Gene51
Loc: Yonkers, NY, now in LSD (LowerSlowerDelaware)
Architect1776 wrote:
Has anyone actually done a real world test themselves or is it all just a bunch of speculation?
I use medium format lenses on my FX cameras all the time - a PC-E lens will cover medium format but is intended for use on a FX camera.
I've used a number of FX lenses on D70, D200, D300 - all of them long lenses - since no one makes DX telephoto primes, without any downside.
If you read my earlier post about Tony Northrup, you'd see that DXO Labs backs up the notion that the center of the lens is sharpest, and that using an FX lens on a DX camera will result in better image quality across the field.
Gene51
Loc: Yonkers, NY, now in LSD (LowerSlowerDelaware)
PHRubin wrote:
Misleading since lenses don't have megapixels. In fact, a crop body camera uses the centermost area of a FF lens, which is the best part. No image quality is lost. The lens projects an image onto the sensor as good as, if not better than, what would be on a FF sensor.
I think you missed something - a DX lens on FX body will only cover the center of the sensor - and in only using part of the sensor there will be a loss of image quality. There will most definitely be fewer pixels in the crop. No different than if you cropped the picture in post processing.
Gene51
Loc: Yonkers, NY, now in LSD (LowerSlowerDelaware)
amfoto1 wrote:
I put VERY LITTLE faith in DXO.... would hardly call it a "defacto standard".... and almost never reference it at all. It just doesn't make sense or IMO have much bearing on "real world" gear performance. (Same with some of Tony Northrup's BS.)
In my dozen or more years experience using the same lenses on BOTH full frame and crop cameras, it's certainly no myth. Crop cameras DEFINITELY use the "best" portion of most lenses. If you look at the MTF charts for virtually any lens, you'll see that the mid-frame and extreme edges have anywhere from a little to a lot lower resolution than the center. And the crop camera only utilizes that central portion.
Because of higher pixel density of crop sensors, they're more demanding of lenses.... so it's good that they "crop away" the softer corners and edges. For example, a 24MP Nikon DX/crop sensor has close to the same pixel density and potential resolution as would a 56MP FX/full frame sensor (which, of course, doesn't exist... the highest resolution Nikon FX currently is 46MP). A 24MP Canon APS-C sensor, which is slightly smaller than Nikon's DX, has the same density and potential resolution as would a 61.5MP full frame sensor (which also doesn't exist... the highest resolution Canon FX currently is 50MP).
I put VERY LITTLE faith in DXO.... would hardly ca... (
show quote)
Another person that is confused by their complex analysis. Their stuff is 100% on target for what they publish - unfortunately they seldom tell the whole story. Their analyses do take into consideration the greater magnification necessary for crop bodies to get to the same size a comparable FF image needs.
joer
Loc: Colorado/Illinois
Gene51 wrote:
I use medium format lenses on my FX cameras all the time - a PC-E lens will cover medium format but is intended for use on a FX camera.
I've used a number of FX lenses on D70, D200, D300 - all of them long lenses - since no one makes DX telephoto primes, without any downside.
If you read my earlier post about Tony Northrup, you'd see that DXO Labs backs up the notion that the center of the lens is sharpest, and that using an FX lens on a DX camera will result in better image quality across the field.
I use medium format lenses on my FX cameras all th... (
show quote)
Their data only shows an improvement in vignette, but not in the other categories. Sure its using the best part of the lens but the sensor can't take full advantage of it. You've seen their charts.
BebuLamar wrote:
I think people in this thread concentrate on Canon and Nikon APS-C cameras which both Canon and Nikon make very few high quality lenses for them. Most of the high quality lenses are for the full frame format. If they were to put the same effort and cost into a DX or EF-S lens that has the same focal length as the full frame version it would outperform the full frame lens.
True. Sadly, they save most of their best glass for FX and Full Frame cameras. Meanwhile, Panasonic, Olympus, Sony, and Fujifilm DO make premium glass for their mirrorless cameras.
Also, mirrorless system lenses never need focus micro-adjustments.
burkphoto wrote:
True. Sadly, they save most of their best glass for FX and Full Frame cameras. Meanwhile, Panasonic, Olympus, Sony, and Fujifilm DO make premium glass for their mirrorless cameras.
Also, mirrorless system lenses never need focus micro-adjustments.
But may be that is why they are the real crop sensor cameras that use less than the image circle formed by the lenses. Not in the case of the 4/3 system.
joer wrote:
Their data only shows an improvement in vignette, but not in the other categories. Sure its using the best part of the lens but the sensor can't take full advantage of it. You've seen their charts.
That is not true. Check MTF charts for any FX lens to see the significant improvement. The DXO charts, which you are clearly not understanding, are NOT even attempting to show characteristics you say show no improvement.
Note that DXO's "Perceptual Megapixel" measurements are specific to a camera. They help compare multiple lenses on one single body. They do not compare or have any meaning at all between cameras with different sized sensors.
joer
Loc: Colorado/Illinois
Apaflo wrote:
That is not true. Check MTF charts for any FX lens to see the significant improvement. The DXO charts, which you are clearly not understanding, are NOT even attempting to show characteristics you say show no improvement.
Note that DXO's "Perceptual Megapixel" measurements are specific to a camera. They help compare multiple lenses on one single body. They do not compare or have any meaning at all between cameras with different sized sensors.
Where did you get that idea?
Thanks everyone for your great responses. I found this
http://www.clarkvision.com/articles/does.pixel.size.matter/ (the part about this topic is towards the end of the page - there's a lot here). He's a scientist that has performed tests about this and many other things of interest. Seems like there is not a noticeable quality loss when using FF lenses on cropped.
joer wrote:
Where did you get that idea?
DXO does explain what it is, though it requires significant technical background to catch what they are saying.
Apaflo wrote:
... Note that DXO's "Perceptual Megapixel" measurements are specific to a camera. They help compare multiple lenses on one single body. They do not compare or have any meaning at all between cameras with different sized sensors.
They actually do. You just need to understand what they are telling you.
The Perceptual Megapixel score combines the lens resolution with the sensor resolution in approximately the following way:
(1/PM) = (1/LM) + (1/SM) where: PM=Perceptual Megapixel, LM=lens megapixel and SM=sensor MP
For example, a 24MP lens on a 24MP sensor would end up with a 12MP Perceptual Megapixel score because (1/12) = (1/24) + (1/24)
But a "perfect" lens (extremely high LP) the 1/LM approaches 0 and PM approaches SM.
To calculate the lens's MP you can rearrange the terms:
(1/LM) = (1/PM) - (1/SM}
This calculation is only approximate. It cannot account for a difference in the actual resolution of the sensor itself which is affected by the Bayer array and whether there is an AA filter or other factors. And it assumes that you are comparing sensors of the same format (size).
But the biggest fallacy is in assuming that a 36 MP camera with a Bayer array automatically delivers 36 MP. But there are only 18 MP of green sensors and 9 MP each or red and blue sensors. The Bayer array actually degrades the resolution by about half and it has to be recreated by the software. A 24 MP monochrome sensor actually starts with 24 MP that have not been degraded by a Bayer array.
However the formulas explain why a non-perfect lens will
always degrade the performance of the sensor.
But getting back to the OP's original question, just about everyone is correct that by using only the better center portion of the image circle will result in a higher resolution. You can confirm this with a superficial understanding of MTF curves. The thing that throws a monkey wrench into the works is the size of the pixels themselves and the lower diffraction limits of a crop sensor compared to a full frame sensor.
"They actually do. You just need to understand what they are telling you."
You first say they do and then demonstrate that they do not!
"The Perceptual Megapixel score combines the lens resolution with the sensor resolution ..."
That is why they cannot be used to compare the same lens on different size sensors. Simply stated Perceptual Megapixels are not megapixels at all, despite the name.
"But the biggest fallacy is in assuming that a 36 MP camera with a Bayer array automatically delivers 36 MP. "
A 36 MP sensor absolutely does deliver 36 MP. Don't confuse that with Perceptual Megapixels. No sensor can produce an image where they are equal. Megapixels are hardware while Perceptual Pixels are conceptual.
Much of your article was totally invalid misperception about Bayer encoding. I have ignored that without comment as it has no bearing on understanding the concept of Perceptual Megapixels.
Apaflo wrote:
"They actually do. You just need to understand what they are telling you."
You first say they do and then demonstrate that they do not!
"The Perceptual Megapixel score combines the lens resolution with the sensor resolution ..."
That is why they cannot be used to compare the same lens on different size sensors. Simply stated Perceptual Megapixels are not megapixels at all, despite the name.
"But the biggest fallacy is in assuming that a 36 MP camera with a Bayer array automatically delivers 36 MP. "
A 36 MP sensor absolutely does deliver 36 MP. Don't confuse that with Perceptual Megapixels. No sensor can produce an image where they are equal. Megapixels are hardware while Perceptual Pixels are conceptual.
Much of your article was totally invalid misperception about Bayer encoding. I have ignored that without comment as it has no bearing on understanding the concept of Perceptual Megapixels.
"They actually do. You just need to understa... (
show quote)
You don't know the difference between lens resolution, sensor/film resolution and combined system resolution.
All you need to know is that either a lens or the sensor or both can limit the resolution of a system. You should also know that the combined system will be less than either of the two components taken separately.
In other words, no matter how good the lens is, it will never deliver more resolution than the sensor. Likewise, no matter how good the sensor is, it will never deliver more resolution than the lens. The formula I presented is an approximation of that relationship.
What I demonstrated is an
approximate way to isolate the two resolutions. You can confirm this by picking any* lens on the DxOMark display and selecting different cameras in the "mounted on" drop-down.
* Any lens except where the Perceptual Megapixel is not too close to the nominal sensor megapixels, which it is for all but a handful of the FX mount type lenses.
If you want to reply, then
register here. Registration is free and your account is created instantly, so you can post right away.