How much better is Full-Frame really?

bentcountershaft · Jul 30, 2010

If you go full frame the heavens will open and glory will shine down upon you. You will become the Dos Equis guy. You will be the most interesting man in the world. Your blood will smell like cologne. If you mail a letter with no stamp, it will still be delivered. Sharks will dedicate a week to you. Your personality will be so magnetic you won't be able to carry credit cards. The bulls in Pamplona will run from you.

Big Mike · Jul 30, 2010

Sharks will dedicate a week to you

:lmao:

Rekd · Jul 30, 2010

Great question, Neil. Glad you asked.

Can someone educate the ignorant here? What physically makes the crop less capable in low light vs a FF? All I can figure is there's less area on the FF for the light to be dispersed across? (Guessing, of course)

Thanks.

Robin Usagani · Jul 30, 2010

You can just compare them side by side using same lens and same object and low light. Bump up the Iso really high. Full frame will do better for sure.

Derrel · Jul 30, 2010

Rekd said:
Great question, Neil. Glad you asked.

Can someone educate the ignorant here? What physically makes the crop less capable in low light vs a FF? All I can figure is there's less area on the FF for the light to be dispersed across? (Guessing, of course)

Thanks.

The pixel "wells" on a sensor capture light up to their maximum capacity. A good analogy is a bucket...like take a Canon 5D or a Nikon D3...12.8 or 12.2 million pixels that cover a surface area roughly 24x36mm wide, or 864 square millimeters...the "wells" are very LARGE ones....

then take a Nikon D2x, 12.2 million pixels that cover a sensor that is only 15.7mm x 23.6mm in dimension, or 370 square millimeters in area....the pixels that are dispersed across that smaller sensor are smaller pixels, which can "catch" less light, per-pixel, than the significantly larger pixels of the FF sensor.

Smaller sensors are covered with smaller, lower-capacity pixel "wells" or "buckets" which can capture and hold LESS light than wider, "bigger" wells or "buckets".

An analogy might be found in engines.An old-school, 1960's 396 cubic inch V-8 might have say, 340 horsepower, while a modern 2.8 litre 4-cylinder might also produce 340 horsepower by virtue of fuel injection, turbocharging, and double-overhead valves--and yet, the old-school big-block V-8 might well deliver more foot-pounds of TORQUE, which is what HIGH-ISO performance is most like, in my book...

Statistics and figures can be misleading: really,really high megapixel counts are kind of like engine RPM figures...say you have an engine that can rev to 7,000 RPM....and it's a 1,000 cc motorcycle engine....and you have a 1949 John Deere Model M 3-cylinder tractor that delivers only 3,500 RPM wide-open...which will have greater towing capacity? Answer: it ain't the high-revving (high-MegaPixel count) motorcycle engine...it's the bigger-cylindered, slower-revving, slightly larger displacement tractor engine...

Similarly, figures can be misleading: I know of several 350 car and truck horsepower engine configurations...NONE of which can apply the same amount of raw, pure "pulling power" as say, an old tractor where the horsepower is figured "at the drawbar", not on the dyno...a 15 (fifteen) Horsepower 1940's tractor can out-pull a 400 HP pickup truck...because the tractor's HP is measured at the DRAWBAR, which is AFTER the rubber has hit the road...

So, back to megapixels: two engines, EACH with 6 cylinders. Same cylinder count, 6. One has six very small cylinders and pistons. The other has very LARGE cylinders. Or, in other words, 12.2 megapixels, crop-frame sensor, and 12.2 megapixels, full frame sensor. Bigger "buckets". Same number of them. B oth engines go up to 3,200 RPM (or ISO 3,200) Which performs better at low-speed? Which has higher torque per-cylinder?

Rekd · Jul 30, 2010

Thanks, Derrel. Not bad for a Nikon guy.

So the difference between the engine's cylinder size helps me understand, but with engines, if you have 10 or 12 tiny cylinders pushing 20,000 rpm you get a very substantial rate of acceleration.

What's the trade-off between the FF and the crop? I enjoy being able to crop motorsports shots (my main subject). If you were to compress the density of the crop to match that of the FF would they be even? What if you took into account the fact that the 7D has 18mp, would the quality get better in an apples to apples comparison when the 18mp quality was compressed (or scaled??) to match the physical size of a 12mp?

gsgary · Jul 30, 2010

Derrel said:
Rekd said:

Great question, Neil. Glad you asked.

Can someone educate the ignorant here? What physically makes the crop less capable in low light vs a FF? All I can figure is there's less area on the FF for the light to be dispersed across? (Guessing, of course)

Thanks.

Click to expand...

The pixel "wells" on a sensor capture light up to their maximum capacity. A good analogy is a bucket...like take a Canon 5D or a Nikon D3...12.8 or 12.2 million pixels that cover a surface area roughly 24x36mm wide, or 864 square millimeters...the "wells" are very LARGE ones....

then take a Nikon D2x, 12.2 million pixels that cover a sensor that is only 15.7mm x 23.6mm in dimension, or 370 square millimeters in area....the pixels that are dispersed across that smaller sensor are smaller pixels, which can "catch" less light, per-pixel, than the significantly larger pixels of the FF sensor.

Smaller sensors are covered with smaller, lower-capacity pixel "wells" or "buckets" which can capture and hold LESS light than wider, "bigger" wells or "buckets".

An analogy might be found in engines.An old-school, 1960's 396 cubic inch V-8 might have say, 340 horsepower, while a modern 2.8 litre 4-cylinder might also produce 340 horsepower by virtue of fuel injection, turbocharging, and double-overhead valves--and yet, the old-school big-block V-8 might well deliver more foot-pounds of TORQUE, which is what HIGH-ISO performance is most like, in my book...

Statistics and figures can be misleading: really,really high megapixel counts are kind of like engine RPM figures...say you have an engine that can rev to 7,000 RPM....and it's a 1,000 cc motorcycle engine....and you have a 1949 John Deere Model M 3-cylinder tractor that delivers only 3,500 RPM wide-open...which will have greater towing capacity? Answer: it ain't the high-revving (high-MegaPixel count) motorcycle engine...it's the bigger-cylindered, slower-revving, slightly larger displacement tractor engine...

Similarly, figures can be misleading: I know of several 350 car and truck horsepower engine configurations...NONE of which can apply the same amount of raw, pure "pulling power" as say, an old tractor where the horsepower is figured "at the drawbar", not on the dyno...a 15 (fifteen) Horsepower 1940's tractor can out-pull a 400 HP pickup truck...because the tractor's HP is measured at the DRAWBAR, which is AFTER the rubber has hit the road...

So, back to megapixels: two engines, EACH with 6 cylinders. Same cylinder count, 6. One has six very small cylinders and pistons. The other has very LARGE cylinders. Or, in other words, 12.2 megapixels, crop-frame sensor, and 12.2 megapixels, full frame sensor. Bigger "buckets". Same number of them. B oth engines go up to 3,200 RPM (or ISO 3,200) Which performs better at low-speed? Which has higher torque per-cylinder?

Trust Derrel to come back with a bloody esay :lmao:

Rekd · Jul 30, 2010

gsgary said:
Trust Derrel to come back with a bloody esay :lmao:

Yeah, but sometimes it's almost worth it! :thumbup:

TiCoyote · Jul 30, 2010

Which lens do people most commonly use to shoot weddings on a 1.6?

I imagine a 24-75 would be a pretty standard range on a FF. Does that mean that on a 1.6 people would use a 17-55? It seems that something that wide would be unflattering at the wide end.

Derrel · Jul 30, 2010

Rekd said:
Thanks, Derrel. Not bad for a Nikon guy.

So the difference between the engine's cylinder size helps me understand, but with engines, if you have 10 or 12 tiny cylinders pushing 20,000 rpm you get a very substantial rate of acceleration.

What's the trade-off between the FF and the crop? I enjoy being able to crop motorsports shots (my main subject). If you were to compress the density of the crop to match that of the FF would they be even? What if you took into account the fact that the 7D has 18mp, would the quality get better in an apples to apples comparison when the 18mp quality was compressed (or scaled??) to match the physical size of a 12mp?

I also own some Canon stuff too...but thanks. :lol:

Indeed, you can get a high rate of acceleration...but the fact is, the 17.8 MP sensors perform worse at ultra-high ISO settings than the FF, 12.2 lower-MP count D3s sensor does...same with the Canon 1.3x 1D Mark IV versus the D3s--even though the Canon 1D-IV has a higher MegaPixel count (R.R.M), its sensor begins to suffer from noise as ISO levels go upward...

As to your question, "What is the tradeoff between the FF and the crop?" In a practical sense, we have to look at the cameras actually available for some kind of answer. As I see it, the 17.8 MP sensors do not offer that much more resolution advantage over the 12.2 sensors Sony makes for Nikon in the D300s...the higher density of the 17.8 versus the 12.2 MP, and the slightly larger surface area of the Nikon sensors (1.5x = 370 sq. mm) versus the 1.6x or 329 sq. mm sensors Canon uses can be compared directly to the 16.1 MP Canon 1D-IV which is a 1.3x APS-H sized sensor = 548 sq. mm. and then, the Full Frame D3s sensor 1.0x= 856 sq. mm at 12.2 MP.

So...what we see in real-world is that the D3s offers lower NOISE at much higher ISO values than the 1D-IV can do on a 16.1 MP 1.3x sensor, and the FF 12.2 D3s can hold its image quality at much higher ISO levels than the 17.8 MP Canon sensors in the 7D and T2i. It's hard to compare without bringing in Canon and Nikon camera models, since the two companies use alternative approaches,and both represent the absolute best possible results given different approaches (this isn't a brand war kind of thing!)

If the T2i or 7D sensor's small pixel size and high pixel density were to be carried over to a 24x36mm sized FF sensor, I think it'd be a 39.5 megapixel sensor, if my memory serves me right. At that pixel density, today's Canon lenses would not, for the most part, be good enough to actually utilize that high of an MP count, and diffraction would begin to limit picture sharpness at probably around f/2.5....meaning almost no lens in current production would be at its absolute best performance, and the images shot at smaller apertures like f/4.5 even would look crummy...It's almost like a street legal car that can go 195 miles per hour...a driver, and a road, that could handle that kind of capability would be far, and few between.

As to the 17.8 MP and cropping...the image is already somewhat small to begin with...2.62 times smaller in area than a FF image...and so the pixels are very small..meaning the lens has to deliver an incredibly detailed, high-resolution image to that small sensor, in order to take advantage of the 17.8 million tiny pixels...which is do-able in good light, at moderate to higher ISO levels, but NOT at the ultra-high ISO settings the FF cameras can reach...

The answer to your question is difficult to state,exactly...suffice it to say that MANY professional sports shooters like the Nikon D300s because of the crop-factor sensor and what that does to angle of view...and MANY sports shooters also like the Canon 7D because of the high MP count and crop-sensor...but people who shoot at really DARK venues where absolute,pure ISO value at f/2.8 or f/2 or f/1.4 is "the key" to getting action stopping, clear, low-noise images are kind of gravitating toward the Nikon D3s, which has about a 1.5 stop advantage in absolute noise over the Canon 1D-IV at 16.1 MP on a 1.3x sized sensor...

The extremes are where the sensor sizes seem to shake out...at "Moderate" ISO values, almost anything current can make good images...the diffraction limit, and the noise limit seems to be around ISO 3,200 for APS-C, while the FF cameras can reach formerly undreamed of ISO values and the pictures still hold up. From what I have seen, the 7D's 17.8 MP on 1.6x is already past the point of most lenses, at only f/4.5....meaning stopping down past f/4.5 does not bring ANY MORE sharpness, just more depth of field. Pictures made on a 17.8 MP sensor at smaller apertures do NOT look any higher-resolution than stuff shot on 12.2 MP sensors or 15 MP sensors at slightly larger apertures.

Same with Nikon's D3x at 24.6 MP on FF...that sensor demands the absolute best Nikkor lenses, and diffraction creeps in very fast...smaller f/stops look like crap. Thom Hogan has shown that a 12.2 MP D3 file shot with a good lens can be up-sampled to D3x pixel count, and look virtually identical to most people. Hogan has found that it takes the newest, most-modern,highest-resolution NIkon lenses like the 200 f/2 VR and the 24mm PC-E to show what the D3x sensor is actually *capable of resolving*.

So...the 17.8 7D sensor...it would seem at first glance that it's an ideal candidate for cropping images--but it is NOT "ideal" for several reasons. Lens performance, diffraction issues, small sensor capture area, higher than possible noise, lower than possible color saturation--all of those issues make the 17.8 MP sensors NOT ideal for cropping, but instead IDEAL for using a high-quality zoom lens and filling the frame with the desired subject. That is the ideal way to use a crop-body camera for action/sports...fill the framing area with subject, using the best-possible lenses you can get for your system. Nikon, Olympus, and Sony,and Canon to an extent, have found that 12.2 MP on APS-C and 12 to 13 MP on FF (Nikon D3 series, D700,Canon 5D)
have found that, with ALL factors considered, a 24x36 capture size offers a fantastic mix of enough resolution for actual lenses, across the actual best-performing lens apertures on lenses in circulation now, and at ISO ranges from 100 to 3,200. Because the lenses do well on FF of 12 to 13 MP, cropping can be done, and so can up-sampling, which demands a quality, high-resolution image, and currently, the best lenses are the ones that can make a 12 MP image and up-sample that to the 24-25-26 MP size,with good results.

I'm sorry for the length of the response, but I can't reduce it to a one-line reply like gsgary can....gary can summarize like nobody's business!

The "tradeoff" between FF and crop for me though, is more about lens angles of view and depth of field...to me, I prefer the older approach of a 24 as wide, a 35 as semi-wide, a 50 as normal, 85 as short tele, 100 to 135 as medium, 135 and 180 and 200 as longer, and 300 as "long tele", 400 as super-tele...on 1.6x, those things go out the window, and you start needing to stand wayyyyyy far away from people to use a 50 or an 85mm lens...and on APS-C, the backgrounds have that always in-focus look that to me, looks bad. To me, the tradeoff is that with FF, my lenses are exactly what they were designed to be, and on APS-C I am always having to adjust every,single thing, to compensate for the 2.33x to 2.6x smaller format.

Restomage · Jul 30, 2010

Once you turn to Full Frame there is no going back! LOL

Robin Usagani · Jul 30, 2010

TiCoyote said:
Which lens do people most commonly use to shoot weddings on a 1.6?

I imagine a 24-75 would be a pretty standard range on a FF. Does that mean that on a 1.6 people would use a 17-55? It seems that something that wide would be unflattering at the wide end.

Thats why canon makes EF-S and EF mount.

Neil S. · Jul 30, 2010

bentcountershaft said:
If you go full frame the heavens will open and glory will shine down upon you. You will become the Dos Equis guy. You will be the most interesting man in the world. Your blood will smell like cologne. If you mail a letter with no stamp, it will still be delivered. Sharks will dedicate a week to you. Your personality will be so magnetic you won't be able to carry credit cards. The bulls in Pamplona will run from you.

Well in that case.....

Better go buy one lol. :thumbup: