Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

My posts · My subscriptions
  

  Previous versions of gdanmitchell's message #12016771 « Why Do Canon Lenses Score so Low? »

  

gdanmitchell
Offline
Upload & Sell: On
Re: Why Do Canon Lenses Score so Low?


Regarding the issue of lenses out-resolving sensors (or vice versa) it is important to think this through a bit. There are only three possibilities that I can imagine:

1. Lens and sensor resolution are equal - This is essentially impossible for a bunch of reasons. Not all lenses have the same resolution, no lens has the same resolution across the entire frame, and aperture and other factors create more variables. I suppose the closest you might imagine coming to this ideal would be the very best portion of the frame on your very best lens at the optimum aperture might equal the purported sensor resolution. But in the real world, equal resolution is not reality.

2. The lens out-resolves the sensor - Even considering the factors above, it is possible to imagine a situation in which the lens out-resolves the sensor. (This is the scenario most people are referring to when they complain that the sensor cannot resolve the detail provided by the lens - we've seen that issues raised in this thread.) It certainly would not be a bad thing for the lens resolution to exceed that of a sensor, and given the way that lens resolution is measured, it could be argued that this is perhaps the best option. And you might well keep a lens long enough to use it on a series of bodies, so a sensor that is out-resolved by the lens today could be replaced by a higher resolution sensor in the future.

3. The sensor out-resolves the lens - This also is not necessarily a bad thing. (Well, maybe it is if your lens is so awful that any sensor out-resolves it!) There are a number of potential advantages from sensors that out-resolve lenses, and even from those that might do so by a very large amount. For example, I'd be happy to accurately render the gradients that occur at the edges of objects in the image. While the system may not be able to resolve more subject detail, it could resolve more accurately the image projected on the sensor. There is the possibility of smoother rendering of gradients. It is also possible that a much higher resolution sensor could reduce or eliminate the need for an anti-aliasing filter.

So, I think it might just be time to stop throwing out the "sensor can't resolve the lens resolution" notion every time this subject comes up. First, it seems predicated on the impossible ideal where sensor and lens resolution will be equal. Second, there are potential advantages in increasing both lens and sensor resolution in ways that are not linked to perfectly equal resolving power.

And, yes, I do realize that much of this is moot for all but those who are working very carefully and making very, very large prints. Current lenses and sensors from many manufacturers provide enough and more than enough resolving power for the ways that the majority of photographer are using them.

Take care,

Dan

Scott Stoness wrote:
It has been said before but I will repeat it:

1) Dxomark is a good site that I look to often but you have to know what their numbers mean. The aggregate score is useless and has an arbitrary weighting as applied by Dxomark. The sharpness score is the total equivalent resolution achieved. Thus when a d800 achieves 28, it is failing to make 36mpx (its mpx) that means either that the d800 is failing in achieving its maximum of 36 or the lens is being outresolved by the mpx. When you look at the sharpness of the best canon 5diii you will find that they often hit 22mpx which indicates either the camera was perfect in achieving all that it could and likely the lens can deliver more and was limited by the camera. The distortion and CA numbers are helpful becuase they were done in a similar manner for all lens however, its not clear how they weighted the corners vs the centre. I like to use photozone.de who actually publish the edges seperately from the centre and the CA and vignetting at different fstops as more useful.

2) Often technique makes more resolution not useful. Eg. if you are shooting a 600mm lens with great iq at 1/500, the resolution will not show up. You can only achieve the specified resolution with no wind, tripod bolted down. etc.

3) The resolution of the bodies will rise and fall. Currently the d800 and 14-24 delivers the best landscape (ignoring the fact that they do not have a good UWA ts which means they are not useful to me) but you don't want to be flipping in and out of bodies when you buy $3000 lens. Just wait a year and Canon will pass them and then Nikon will pass them and then Sony will pass them and then Canon will pass them but you will not get an average person to tell the difference. So the right solution is stick with a market leader (nikon or canon) and buy the best in each system and don't look over the fence.

4) Currently my view is that Canon has best tilt shift lens. Nikon does not even have a 17ts. And I think the 5diii is the best all round camera. And Canon has the best (newest design) long lens [600v2, 500v2, 200-400]. The d800 is not great for wildlife. So so might say Canon is best (Ts17, TS24, 600v2, 500v2, 200-400) and some might say Nikon is the best (d800 with 14-24) and some might say sony is best with zeiss lens and a7r. So which is best depends on what you are shooting. For me Canon is best for wildlife. And best for me for landscape (TS17).



Dec 22, 2013 at 03:25 AM
gdanmitchell
Offline
Upload & Sell: On
Re: Why Do Canon Lenses Score so Low?


Regarding the issue of lenses our-resolving sensors (or vice versa) it is important to think this through a bit. There are only three possibilities that I can imagine:

1. Lens and sensor resolution are equal - This is essentially impossible for a bunch of reasons. Not all lenses have the same resolution, no lens has the same resolution across the entire frame, and aperture and other factors create more variables. I suppose the closest you might imagine coming to this ideal would be the very best portion of the frame on your very best lens at the optimum aperture might equal the purported sensor resolution. But in the real world, equal resolution is not reality.

2. The lens out-resolves the sensor - Even considering the factors above, it is possible to imagine a situation in which the lens out-resolves the sensor. (This is the scenario most people are referring to when they complain that the sensor cannot resolve the detail provided by the lens - we've seen that issues raised in this thread.) It certainly would not be a bad thing for the lens resolution to exceed that of a sensor, and given the way that lens resolution is measured, it could be argued that this is perhaps the best option. And you might well keep a lens long enough to use it on a series of bodies, so a sensor that is out-resolved by the lens today could be replaced by a higher resolution sensor in the future.

3. The sensor out-resolves the lens - This also is not necessarily a bad thing. (Well, maybe it is if your lens is so awful that any sensor out-resolves it!) There are a number of potential advantages from sensors that out-resolve lenses, and even from those that might do so by a very large amount. For example, I'd be happy to accurately render the gradients that occur at the edges of objects in the image. While the system may not be able to resolve more subject detail, it could resolve more accurately the image projected on the sensor. There is the possibility of smoother rendering of gradients. It is also possible that a much higher resolution sensor could reduce or eliminate the need for an anti-aliasing filter.

So, I think it might just be time to stop throwing out the "sensor can't resolve the lens resolution" notion every time this subject comes up. First, it seems predicated on the impossible ideal where sensor and lens resolution will be equal. Second, there are potential advantages in increasing both lens and sensor resolution in ways that are not linked to perfectly equal resolving power.

And, yes, I do realize that much of this is moot for all but those who are working very carefully and making very, very large prints. Current lenses and sensors from many manufacturers provide enough and more than enough resolving power for the ways that the majority of photographer are using them.

Take care,

Dan

Scott Stoness wrote:
It has been said before but I will repeat it:

1) Dxomark is a good site that I look to often but you have to know what their numbers mean. The aggregate score is useless and has an arbitrary weighting as applied by Dxomark. The sharpness score is the total equivalent resolution achieved. Thus when a d800 achieves 28, it is failing to make 36mpx (its mpx) that means either that the d800 is failing in achieving its maximum of 36 or the lens is being outresolved by the mpx. When you look at the sharpness of the best canon 5diii you will find that they often hit 22mpx which indicates either the camera was perfect in achieving all that it could and likely the lens can deliver more and was limited by the camera. The distortion and CA numbers are helpful becuase they were done in a similar manner for all lens however, its not clear how they weighted the corners vs the centre. I like to use photozone.de who actually publish the edges seperately from the centre and the CA and vignetting at different fstops as more useful.

2) Often technique makes more resolution not useful. Eg. if you are shooting a 600mm lens with great iq at 1/500, the resolution will not show up. You can only achieve the specified resolution with no wind, tripod bolted down. etc.

3) The resolution of the bodies will rise and fall. Currently the d800 and 14-24 delivers the best landscape (ignoring the fact that they do not have a good UWA ts which means they are not useful to me) but you don't want to be flipping in and out of bodies when you buy $3000 lens. Just wait a year and Canon will pass them and then Nikon will pass them and then Sony will pass them and then Canon will pass them but you will not get an average person to tell the difference. So the right solution is stick with a market leader (nikon or canon) and buy the best in each system and don't look over the fence.

4) Currently my view is that Canon has best tilt shift lens. Nikon does not even have a 17ts. And I think the 5diii is the best all round camera. And Canon has the best (newest design) long lens [600v2, 500v2, 200-400]. The d800 is not great for wildlife. So so might say Canon is best (Ts17, TS24, 600v2, 500v2, 200-400) and some might say Nikon is the best (d800 with 14-24) and some might say sony is best with zeiss lens and a7r. So which is best depends on what you are shooting. For me Canon is best for wildlife. And best for me for landscape (TS17).



Dec 21, 2013 at 08:10 PM



  Previous versions of gdanmitchell's message #12016771 « Why Do Canon Lenses Score so Low? »