Downside of using FF lenses on crop bodies

gonzogolf wrote in post #17928895
There is no reason to make ef-s primes lomger than 28 and there seems little demand for any wider.

The desire to have fast wide options is the primary driving force for me wanting to go FF. There just isn't a 24/1.4 equivalent option for crop. The Tokina 14-20/2.0 is now the closest option. I guess you could say Canon realizes this and would rather sell me more expensive FF gear.

wallstreetoneil wrote in post #17924512
the important observation is that the APCS lens focuses all the gathered light onto the smaller sensor which is of course not remotely true of a FF lens that is only projecting a fraction of the gathered light onto the smaller sensor

more light = more detail

What makes you think that all gathered light is delivered to the sensor?

.

The only downside to using FF lenses on crop bodies is that you *may* pay more money for a lens. Ignoring every other feature/function of a lens and basing your choice of lens on IQ alone is like choosing a shifter cart to commute to the grocery store because it's faster than your passenger sedan.

Tom Reichner wrote in post #18018193
What makes you think that all gathered light is delivered to the sensor?

.

Yep, those round sensors are great!!!

Wilt wrote in post #17925622
OK, that is fine. He did state, as a generality, "Lenses designed for smaller sensors will give you sharper results because they're focusing all that light that is coming into the lens onto the sensor itself." not true.

And what about the absolute untruth in the statement that you have to multiply the f/stop as well, as it transmits less light?!

I'm one to think he misspoke and meant to say that EF-s lenses were built more with pixel density in mind and "usually" better (watch his other videos, he has a clear grasp on all matters photography). Would have been true before canon started making their modern lineup of lenses, now ef-s lens advantage are price and size.

I didn't watch the whole video the BS was really rather getting to me. The biggest problem with all of this is the fact that none of the "usual" review sites actually measures the resolution of the lens! Not only do they not measure the resolution of the lens, they also throw away a century or two's normal practice when measuring resolution and change the normalisation from resolution per unit distance, to resolution per image. The trend for measuring the resolution while using a camera body, and normalising the resolution to the image size simply means that the body with the highest total number of pixels will win. Of course that used to be great for the "FF" fans, as the 1.6× linear increase in sensor size used to insure that even though the 35mm format sensor was using significantly larger sensels it would still have more pixels than the smaller sensor. Now in the Canon world at least, we have several APS-C sensors with a significantly higher sensel count than the popular 35mm format sensors things could start to get interesting, although the 5DS with it's 50MP still manages to keep the Canon 35mm sensors out in front on the resolution front.

Personally I'm not too interested in resolution normalised to the size of the image. I spend so much time focal length limited at 600mm that I am often stuck with only being able to use approximately 15×10mm of the sensor. So actually I'm really interested in having the highest possible sensor resolution. There is no 900mm lens, which is what it would take to fill just an APS-C sensor with my subjects, and I cannot get any closer, it would be illegal. There are some 800mm options but I can neither afford them, or am I even physically capable of lifting let alone using one. To fill a 35mm frame I would need a 1440mm lens working at f/6.3 for the exposure, even if I would only need to work at f/15 for DoF. The problem is that it is the exposure that is the limiting factor in the lens design, and a FL of 1440mm @ f/6.3 requires an entrance pupil of 229mm, that's 9"!

Oh and when it comes to lens resolution I would really much rather know exactly what the true optical resolution was that it could deliver, normalised to a true linear measurement. That way I could gauge it's output, for both black/white output and also colour information with regard to the Nyquist limit of the sensor in the camera that I am using. If we do not know the resolution in LP/mm that the lens can present to the sensor, how would I know if I were to want to pick the 5DS over the 5DS-r? IIRC for a Black/white signal the Nyquist limit for the 5DS-r is around 127 LP/mm, for a coloured signal though, the Nyquist limit is only 63.5 LP/mm for full colour depth information. Clever processing could probably reduce that difference by around 50%, thanks to the configuration of the Bayer Colour Filter Array, but at the cost of a reduced colour depth. Pretty much all of the published testing makes any of this impossible.

Alan

BigAl007 wrote in post #18018653
I didn't watch the whole video the BS was really rather getting to me. The biggest problem with all of this is the fact that none of the "usual" review sites actually measures the resolution of the lens! Not only do they not measure the resolution of the lens, they also throw away a century or two's normal practice when measuring resolution and change the normalisation from resolution per unit distance, to resolution per image. The trend for measuring the resolution while using a camera body, and normalising the resolution to the image size simply means that the body with the highest total number of pixels will win. Of course that used to be great for the "FF" fans, as the 1.6× linear increase in sensor size used to insure that even though the 35mm format sensor was using significantly larger sensels it would still have more pixels than the smaller sensor. Now in the Canon world at least, we have several APS-C sensors with a significantly higher sensel count than the popular 35mm format sensors things could start to get interesting, although the 5DS with it's 50MP still manages to keep the Canon 35mm sensors out in front on the resolution front.

Personally I'm not too interested in resolution normalised to the size of the image. I spend so much time focal length limited at 600mm that I am often stuck with only being able to use approximately 15×10mm of the sensor. So actually I'm really interested in having the highest possible sensor resolution. There is no 900mm lens, which is what it would take to fill just an APS-C sensor with my subjects, and I cannot get any closer, it would be illegal. There are some 800mm options but I can neither afford them, or am I even physically capable of lifting let alone using one. To fill a 35mm frame I would need a 1440mm lens working at f/6.3 for the exposure, even if I would only need to work at f/15 for DoF. The problem is that it is the exposure that is the limiting factor in the lens design, and a FL of 1440mm @ f/6.3 requires an entrance pupil of 229mm, that's 9"!

Oh and when it comes to lens resolution I would really much rather know exactly what the true optical resolution was that it could deliver, normalised to a true linear measurement. That way I could gauge it's output, for both black/white output and also colour information with regard to the Nyquist limit of the sensor in the camera that I am using. If we do not know the resolution in LP/mm that the lens can present to the sensor, how would I know if I were to want to pick the 5DS over the 5DS-r? IIRC for a Black/white signal the Nyquist limit for the 5DS-r is around 127 LP/mm, for a coloured signal though, the Nyquist limit is only 63.5 LP/mm for full colour depth information. Clever processing could probably reduce that difference by around 50%, thanks to the configuration of the Bayer Colour Filter Array, but at the cost of a reduced colour depth. Pretty much all of the published testing makes any of this impossible.

Alan

dare I say that you shooting scenario is outside the norm of 99% of all shooters

tony is laying down content in laymen terms, and I think he's pretty accurate.

Charlie wrote in post #18018666
dare I say that you shooting scenario is outside the norm of 99% of all shooters

tony is laying down content in laymen terms, and I think he's pretty accurate.

And I think he is talking out of his lower Sphincter.

Alan

BigAl007 wrote in post #18018677
And I think he is talking out of his lower Sphincter.

Alan

revisit his videos with an open mind, his interpretation is unconventional, but he hits his mark.

Wilt wrote in post #17925482

Using photozone.de tests of 35mm f/1.4 lens on both 20MPixel APS-C and 20MPixel FF bodies,

the MTF value at f/2.8 is 3869 lp/ph (line pairs per picture height) on 20MPixel FF while it is 3380 lp/ph on 20MPixel APS-C body. In other words, with a 1.6x smaller FOV image from APS-C has 13% less total detail on the same size print.

Looking back at this post and I realized you are comparing the 35II on crop to the original 35/1.4 on FF so not a fair comparison, which further proves your point. However I don't think many people doubt that the FF body is going to give better resolution in the center of the frame. The trick is how the corners do. One argument is that the EF lenses on a crop only use the better resolution area in the center. From your comparison above, the edge of the frame with the 35/1.4 on FF, the resolution falls off a cliff giving only lp/ph1437. Unfortunately Klaus doesn't test the 35 I on 20mp crop or the 35 II on 20 mp FF to do a fair test, but the 35 II on 20 MP crop in the corners out performs the 35 I on FF 2398 versus 1437.

This however still really doesn't tell us anything because you wouldn't use a 35mm lens to give the equivalent image on both formats. It would be nice to see for instance a 20 MP crop test of the 24/1.4 Art or better yet the 18-35 at 35mm equivalent and compare it to a 20 MP FF test of the 35/1.4 Art. That way you have lenses of the same generation. If you compare the 35/1.4 to the 24/1.4, the 24/1.4 is a newer design and the 35 II is newer than the 24/1.4

I watched the video a while back.. and it all made sense to me I suppose, some of more scientific geeky equations I got lost on but I understood the point.

That being said , my first camera was the t4i, a year before i bought full frame ,I bought the 24-70 mkii.. I was very impressed with the results ,, prior to that I had what many consider the sharpest lens for apsc (sigma 18-35)..
So it's not like I went from kit to 2400 lens ..
I went back to look at picturesome taken with t4i and 24-70.. I'd do it again. , I know there are cheaper lighter lenses for apsc... unless I was never planning on moving from the t4i.. then yeah there are better options , not because of IQ, but cost and weight

I don't see how there can be any exposure differences when using a full frame lens on a ff sensor versus a crop sensor.

Think of it this way: When using a ff lens on a ff camera, every part of the sensor gets the exact same exposure (at least theoretically, and not considering vignetting or minor manufacturing inconsistencies in the lens). If, for example, the entire sensor is correctly exposed for a 18% gray wall, every part of that sensor will be correctly exposed. So no matter which part of that sensor area you choose to use -- including any crop sensor sized portion - it will have the exact same exposure as any other part of the sensor or even the whole sensor.

If all other variables remain the same -- such as the lens to sensor distance and aperture -- change between ff and crop sensor should have no effect on exposure.