EOS 70D autofocus tracking???

[Bazsl]

Will the phase detect 19 point AF system in the 70D track subjects moving toward the camera at 100 miles per hour in AI Server mode? In other words, is the 70D usable for still photos of aircraft in flight, horse racing and automobile racing? Any references to articles on the 70D AF sustem that discuss its ability to track subjects moving toward the camera would be appreciated. Thanks.

Bill

[cicopo]

Reporting back re printing. This is one of a set I printed & donated to the local Legion last Rememberance Day. From the math I kept 32% of the original pixels. The photo was taken with a 1D mark lV and 100-400 L IS and printed with an Epson 3880 Pro at 13" X 19" and was crystal clear & very sharp. I just ran it in Photoshop as if I would make another print & it says it would be printing at 266 DPI & I've been taught to stay at 300 DPI or higher, but based on my own print it looks like a little less may still work on a good printer. This is a low res sample of the full frame I started with.

The file I printed can be viewed here.(it's at web resolution but large enough to see it was sharp even after cropping.)

http://plus.google.com/photos/114565060647135800760/albums/5931774341532190993/5931774363069143538?b...

.

[garryjames]

Thanks for your response.  Everything you say makes sense.  

There are two limiting factors, lens resolving power / sharpness ( relaize they are different but complementary)  versus the pixel count in the final cropped image.

I guess the question is this:  Which rules, lens resolving power or pixel count as you compare FF with ASP-C?

I've read that the best lenses, and I count the Canon 300mm F/4.0 IS II USM in this group, are capable of resolving 120 to 140 lines per millimeter at a contrast level of 30%.  30% sounds pretty low, but I figure that contrast can be tweaked a bit in post.  So for a full frame cropped to 40%, i.e. Canon APS-C sensor measures 22.2 x 14.8 mm.  if we multiply 130 squared (effective resolved dots per square mm) times the number of square mm's, we get get 5.5 effective megapixels in an APS-C frame, based on lens capabilities.  Compare this with the number of pixels in a cropped FF, which you pointed out is 8 megapixels.  So it appears that the lens limits the resolving power of the system.

The same is true of the APS-C camera, e.g. the Canon 70D.  the lens will give 5.5 megapixels, while the sensor will provide 20.2 megapixels.  Once again the lens is the limiting factor.  The extra pixels are, what I have heard called "empty magnification" as they don't add to the resolving power of the system.

Seems like the only way to get max resolving power is to get a 600mm lens and shoot full frame.  If we calculate the effective pixels for that system (i.e. 130 squared times 35.8 x 23.9mm) we get 14.5 effective megapixels.  Here again we are limited by the resolving power of the lens.  

My question is this, do real world results support this conclusion?  I got a response that using an APS-C body won't give improved results over a FF sensor cropped down to APS-C size.  Yet, there are lots of YouTube videos, Matt Granger and Tony Northrup included, who recommend using APS-C bodies to "extend the reach of lenses."  This contradicts the theoretical calculations above.

Do we know of a quantitative test that addresses this question?

[garryjames]

Thanks for your response.  I too was attracted to the Tamron 150-600mm F/5-6.3 VR Zoom.  but I just watched a video by Tony Northrup who gave it bad performance reviews.  The conclusion of the video was:

The Canon 400mm F/5.6 (cropped down to 600mm frame size) gives better resuts than the Tamron 150-600mm F/5-6.3 VR at 400mm, 500mm and 600mm.

Here's the link:

https://www.youtube.com/watch?v=1fmMG5jgDwk

[garryjames]

Thanks so much for taking the time to answer the questions.  Nice work on the photo.  You must have been shooting at 1/500 second or slower, else the prop would have been frozen.  You must have a steady hand.  

Here's a shot I took at 1/2000 seconds, nearly rozen propeller.  It represents 16% of the frame or 1.2 megapixels.  

Here's a shot I took two days ago, uncropped.

Here's some detail from the cockpit. It represents 3% of the original area of the frame.  or about 0.6 megapixels.

[cicopo]

I've never even thought about the resolving power of any given lens & I rely on superzooms and get results which is what counts. I have worked my way through a lot of quality gear finding what works for my needs & there aren't a lot of people concentrating on radio control events locally so the clubs appreciate my work. If you haven't taken the time to read the links I provided earlier you need to based on your samples. You also need to understand the limitations of what you can hand hold for several hours at an event plus that there is atmospheric haze influencing sharpness on distant shots.

Here's another link to what will show larger than we can place here, and again it's from the 1D4 & 100-400 at 390 mm & 1/1000. I have since decided to shoot jets at 1/1600-1/2000 simply because it improves the odds for a sharper photo. My max shutter speed for prop planes is 1/320 but I try to stay at 1/250 or slower.

http://picasaweb.google.com/114565060647135800760/VintageWingsWingsOverGatineau2012TheMilitarySet#57...

[garryjames]

Great Photos!  

It looks like a lot of those pictures were taken from another aircraft.  You must be panning.  do you use IS-1?

[cicopo]

Thank You. All photos shot from the ground in the normal spectator area, hand held, IS in mode 2, Tv mode & EC as required for decent exposures and generally only minor post processing to the jpg's. Depending on importance I will usually (but not always) start from the RAW image if a print is needed. That allows me to get event sets on line within a few days after they end & before the next one is held. For my R/C events I am allowed on the flight line so I'm up close for take offs & landings which is why I depend on superzooms. (old 35-350 L which has now been replaced by the 28-300 L IS (in mode 2 on 1 body & the 100-400 L IS on another body)

Here's the link to an album which demonstrates how atmospheric haze cuts clarity. These wouldn't make an event album but were used to see what went wrong & when it began. The pilot forgot to install 1 wing bolt & the 14th photo shows the wing working it's way out of the pocket in the fusilage (note the air gap on the back of the wing). It's performing a judged aerobatic routine and that's done well up & away from the flight line & I would have stopped following it but saw (& heard unexpected throttle changes) he was in trouble & stayed on it until the buffer filled. 7D, 100-400, at least 1/2 mile out. I had the advantage of being able to see the wing rotate through the viewfinder but the others watching couldn't see well enough to know what went wrong. 

http://picasaweb.google.com/114565060647135800760/MarioSFlight?authkey=Gv1sRgCJ-06JTQ7pW4EQ

[garryjames]

Thank you for this post.  I'm learning a lot.  I do appreciate it.

Too bad the pilot damaged his craft...  Great photographic work!  I'm sure the pilot appreciated your documenting the fault.  

Do you ever use IS-1?  I read that it's valuable for panning.  It allows the field to blur in the right / left direction while holding sharpness in the up / down direction.

[ebiggs1]

“Which rules, lens resolving power or pixel count as you compare FF with ASP-C?”

“I got a response that using an APS-C body won't give improved results over a FF sensor cropped down to APS-C size.”

The actual physical size of the APS-C sensor is 22.3mm and 14.9mm. The sensor in a FF is 36mm and 24mm.

The FF sensor is approximately 1.6x larger than the crop-sensor.  OK, we got it.

Pixels, or photosites, are not all the same, and they have an individual size too, it is referred to as pixel pitch. A FF, like a 5D Mk II has large pixels, about 6.4 microns. The much smaller crop-sensor, like a 7D, has photosites, about 4.3 microns across.

The photosites, or pixels, on the 7D sensor are much smaller than those of the 5D sensor. Granted, OK?   If you took the larger pixels of the 5D Mk II, and divided up a 7D sensor using those pixels, you would only get 8 million of them to fit.

When comparing the FF, 5D Mk II image with that of the APS-C sensor, 7D, remember the sensor cannot change the focal length of the lens.

The larger FF sensor captures 36mm x 24mm of what the lens projects. Pretty much corner to corner of the image circle. The crop sensor captures the middle portion of that circle which is 22.3mm x 14.9mm.  Using just a portion of what the lens sees. The FF image can show superior fine detail because of this.

When the FF image is 'cropped' to where it matches the “subject size” of the same image from the crop-sensor, there is not a great deal of difference in the image quality.  In this case of a 5D Mk II vs a 7D.

It is only comparable to see and compare images taken with the two cameras, side by side. And it should be clear that to be a fair test there should be a similar MP count. IE 18 MP crop vs a 18 MP FF. Certainly the results would be different and unreliable if the two were of completely differing sizes.

Thus leading to the thought that a crop sensor is sharper. In certain cases it might well be and is.

After extensive use of both of these camera bodies, and my own findings in actual shooting conditions a crop sensor gives no real benefit with telephoto lenses. And it has a distinct disadvantage when using WA.

And very much, a point that has been not talked about here. When you don't crop the FF image, a FF will be considerably sharper.  No contest.

Think of this, Canon's flagship cameras are both Full Frame. Canon could make this body with any sensor it wanted to.

Yet it chose a 18 MP FF for the 1Dx.

If I had t go out and buy a camera to shoot super telephoto lenses today, it would be the 5D Mk III with no hesitation.

Hands down the best.

All this said I tooseem grab my 7D when I go to shoot with my big lenses. Maybe old time myths die hard.  Even with old time photographers.

One other thing to point out is Mr. cicopo is an experienced and accomplished photographer. He knows what he is doing and probably could get the best from any camera!

"A skill is developed through constant practice with a passion to improve, not bought."

[Skirball]

---

@ebiggs1 wrote:

When comparing the FF, 5D Mk II image with that of the APS-C sensor, 7D, remember the sensor cannot change the focal length of the lens.

---

Nobody cares about the actual focal length, only the relative effect of using a XXX mm lens on an APS-C sensor vs. using the same focal length lens on a FF.

---

@ebiggs1 wrote:

When the FF image is 'cropped' to where it matches the “subject size” of the same image from the crop-sensor, there is not a great deal of difference in the image quality.  In this case of a 5D Mk II vs a 7D.

---

That's not necessarily true, not if the resolving power of the lens is good enough to not limit the crop sensor, and providing we're at low ISO.  You're now comparing an 8 MP image to an 18 MP.  It's a significant enough difference to be able to see.

---

@ebiggs1 wrote:

It is only comparable to see and compare images taken with the two cameras, side by side. And it should be clear that to be a fair test there should be a similar MP count. IE 18 MP crop vs a 18 MP FF. Certainly the results would be different and unreliable if the two were of completely differing sizes.

 

---

But it's not, because Canon doesn't offer a FF camera with enough resolution to compare similar MP count.  It would take a 46 MP sensor to allow you to crop it down to APS size and still have 18 MP.  The camera doesn't exist (for Canon).  So, if you're shooting something where you always are cropping down (such as birds or planes), and you're shooting Canon, you would be better off using a crop sensor camera.  Which is why most people that shoot this type of work use crop sensor.