Crop Factor VS Full Frame dSLR - when, why and how...

[Lin Evans]

Crop Factor Sensor VS Full Frame (36mm x24mm) so called 35mm Sensor

This is just a brief explanation of a number of issues and questions facing the digital photographer today. I will include an easy “formula” for determining the actual number of pixels painting a subject when a so called “Full Frame” sensor image is cropped to the field of view (FOV) of various so called “Crop Factor” sensors. Also I will discuss advantages and disadvantages of each in a simplified fashion.

Today, there are a number of different “sensors” for digital cameras ranging from quite expensive professional model dSLR (digital single lens reflex) models to much less expensive consumer models. What I want to concentrate on are primarily the dSLR models rather than the mirrorless and consumer cameras. The reason for this explanation is that there appears to be a lot of confusion about which, if either, is “best,” and why some cost so much more than others…

A so called full frame dSLR has a sensor approximately the size of 35mm film which is 36mm in width and 24mm in height. This works out to an aspect ratio of 3:2 and is found in the most expensive dSLR models made by the major camera manufacturers.

Below are some typical sensor sizes:

36x24 mm (approximate) full frame sensors are found variously in dSLR models by Canon, Nikon, Pentax and Sony.

Canon: 1X (FF), 1.3X and 1.6X

Nikon: 1X (FF), 1.5X

Pentax: 1X (FF), 1.5X

Sony: 1X (FF), 1.5X

Sigma 1.7X, 1.5X

Olympus 2X

Above are typical dSLR’s and their various iterations of sensors. There are also 2.7X sensors such as found in the Nikon 1 series and in a few other major manufacturer models and of course a host of 5x, 6x, etc., which are found in consumer model digital cameras.

So what does this all actually mean? There are two main aspects to digital sensors which are commonly used in advertising. First there is the term “megapixel.” For the purposes of this explanation, we will assume that a megapixel is one million pixels. So a camera which is called a ten megapixel camera has ten million pixels used to capture the image while a camera which has thirty six megapixels uses thirty six million pixels to capture an image. The actual size of the sensor pixel which is the tiny “well” which gathers light on the silicon determines what is called pixel pitch. For example, a tiny sensor the size of perhaps your smallest fingernail, might have eighteen million pixels packed onto this tiny space. This would represent a very dense “pixel pitch” while a full frame sensor on say a Canon camera could also have eighteen million pixels but would have comparatively much larger pixels. So what is the difference between larger and smaller pixels?

Larger pixels gather more light. Having more light means, generally, a better “signal to noise” ratio which in practical terms means better low light performance allowing greater amplification of the signal without a corresponding huge increase in noise. So a full frame dSLR with an eighteen megapixel sensor will generally allow the photographer to shoot in lower light conditions using higher ISO numbers and still get superior images which can be enlarged a great deal more than the same image, for example as from a tiny sensor in a consumer level digital camera.

So we come to the first advantage of the larger sensor. Better low light performance with correspondingly less noise in the image. So what then is the advantage of the smaller sensor such as the so called “crop factor” sensors in the dSLR lines, if any? Primarily what is referred to as “telephoto boost.” A 100mm lens attached to a full frame dSLR gives a 100mm field of view. But when this same lens is attached to say a 1.6x crop factor Canon dSLR, it gives a 160mm field of view which would encompass the same geography as a 160mm lens mounted on a FF camera and shot from the identical position at the identical subject. The actual focal length has not changed, it’s still 100mm and the true “magnification” has not changed, but the subject appears much closer because a large proportion of the field of view as would be seen from the full frame camera has been lost to the crop. Because the sensor itself is smaller, the light falling on this crop frame sensor from the circle of light gathered by the lens does not included much of the geography which would be seen had there been a 36x24 mm sensor rather than the perhaps 22.3 x 14.9 mm 1.6x Canon crop sensor. Assuming the same number of pixels, but a greater pixel density (more pixels squeezed into a smaller space), there will be the same number of pixel painting a smaller subject area. But because the pixels are smaller and therefore gather less light, there will be more noise and correspondingly less image quality especially when shooting in low light conditions.

Advances in sensor technology and electronic processing have closed the gap a great deal between the actual quality of image which can be captured by the full frame sensor versus the dSLR crop factor sensor. There are a much smaller differences today than in the past and because of the telephoto “boost” many photographers who shoot distant or small subjects (wildlife, birds, etc.) often use one of the crop factor sensors rather than a full frame sensor. On the other hand, landscape photographers and portrait studios often favor the full frame sensor because of the admittedly now smaller advantages in image quality as well as the correspondingly better ability to control depth of field, and better wide angle performance of the full frame sensor.

Before getting into the math on how to determine pixel counts when cropping full frame images to crop factor fields of view, let me briefly talk about advantages to some wildlife photographers when using crop factor sensors with dSLR’s.

I am primarily a wildlife photographer who also shoots landscapes. I work in very rugged terrain at high altitudes and often spend a number of days in the back country shooting at altitudes above 13,000 feet and sometimes above 14,000 feet. I must carry my camera equipment plus my survival and camping equipment all on my back while climbing over boulders, traversing dangerous forty five degree angle slippery scree fields where one misstep could mean death or serious injury. Sometimes I will walk ten to fourteen miles in a day of shooting then camp and begin again before sunup and I may do this for a week or so. This is extremely hard physically and the older I get, the more difficult it has become. When I was a young man and very strong, I thought nothing of hiking twenty five or even thirty miles per day, but even then, because of the altitude and difficult terrain, every pound of equipment I had to carry was very important. Let’s put this into photographic perspective.

A good full frame dSLR may weigh around three pounds. To get the images I want, I need at least 800mm focal length. A Sigma 300-800mm F5.6 lens weighs more than twelve pounds. To use this lens it must be mounted on a sturdy tripod with a relatively large and heavy tripod head. All together this combined weight (tripod, head, lens, camera) is around 23 pounds as a conservative estimate and it can’t be used effectively without the tripod. To complicate things, often I have less than five seconds to raise my camera and press the shutter before my elusive prey disappears so I must generally work hand-held. My “go to” camera these days is a 24 megapixel Nikon D7200 and if I couple this with a Sigma 150-600 mm stabilized lens, I have about an eight pound combination which gets me an equivalent 900mm hand held focal length. I shave about fifteen pounds off the weight I must carry plus a huge bulk difference and gain the ability to hand hold and quickly raise my camera and get the frame before my subject disappears. In addition, I get my full 24 megapixels vested in the 1.5x cropped image. Just for a few seconds, let’s calculate what I would get in terms of the number of actual pixels on the subject should I use the same lens and say a full frame 36 megapixel Nikon D810 camera and a 1.4x telephoto converter which gives me 840mm.

I’m going to keep it simple with the math and let the reader see how the calculations are done; So we take the full frame size of 36x24 mm as a ratio to 36 megapixels. 36 time 24 is 864 so we have 864/36. Now when we crop this to the 1.5X we have to take the size of my 1.5x crop sensor as a ratio to “X” to determine the number of actual pixels we would have on the subject. This is 23.5 x 15.6 mm which when multiplied equals 366.6. So solving the ratio and proportion we have 864/36 is as 366/x. Solving for “X” we get 864X = (366.6x36) or 13197.6/864 = 15.275 megapixels painting the subject. So, using the Sigma 300-800mm lens I lose 8.725 million pixels without the 1.4x teleconverter and I lose sixty mm of focal length with it. Using the Sigma 150-600mm lens with the FF Nikon I lose a stop of light and the ability to get reasonably good hand held frames. Which then is the better tool for me? For me, experience with both tells me that the 1.5X crop factor D7200 in this scenario far outstrips the 36 megapixel D810 for my purposes. I get my full 24 megapixel resolution, and I don't have the one stop light loss of the 1.4x teleconverter plus some image quality loss because of the teleconverter. But primarily, I don't have to carry the big tripod, head and much heavier and larger lens and I don't have to set it all up to get an image of an elusive subject.

So in this case, my 24 megapixel D7200 for wildlife at high altitudes is the better choice because of weight, bulk, resolution and photo quality for enlargement. Were I shooting landscapes, undoubtedly the D810 with the Sigma 300-800 f/5.6 would be the far better choice.

My main point here is not to decide for anyone which is the better way to go but to explain how to do the relatively simple calculations. You just take the full frame sensor size of 36 mm by 24 mm which when multiplied gives 864 and use this as a ratio over your resolution in megapixels. Then find the actual size of your crop factor sensor in mm and multiply the width by the height and put this number over the unknown represented by “X”. The solve the simple ratio/proportion and you will have the actual number of pixels the full frame sensor will give you at the same field of view of your crop factor sensor.

So the actual "formula" is 864/number of full frame pixels as a ratio to crop factor expressed as (length times width of your crop factor sensor in mm) over "X". Solve for "X" which gives you the actual number of pixels painting the subject from the FF sensor when cropped to this crop focal length. An example below:

864/36 (full frame 36 mp sensor) 366.6/X (Nikon 1.5x crop) divided by "X" so 864/36 to 366.2/X 864x=366.6 times 36 X equals 13197.6/866 equals 15.24. So when you crop a 36 mp Nikon FF sensor to the field of view of a 1.5x crop factor it paints that FOV with 15.24 megapixels.

Hopefully this will somewhat simplify the mystery for some of how to calculate these values.

Best regards,

Lin

[JudyKay]

Good job Lin. Thanks.

One other factor to toss in regarding FF vs CF sensors is the speed of writing data to the CF card. Shooting wildlife this might matter when a FF maxes out at 4-6 fps (Except the newer 1D's) and a CF sensor provides maybe 8 fps. I get in far away places doing street photography where that matters...maybe. I sure wouldn't be out there with a 1DX--not that I have one.

[Guest Yachtsman1]

I use a Sandisc Extreme Pro in my Panasonic FZ150, never had any problems with stills, bursts or video. I shoot at 16-9 aspect ratio.

Yachtsman1.

[davegee]

Lin,

Nikon sensors are NOT 3:2; a fact that a most/ a lot of people miss when resizing.

DG

[Lin Evans]

Hi Eric,

But this is about dSLR's - the consumer cameras like the FZ150 are a totally different animal...

Best regards,

Lin

[Lin Evans]

Hi Dave,

Some are, some are not - it depends on the the Nikon... At least the DX Nikon cameras I have are 1.5x crop according to Nikon... Could you explain further? My DX Nikons are D7000, D5300 and D7200 - they have 15.8 x 23.6 mm sensors, while 35mm film and FX digital sensors measure 24 x 36mm. Doesn't that equal a 1.5x crop?

Best regards,

Lin

[davegee]

Hi Lin,

I have owned D70,200,300,700 and now 600.

None have been 3:2.

All of them have had more pixels on the long side than they should have.

Example - D70 - 3008x2000.

But aspect ratio and crop factor are different subjects. Are they related?

DG

[Lin Evans]

Hi Dave,

I guess we would have to look at a large number of Nikons to be certain, but for example this is the reported aspect ratio of my older D800E which I traded off a few months back: Below per dPReview which shows both 5:4 and 3:2 aspect ratios (improperly - in my opinion, referred to on dPReview's chart as "other resolutions.").

Other resolutions 6144 x 4912, 6144 x 4080, 5520 x 3680, 4800 x 3200, 4608 x 3680, 4608 x 3056, 3680 x 2456, 3600 x 2400, 3072 x 2456, 3072 x 2040, 2400 x 1600 Image ratio w:h 5:4, 3:2 Sensor size Full frame (35.9 x 24 mm) Sensor type CMOS Processor Expeed 3 Color space sRGB, Adobe RGB Color filter array Primary Color Filter

Crop factor and aspect ratios are somewhat related - of course not exactly the same thing but the sensor is so close to 34x24 mm that for all practical purposes I would refer to it as 3:2 myself.

The D600 specifications are given below: It also appears that Nikon (where these figures come from) consider it 3:2 so let's see how the math works out:

Below Edited later:

3/6015 as X/4016 ... solving for "X" we get 12048/6015 = 2.0029925187032418952618453865337

That's probably about as close to 3:2 as we can get with little light wells I suspect.... ??? (this is the wrong horizontal pixel count - looking again I should have used 6016 which then works out to a perfect 3:2 aspect ratio...)

Best regards,

Lin

D600:

6016 x 4016 Other resolutions 4512 x 3008, 3936 x 2624, 3008 x 2008, 3008 x 1688, 2944 x 1968 Image ratio w:h 3:2 Effective pixels 24 megapixels Sensor photo detectors 25 megapixels Sensor size Full frame (35.9 x 24 mm) Sensor type CMOS Processor Expeed 3 Color space sRGB, Adobe RGB Color filter array Primary Color Filter

[Guest Yachtsman1]

Hi Eric,

But this is about dSLR's - the consumer cameras like the FZ150 are a totally different animal...

Best regards,

Lin

What's in a name, a camera is a camera is a camera-----

Yachtsman1.

[Lin Evans]

Hi Eric,

Not really - there are myriad differences and many types of "cameras." There are Color Filter Array (CFA) sensors some with and some without anti-aliasing filters. There are Foveon sensors which rather than using a CFA have a three layer sensor with no color filter array or AA filter which detects RGB differentially on each of the three layers of silicon. There are various types of "film," etc. There are large arrays of sensors in some cameras and tiny sensors in others. I would say absolutely no - a "camera" is not a "camera." Saying "a camera is a camera" is like saying a boat is a ship... there's a bit of difference between a row-boat and the Queen Elizabeth. Most would agree on this...

Best regards,

Lin

[davegee]

Hi Lin,

This thread could easily go "off-topic".

If you compare "sensor sizes" for N brand vs C brand you will see that C brand is a true 3:2 sensor whereas the N brand is not.

I only shoot RAW Full size images so the other sizes don't mean thing to me, but I am pretty sure that they are "derived" from the full size sensor and crops applied where necessary.

The only time it becomes a problem is when you try to add an uncropped Nikon image to a 3:2 project in PTE. If you have a light coloured bg it will show on the edges of the image. I found a long time ago that you cannot ignore it or say it is "close enough" - it isn't.

DG

[Lin Evans]

Hi Dave,

Isn't that why Igor included the "Low Quality of Resizing" check box in the Properties Tab of O&A? It seems we had some discussions about this a long time ago - can't remember exactly.

Yes there are microscopic differences in aspect ratio between Canon and Nikon and as it relates to PTE it's a good point. On the other hand I've never worried much about it - I just size the image visually in PTE and usually drag it out to a very slight crop for normal display. Of course that won't work so well when one is doing precise geometry for animations so I see where you're coming from.

Best regards,

Lin

[davegee]

I just click on the crop tool at 3:2 and hit enter without adjusting.

DG

[Lin Evans]

Hi Dave,

Yep, that should do it for an exact 3:2 ...

L

[Lin Evans]

Eric,

The differences as Judy was explaining concern the amount of data being processed. Your camera is a 12 megapixel model with a tiny sensor and the file size which must be processed is tiny compared with say a Nikon or Sony 36 megapixel dSLR or even one of the Canon 50 megapixel files which has four times the file size of the consumer model you are discussing. For example, RAW files from my Sigma DP2 Merrill are close to 60 megabytes in size. Processing these internally in the camera is a bit different than processing the relatively tiny files from the 12 mp FZ-150..

Best regards,

Lin

[davegee]

Hi Lin,

Here's a case where it gets interesting.

If you resize a D700 image to 1080 high you get a 1623x1080 image.

If you then apply the crop tool at 3x2 and zoom in you see that the crop tool is highlighting all but 1.5 pixels on each side. Now, if that is the way it works then cropping is going to necessitate a complete remapping of the whole image because half pixels do not exist.

Doing the same thing with a D70 image your crop tool only has to "discard" 2 pixels from each side (1624x1080) - no "remapping" required (?). Would I be correct in assuming that the processor has to work (relatively) harder to crop a resized D700 image over a D70 image??

To those of you who just don't care - that's OK - just don't reply.

DG

[Lin Evans]

Hi Dave,

Sorry - I've been away for a couple hours - I'll have to think about this on for a bit and get back...

Later:

I would assume that in order to crop precisely either complete remapping or perhaps another technique might have to be used such as done in photoshop with "Edit" "Fill" "Content Aware" where those pixels around the sides which are affected by the crop are remapped such as by borrowing the values from adjacent pixels and duplicating those values. This is honestly only a wild guess because I really have no idea how the designer of the tool would proceed. One has to wonder whether perhaps the entire image is resized then cropped or only a content aware fill is used such as in the suggestion above. Since only a scant few pixels are necessarily involved and since all the changes can be done on the extreme periphery it's difficult to know how designers of the firmware or software might elect to proceed.

Interestingly - my two 24 megapixel Nikons (D5300 and D7200) each produce a 6000x4000 pixel image which works out precisely to 3:2 however, when I bring one of my full sized images into PTE with the aspect ratio set for 3:2, use a white background and expand the view to 1000% (type it in manually from 500%) and look at the periphery of the image, I see an approximate couple pixels of white around the right and left sides. The bounding rectangle sets outside of this view slightly. When I set it to Automatic everything seems to line up perfectly. At 500% view there is a small amount of background visible on the left but not on the right. Assuming that the original image is being sized appropriately, that leaves either the background in question or the bounding rectangle math may not be absolutely precise sufficiently for the enlargement. So the answer to your question may have to do with more than just the original images pixel count on the horizontal...

Looking back at my calculations it appears that for the D600 I substituted 6015 pixels on the horizontal for the 6016 which dPReview gives as the horizontal pixel count. If that's the case and there is 4016 pixels on the vertical, doesn't that work out to a perfect 3:2 aspect ratio?? I'm not certain what's actually happening now because it appears to be a correct number of both horizontal and vertical pixels to make a perfect 3:2 aspect ratio....

I looked up your D700 specs and dPReview gives 4756 horizontal by 2832 vertical which works out to an aspect ratio of 3 to 1.99624 so the D700 definitely isn't a perfect 3:2, but the D600 appears to be a perfect 3:2... (EDIT) Actually it isn't a perfect 3:2 but rather 3:2.003... bummer....

L

[davegee]

After resizing a D600 image (without crop) to a 3:2 AR you should get 1620x1080.

Depending on which way you do it (long edge or short edge) a straight RESIZE without crop (preserving the original AR) gives you either:

1620x1081

or

1618x1080

With a D700 image you get either 1620x1077 or 1623x1080.

DG

P.S. Regarding the D5300 and D7200 - it looks like someone has been paying attention to what I have been saying for so long

[Lin Evans]

Yes, even though it appears intuitively that 6016 by 4016 "should" be a perfect 3:2 it's actually 3:2.003 I'm unsure why Nikon needs the extra pixels on the sensor. Seems like they would have made it 6000x4000 just like the D5300, D7100 and D7200... If by no other way, just by masking off the additional pixels. Since it's not possible in a practical sense (acually you can have sub pixels) to have a fractional portion of a pixel I guess they couldn't have made it 6016 x 4010.6. Who knows what their engineers were drinking before final design... probably some kind of sake - LOL

Best regards,

Lin

[davegee]

Lin,

Just out of interest, take your d5300 or d7200 nef into nx2 and add a size/resolution Edit Step.

What does that say?

DG

[Lin Evans]

Hi Dave,

I've got a bunch of things I have to do today so won't get a chance to test it. Here's a link to one of my D5300 nef files if you want to try it. NX2 doesn't work with my d7200 files....

http://www.lin-evans.org/dave/DSC_004.NEF

Best regards,

Lin

[davegee]

I downloaded a jpeg from a review site and it was 6000x4000 as per spec.

However, my d70, 200,300, 700 and 600 images remain something other than 3:2

I will keep an eye on new cameras to see if this new trend continues.

I am not likely to upgrade the d600 for anything less than Nikon's 50MP sensor and only then when it has been proven. The other 50MP sensor took a bit of a beating in the press today so just a big hike in MP is not going to make me jump any time soon.

DG

[davegee]

Lin,

A quick follow up:

All of the current Nikon DX Cameras (except the D300s) have true 3:2 sensors wheras none of the current Nikon FX range have true 3:2 sensors.

DG

[Lin Evans]

Interesting - I've never really thought about the small differences between the nominal 3:2 aspect ratios. I suspect that Nikon probably didn't either possibly thinking that in the vast majority of cases such as in prints, the image would be cropped to a different aspect ratio anyway. It's nice though that the newer DX cameras have a true 3:2 even if they haven't yet implemented that for the FX models.

Yes, Canon took a bit of a beating when the DXO folks demonstrated that their new top-of-line 50 mp sensor didn't even measure up to the DX Nikon models for IQ determiners. I'm perfectly happy with my D7200 for my own personal purposes. I may eventually get another 36mp such as the D810 when prices drop and if I have the money. For the majority of my own use, I prefer the crop sensor for the telephoto boost and for my landscapes I use the Sigma Merrill cameras which give me the equivalent of 30 mp in resolution but with far better sharpness across the full image than I can achieve with any current Nikon or Sigma lens on my Nikon's.

For several years I used medium format digital but the costs became prohibitive since I've retired. I would love a new Hassy 80 mp but it's far out of reach these days.

Best regards,

Lin