The Reality of the "Crop Factor"

[HDNitehawk]

First let me say that there is no advantage in having a "Crop Factor". It is one of those phrases that is thrown around as if you magically get more reach. In reality it is a smaller sensor that wastes or throws away the outside edges of a Full Frame sensor. It just crops your full frame picture down to a smaller size.

However, there is an advantage in having a sensor with a denser pixel count. While I wouldn't have agreed with this back in the 7D days, I would have to say now there is an advantage of having the 7D II body and its increased resolution.

On Jan 1st I was bored and I had a test I wanted to perform to confirm some of the things I was seeing with the 7D II. I set up a test with the 500mm, shooting in manual live view focus mode, shooting at F/4 I took a make shift ISO 12233 chart I had printed off line. I started at 15 feet and took a base shot with the 5D II. Then at 18" intervals I took shots to give comparisons at x1.1, x1.2...etc. I did this with the 1D IV and the 7D II. Then I cropped the center of the pic and resized all samples to the exact same size. My thought process is that this is what we would have done with a real life pic. We would have cropped to get the exact picture we wanted. The end picture would have been either printed or displayed in the exact same size.

What I found is that the 7D II x1.2 was just a hair better than the 5D II x1.0. At the 7D II x 1.3 the 5D II x1.0 clearly beats the 7D II x1.3x. So what I find is that the x1.2 + just a little bit is probably the benefit we see in a crop body where resolution counts.

With the 1D IV the 5D II the 5D x1.0 is much better than the 1D IV x1.1. There is minimal if no additional resolution value using the 1D IV.

The 7D II held an x1.2 advantage both over the 1D IV and 5D II.

Here is an example in real life. The Blue Jay is the same area and about the same distance as the Female Cardinal. The Blue Jay is with the 1D IV and the Cardinal with the 7D II. Shot at 500mm same time of day. Close to the same settings. You can pick the shots apart and say maybe this or that was done wrong, but I am seeing similar shots over and over again.

Notice the Blue Jays eyes compared to the Cardinals. The Cardinals have just a bit better resolution.

BB1H5832 by hdnitehawk01, on Flickr


CQ2A2128 by hdnitehawk01, on Flickr


Now the negative of using the 7D II.

The 7D II is not as precise as the 1D IV. The 1D IV seldom misses. Usually if it misses it is because of me putting it in a bad condition.
The 7D II is random. It will have the bird in focus within the DOF but it will miss the critical point that puts the eye in sharp focus. The 7D II takes a +4 to +6 with the new 500mm. I am thinking about shipping it to Canon to check the AF system. I cannot trust it for a critical shot.

The 7D II is slower than the 1D IV. It also seems like it hunts for focus more.

I haven't given up on the 7D II yet. I have seen the benefit of its increased resolution on small critters and birds in focal length limited conditions. But its AF system has to perform better.

[Kayaker72]

Hi Rick...

I very much appreciate the analysis...and phenomenal pics.

I just want to make sure I understand your methodology. You started at 15 ft, and moved back in 3 ft increments. So is:

• x1.0 = 15 ft
• x1.1 = 18 ft
• x1.2 = 21 ft
• x1.3 = 24 ft?

If this is the case, and the 7D II was barely better than the 5DII at x1.2 vs x1.0, then that would be 21'/15' = 1.4x (+)

So, if that is correct, the 7D II's advantage is roughly equivalent to an optically perfect 1.4x TC without the 1 stop loss of light (but DoF differences)

If that interpretation is correct, not bad at all.

[HDNitehawk]

Hi Rick...

I very much appreciate the analysis...and phenomenal pics.

I just want to make sure I understand your methodology. You started at 15 ft, and moved back in 3 ft increments. So is:

• x1.0 = 15 ft
• x1.1 = 18 ft
• x1.2 = 21 ft
• x1.3 = 24 ft?

If this is the case, and the 7D II was barely better than the 5DII at x1.2 vs x1.0, then that would be 21'/15' = 1.4x (+)

So, if that is correct, the 7D II's advantage is roughly equivalent to an optically perfect 1.4x TC without the 1 stop loss of light (but DoF differences)

If that interpretation is correct, not bad at all.

Woops my bad. 18" is correct not 3' however since I did this on New Years morning and there is a possibility of New Years Eve coming in to play let me go home and make sure the markings on the floor were not 3' instead of 18".

Still if I did make the mistake it makes the benefit appear even greater.

[HDNitehawk]

Brant, it was indeed 18", the 3' was my mistake typing today. I corrected the OP.

I reset the test up tonight and this time took a little more care than I did the first time. I had almost identical results. This time the 5D II crop was just a shade better than the 7D II x1.2 crop. From what I see x1.2 is the number.

There is about a 20% resolution benefit over the 5D II. I wonder how much this benefit would disappear with the 5D III's additional 5% pixel density.

[Kayaker72]

Rick...thanks for the clarification and additional testing.

There is a lot to that conclusion. On one hand, if you want all the reach you can get, 1.2x is something. On the other, it is pretty impressive how far away from the "ideal" 1.6x it really is. And this is just "resolution" and not considering tone, color, noise, etc.

Again, thanks for the comparison.

[Joel Eade]

Doesn't it really boil down to pixel pitch and how many pixels you can get on your subject?

The 7DII has a lot of pixels on a smaller sensor (20.2 megapixels on a 22.4 X 15 mm sensor ) so each pixel is very small compared to say the 1DX (18.1 megapixels on 36 X 24mm sensor)

When you take an image like the birds you took, the 7DII will have a lot more pixels on the bird itself than a full frame camera would.

With more pixels on the subject the IQ holds up much better when cropping...thus the "illusion" of more reach.

Is this thinking correct?

[Joel Eade]

7DII : sensor is 22.4 x 15 = 336 square mm's
20.2 million/336 = 60,119 pixels/sq mm

1DX : sensor is 24 x 36 = 864 square mm's
18.1 million/864 = 20,949 pixels/sq mm

[Kayaker72]

Hi Joel.

That is the argument for the "crop reach" benefit, but it essentially assumes ideal scale up between crop and FF sensors. If ideal conditions existed, two 18 MP sensors with the same framing and same lens would give the same resolution. But they do not.

What ideal scale up neglects is the quality of the pixel on target, different sensitivity to imperfections in the lens (crop uses lens center more), and I am sure other factors.

I think a few people were entertained by this, but AlanF did an analysis of different online crop vs FF resolution test results. What he came up with was that the true benefit of crop vs FF in terms of resolving power was ~1.16x-~1.19x (numbers are fun... ). Very similar to what Rick just observed in his side by side tests, granted they are testing using different methodologies. Actually, Rick's test very much interested me as he was not holding framing to be consistent, yet made similar conclusions.

[conropl]

Joel:

I agree with Brant - not all pixels are created equal. I had read AlanF's bit on CR a while back and it seemed to agree with what I had found with the 1DX vs 7D. That is, although I was loosing reach (so to speak), I could heavily crop the 1DX and still come out with a better image. The 7D was not very friendly to cropping at all. Not to mention, I could raise the 1DX ISO up past the 7D's capability and still crop to get a better image. I would say the 7D-II has better pixels than the original 7D, so it is getting to be a tighter comparison. However, I would not discount that you are effectively magnifying the noise artifacts, and a better set of pixels with less noise is going to hold up to cropping better. If you want that sharp glassy look in the eye of the bird, it is much more difficult to achieve if it is speckled with noise. Good noise reduction procedures like you use are helpful, but they would be even more effective if you start with cleaner set of pixel data.

The rub is, the FF has to loose a lot of pixels to get to the same framing, but with the 1DX vs. 7D, I still found the 1DX's better file to create something more pleasing. I still have the 7D and pick it up occasionally to give my 400mm more "reach", but it does not take long to put it back down and put the lens on the 1DX.

I also noticed you go back and forth with your 1DX and 1D IV. Unfortunately for the IV, on a cloudy day with limited light, for the extra "reach" of 1.3X you need to increase the ISO(compared to the 1DX) to get a higher shutter speed to counter the shake issues with the higher effective Focal length. Now you have more noise compared to the 1DX which can shoot at a lower ISO because the effecting focal length has less shake issues. So now I have a cleaner file that I crop and compare to a noisier file - I would be curious to see which one produces a better picture... I would thing the 1DX, but it may be close.

Pat

[Joel Eade]

I certainly agree all pixels and all sensors are not equal and neither are the down stream electronics so ...... when you bring in all these other factors there is a lot to consider BUT solely for the purpose of resolution (all else being equal) I would think more pixels on the subject would be superior. With so many other variables entering into the comparison it is almost impossible to control for all of them and making comparisons is dicey. Also, most of the images that we are using to compare are very small files for web presentation. The differences may be much more apparent when you go to make a very large print.

With good light and a good photographer the 7DII, 5DIII, 1D4 and the 1DX all seem to make beautiful images to me .... no matter the numbers