the angle of view of the lens did not change. The angle of view recorded changed.

With this in mind, is there really additional reach with a smaller sensor?

Now, if we assume each sensor has the same pixel density (pixels per mm squared) will a print set to 300 ppi look any different if the printed image of the elk's head were set to a constant size - say 6 inches on an 8 X 10??

I am probably spending too much time thinking about all this, but that's the way my mind works sometimes - weird ;)

[Jonathan Michael Ashton]

Very interesting thoughts and observations. No doubt about it the 1D MK3 produces excellent results. So does the 20D, 30D 40D 50D, ( I have heard that Nikon also produce one or two decent cameras too!!:D)I think we can get rather hooked up on technical data striving for that next bit of marketing hype. I know there is a difference between moving up a camera for quality but I often wonder just how often that quality is actually needed, or required let alone perceived.
It would be interesting to see a set of control images , all taken under the same circumstances. Then the first thing many would say is well you can't tell on an internet image. For many I suspect this may prove the point i.e. we can spend hundreds or thousands on camera bodies only to see little difference if any.
I know there will at some/several point(s) there will be a difference(s) in printed images, now this will reflect equipment and processing skills. I suspect in most cases the apparent differences will be due to the latter.
I am not anti progress, quite the reverse I have recently bought a new 50D, I really do not expect it to perform to the same standard as a 1D at ISO 800 but I have to say it seems pretty impressive so far........ but then so is my 20D and even my first digital 300D. I think the quality of image is going to be very similar in many working circumstances, it is only when we begin to stretch horizons in terms of physical conditions that these differences will become more apparent.
I fully expect to be torn to shreds here - there are many people with a much better knowledge and skill base than I. All I can say is just look at the superb range of images on BPN....... and look at the range of cameras.
I think it is good to buy that which one perceives to be the best - I have no issue at all with this but I do wonder sometimes if it really is necessary.

Jon

[John Chardine]

Not all pixels are created equal. 14 bit pixels are better than 12 bit pixels for example. The image quality of the 1D3 with 14 bit pixel depth and the extra reach of the 1.3 crop factor is a combination that is difficult to beat. Add in the 10 per second frame rate and 30 image buffer size and you can see why the 1D3 is a great choice for action photography such as BIF.

As I understand it, a very important factor is sensor-site (pixel) size. Big pixels like those generally found on full-frame cameras are like big buckets that catch more rain per unit time than small buckets. In our case "rain" is light, and light is the signal we are recording on the sensor. More light captured per unit time means you do not have to turn up the sensor amplifier so much to obtain a particular ISO equivalent, thus noise is proportionately much less in relation to the signal. I noticed this very clearly (night and day as they say) when I recently acquired a Canon 5D, 12.8 mp full-frame body. The pixels are over 8 microns across on this sensor, compared to half that size on my 50D. The noise difference is very noticeable, in fact Canon decided not to even offer high-ISO NR built-in to the 5D. It would be surplus to needs.

So all this is to say that rather than comparing megapixels or sensor size, use the pixel size to make choices about camera bodies. Table 2 of BPN member Roger Clark's excellent web site on the subject gives pixel spacing which is equivalent to size in modern sensors. Here are a few from well-known modern camera bodies in order from biggest to smallest (1 micron = 1 millionth of a metre):

Nikon D3/700: 8.5 microns (µ)
Canon 5D/1D mkII: 8.2 µ
Canon 1Ds mkII/1D mkIII: 7.2 µ
Leica M8: 6.9 µ
Canon 5D mkII: 6.4 µ
Nikon D3x: 5.9 µ
Canon 40D: 5.7 µ
Nikon D300: 5.5 µ
Canon 50D: 4.7 µ
Olympus E3: 4.7 µ
"point-and-shoots" are smaller still

[Rick Baumhauer]

I think we're starting to move away from seeing the crop factor as the benefit, but rather the pixel density. With the 5D Mark II, there is now a relatively-affordable full frame camera with the same pixel density as the 20D/30D. While the 40D and 50D up the density further, it's not as much as you'd think, given the difference in megapixels. The 50D has, measured linearly, about 35% higher pixel density than the 5D2. A 5D2 owner could, theoretically, add a 1.4xTC and put just as many pixels on the subject, while having much more freedom with cropping (the same would apply to the 1DS3, of course).

Granted, there are other benefits to the 50D (frame rate, primarily, though it's also a bit smaller), but the 5D2 also has HD video and better high ISO performance. For the moment, I think I'm going to stick with the 5D2 for everything, but I'm open to being convinced by a theoretical 60D at some point in the future :)

[John Chardine]

I look at it the other way around Rick- you pay a cost for modern cropped sensors in terms of image quality and there are few benefits to offset. One benefit is the relatively low cost of a cropped body compared to FF but that's about it. The number of sensor sites that are under the image projected on the sensor is trumped, within limits always, by the size and therefore quality of the sensor sites. I have learned this first hand when comparing images I make from my 50D compared to my 5D. The 5D is equivalent to about a 5mp camera if it had a 1.6 crop sensor yet the images it produces are equal or better than the 50D at 15mp!

Have a look at a thread I started a while back. You may find it interesting:

http://www.birdphotographers.net/for...ad.php?t=27083

[Rick Baumhauer]

John - I certainly understand where you're coming from, but there is a balance in there somewhere between the quality of each pixel and having enough to capture fine detail. Personally, I wouldn't want to go with a lower density than the 20D/30D/1D3/1DS3/5D2 for birding, but I'm not sure that the higher density sensors (40D/50D/450D) make enough of a difference to worry about.

[John Chardine]

I agree Rick, there is a balance between number of pixels to render the subject's fine detail, and pixel quality. It does seem though that modern lower pixel density sensors are up to the task and I don't think they limit ability to render fine detail under normal bird photography circumstances. Many fine bird images have been, and are being, made with Nikon's D700 and D3, and even Canon's older 5D, all of which have a pixel density of around 1.4-1.5 mp/square-cm. That's equivalent to about 120 pixels per mm. The Canon 20D by comparison has about double the pixel density of these yet I don't think you would see any improvement to render fine detail on a bird with this camera.

Ultimately, it seems that all the high-quality camera's out there are capable of making great images so I suppose all of this is a bit academic really!

[Roger Clark]

John - I certainly understand where you're coming from, but there is a balance in there somewhere between the quality of each pixel and having enough to capture fine detail. Personally, I wouldn't want to go with a lower density than the 20D/30D/1D3/1DS3/5D2 for birding, but I'm not sure that the higher density sensors (40D/50D/450D) make enough of a difference to worry about.

Rick,
I have a model that describes these trades. See Figure 9 at:
http://www.clarkvision.com/imagedeta...index.html#AIQ

The model shows that for diffraction limited f/8 lenses, the peak performance would be sensors with a pixel spacing of about 5 microns, right where many DSLRs are these days.

Roger

[Jonathan Michael Ashton]

Rick,
I have a model that describes these trades. See Figure 9 at:
http://www.clarkvision.com/imagedeta...index.html#AIQ

The model shows that for diffraction limited f/8 lenses, the peak performance would be sensors with a pixel spacing of about 5 microns, right where many DSLRs are these days.

Roger

Roger, I have looked at the site. I have read and learned every word;), but just to put my mind at rest you reckon the 50D is OK, but from the data I have seen the "image quality" may be described as inferior to the 20D yet all the Canon hype seems to indicate to the contrary. I hope I have not been suckered! I am assuming that your charts are showing fine degrees if sensitivity, so the apparent image quality in the prosumer models would appear to be visually to be similar(?).
The one that stands out is the 1D Mk3, I have used it and it produced fine images but if I'm being honest and look at my images taken with the 1D Mk3 and similar ones taken with my 20D I would not know which was which without checking the data. I am not trying to imply the Mk3 is a waste of money or anything like that but from my usage point of view I don't see much difference in all the cameras I have used. I do sense the 50D is a perhaps a little noisy though and yet others seem to think it is on a par with the 40D.

Jon

[Andy Wai]

[...] 1D Mk3, I have used it and it produced fine images but if I'm being honest and look at my images taken with the 1D Mk3 and similar ones taken with my 20D I would not know which was which without checking the data. [...]

I have both 1D3 and 20D. The 1D3 is my main camera but I used my old 20D exclusively in a recent trip to China and am going through a huge pile of 20D files at the moment. At low ISO, the 20D is actually quite respectable. But at ISO 800, you begin to lose subtle details. By the time you get to ISO 1600, there is no comparison at all. So how much difference between the two depends on your exact shooting condition.

Andy

[Arthur Morris]

I currently have three MIII bodies and two 50D bodies. In low light/high ISO situations I always opt for the MIII. When photographing flight and action the 50D out-performs the pro body with its superb AI Servo AF accuracy. If you check the recent Bulletins, you will see lots of great 50D images. Is the crop factor real? You have better beleive it. With the 800 F/5.6L IS lens I have an effective 1280mm lens (without using a quality robbing TC) and get to use only the center of the lens elements, the sweetest and sharpest glass. And, I have 15 million pixels. So, no, crop factor cameras are not at all on the way out. All three of my MIII bodies are yellow dot versions. I am selling one of them.

[Roger Clark]

Roger, I have looked at the site. I have read and learned every word;), but just to put my mind at rest you reckon the 50D is OK, but from the data I have seen the "image quality" may be described as inferior to the 20D yet all the Canon hype seems to indicate to the contrary. I hope I have not been suckered! I am assuming that your charts are showing fine degrees if sensitivity, so the apparent image quality in the prosumer models would appear to be visually to be similar(?).

Jon, I think part of the issues with comparing images from different cameras is that people tend to zoom in and "pixel peep." But what should be done is to compare the same size images, e.g. prints. The 50D, for example, has noisier pixels, but there are more of them. Which produces a better 13x19-inch print, 50D or 20D?

For example, the AIQ for the 20D is 39, the 40D 45 and the 50D 53. I bet 50D images at 13x19 look sharper than 20D images (given that the lens was adequate). The big jump is from the APS-C to full frame sensors. For example, the 5D has an AIQ of 76 versus the 5DII at 109. So if one compares 5D pixels at 100% they are clean and pretty noise free, but in a print of the same size, the 5DII will blow away the 5D for detail (again assuming the lens delivers that detail). Same with 50D versus 20D.

The one that stands out is the 1D Mk3, I have used it and it produced fine images but if I'm being honest and look at my images taken with the 1D Mk3 and similar ones taken with my 20D I would not know which was which without checking the data. I am not trying to imply the Mk3 is a waste of money or anything like that but from my usage point of view I don't see much difference in all the cameras I have used. I do sense the 50D is a perhaps a little noisy though and yet others seem to think it is on a par with the 40D.

Yes, the AIQ of the 1D3 is 57 while the 20D is 39, and I would think the 1D3 images to be much cleaner, especially as ISO is increased. I know my 1D2 images are cleaner (less noisy) than my 30D.
Again, compare the images at the same size, like on prints.

Roger

[Jonathan Michael Ashton]

Thanks Roger Artie and Andy, I am in complete agreement with what has been said - and a big thanks to John, I have learned a lot as a result of this thread.

Jon

[Arthur Morris]

No disrepect intended as I respect folks with great technical knowledge: I have been doing this bird photograogy thing for more than 25 years now and have never hear of AIQ until now, yet I have made a few good images along the way. I do feel at times the some folks get so hung up with the technical details that they forget that the idea is to get out and make intersting, dramatic images.

I would imagine that the Q stands for "quotion." IAC, I would love to know what AIQ is. Thanks all for the level-headed discussion and advice above.

[Roger Clark]

I would imagine that the Q stands for "quotion." IAC, I would love to know what AIQ is. Thanks all for the level-headed discussion and advice above.

AIQ = Apparent Image Quality. It is a concept that has been around for decades in one form or another.

With the the digital camera era came new confusion and discussion over image quality. For example, with slow speed film you could drum scan at 8000 pixels per inch and produce 80+ megapixel images. People were confused why an 8-megapixel camera produced better images (for many people) than film. The answers are complex but have to do with true resolution and the signal-to-noise ratio (S/N) in the image as well as dynamic range and tonality and color accuracy. To first order we may perceive an image with higher S/N as better even if it has less actual resolution. It is a human perception problem with no clean answer and considerable variability between people.

I have a model for AIQ that I have been refining over the years. For digital sensors my model predicts with reasonable accuracy the relative image quality:
http://www.clarkvision.com/imagedeta...index.html#AIQ
Film AIQ values are at:
http://www.clarkvision.com/imagedeta....summary1.html

When a new camera comes out, I have been able to predict the basic performance of the sensor and the image quality of the resulting images. My model shows that current DSLRs are operating at near peak performance for the current sensor technology (which is set by quantum efficiency of 30 to 40%). Of course add to that AF and other performance metrics for a camera to get the overall camera quality.

I started this academic exercise to choose the best camera for astrophotography. Astro work pushes all limits for low light work, so choosing a camera can be critical to results.

Roger