• 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Forums > Back > next PZ lens test report: Carl-Zeiss Vario-Tessar T* 24-70mm f/4 OSS ZA
#41
Quote:As an A7r user at least I have the option of the 36 MP FF sensor. Yes often according to the tests you won't be getting anything like full benefit of what the camera is capable of. Leaving aside considerations as to what my technique and capabilities are as a photographer, I will encounter conditions, hand hold, use lenses that won't produce the capabilities of the camera.
 

I'd argue that the situations where you won't produce meaningful 36M pixel images will be much more often than not and that the difference in prints will be insignificant from a resolution point-of-view. In this case FF will give you shallower DOF and more post-processing flexibility.

 

Quote:But the same goes for any other combination of camera and photographer out there. If you were standing next to me with your E-5 or whatever and I am achieving  12 MP on my 36 MP camera, what will you be getting on your 16 MP quarter frame sensor ?
 

A tack sharp 16MP shot from an E-M1 obviously provides superior sharpness to a 12MP FF image, because every pixel carries information which is not the case in a somewhat blurry 36MP image where, say, half of the pixel are duplicates.
--Florent

Flickr gallery
  Reply
#42
Quote:Hi dbm,

 

I'm not sure how much blur is introduced by the shutter shake issue on the A7R, but otherwise I agree with you. You will never get worse results from a higher pixel count sensor. I just think it's a shame to store files containing more or less the same amout of information as smaller files. Keep in mind this is purely philosophical.
Hi Florent

 

Yes I didn't think from your post that you thought the extra pixels could make things worse; I just wanted to clarify for others who are reading coz that mistake floats around (though this is a pretty informed and civilised community here at PZ so maybe it was not needed.)

 

Yeah I agree about the shame; I guess storage is cheap. One could just resize all but the really best files as part of ones workflow (isn't there a reduced size DNG standard?) but complications to workflow are even worse....
  Reply
#43
Quote:We are indeed far from human vision's DR.
No, we are not. Our eye's DR is pretty low. Just when we look around, the brightness gets regulated by closing and opening the aperture. DR of RAW far outperforms our vision's DR. We have pretty contrasty vision. High DR has low contrast.

Quote:I just compared the DR between the Ricoh GR and the E-M1. The GR has indeed a bit more DR until ISO 200, then the MFT has the advantage.  Quite surprising given the sensor size difference.

However, it's probably not significant anyway.

DR is indeed important, but I personally don't like the look of HDR images at all. They look totally unatural to me. I think future cameras will be able to produce very high DR images by shooting (very quickly) the same scene at different EVs and automatically combine the different exposures in RAW.
What you are saying is clashing. HDR is high dynamic range tone mapped into a low dynamic range. That is why it looks unnatural. We can't see a high DR because our vision is contrasty with a low DR. We either perceive a high DR capture as totally bland and not contrasty, or when we adjust the tonal curve, we just see a part of the high DR in a contrasty way and the rest of the DR is lost in perceived white and perceived black.

 

[Image: D8D7184DCEBE40CA83096FFEEF55BE40.jpg]

 

This is an example I made to show what increased DR actually looks like. On the left, a normal contrasty DR quite natural (similar to how our vision "works"), of 8 or so stops. On the right what 14 stops look like. The only two ways to make high DR not bland is to put a steep tonal curve in (doing so limits the DR to 8 or so stops again) or by tonal mapping the high DR into low DR (8 stops again).
  Reply
#44
Thxbb;

 

I think what Jabez02 meant was if a 36MP user was getting 12PMP of resolution out of thirty six (which is easy to achieve: reasonable zooms get 15) you would expect far fewer PMPs our of a 16MP sensor. And indeed the very best M43 lenses on M43 sensors seem to get 8-10.

 

None of that says that you *need* more than 8-10PMp; I used to have an 8MP camera once and I imagine the lens/camera system was probably getting 4PMP and it seemed fine. The point is just that even holding fixed the greater importance of blur, the difficulty of designing lenses etc. the more more MP generally the more PMP in the result, even if often not much more.

 

But of course I agree few need that; there are other reasons for choosing between systems than resolution.
  Reply
#45
Quote:No, we are not. Our eye's DR is pretty low. Just when we look around, the brightness gets regulated by closing and opening the aperture. DR of RAW far outperforms our vision's DR. We have pretty contrasty vision. High DR has low contrast.

What you are saying is clashing. HDR is high dynamic range tone mapped into a low dynamic range. That is why it looks unnatural. We can't see a high DR because our vision is contrasty with a low DR. We either perceive a high DR capture as totally bland and not contrasty, or when we adjust the tonal curve, we just see a part of the high DR in a contrasty way and the rest of the DR is lost in perceived white and perceived black.

 

[Image: D8D7184DCEBE40CA83096FFEEF55BE40.jpg]

 

This is an example I made to show what increased DR actually looks like. On the left, a normal contrasty DR quite natural (similar to how our vision "works"), of 8 or so stops. On the right what 14 stops look like. The only two ways to make high DR not bland is to put a steep tonal curve in (doing so limits the DR to 8 or so stops again) or by tonal mapping the high DR into low DR (8 stops again).
 

Wait brightcolors.

(I have a professional interest in human vision)

 

You are right that the DR of the eye is low.

But the DR of the visual system is huge! 

 

What happens is that we scan different parts of the visual field (the eye only getting high rez information about a tiny area at any one time) and the brain does something a bit like tone mapping (except it can't really be thought of mapping high DR into low for various reasons) so that we get an overall impression of detail in the deepest shadows and brightest highlights. And part of the neural magic is that it still seems like the shadows are much darker than the mid tones or highlights.

 

This is the problem with HDR photography: it doesn't always trigger than neural effect, so it is obvious that the shadows in one area are no darker than what ought to be mid tones elsewhere.

 

Nonetheless very high DR capture can be very useful, if edited by had selectively, and can seem quite natural.
  Reply
#46
Quote:No, we are not. Our eye's DR is pretty low. Just when we look around, the brightness gets regulated by closing and opening the aperture. DR of RAW far outperforms our vision's DR. We have pretty contrasty vision. High DR has low contrast.
 

dbm below explains it better than me. Our brain is just very good at adapting the exposure on the fly which leads to a high DR perception while the eye itself has a DR close to a camera sensor.

 

Quote:What you are saying is clashing. HDR is high dynamic range tone mapped into a low dynamic range. That is why it looks unnatural. We can't see a high DR because our vision is contrasty with a low DR. We either perceive a high DR capture as totally bland and not contrasty, or when we adjust the tonal curve, we just see a part of the high DR in a contrasty way and the rest of the DR is lost in perceived white and perceived black.

 

[Image: D8D7184DCEBE40CA83096FFEEF55BE40.jpg]

 

This is an example I made to show what increased DR actually looks like. On the left, a normal contrasty DR quite natural (similar to how our vision "works"), of 8 or so stops. On the right what 14 stops look like. The only two ways to make high DR not bland is to put a steep tonal curve in (doing so limits the DR to 8 or so stops again) or by tonal mapping the high DR into low DR (8 stops again).
 

I've seen your example before and I perfectly now what DR is.

The fact is a sensor able to capture very high DR would be extremely useful.

Why do you think people use ND grad filters when shooting landscape at sunset?

Why do people use HDR techniques in the first place?

There are situations where high DR is very valuable and there is no way around it. Try shooting a sunset on the beach where the background is very bright while the foreground being quite dark. Good luck with that.

Currently, the only way to shoot such scenes is to either use ND grad filters or HDR.

If we had sensors able to capture the whole range of information in these situation, we would have the ability to just take one shot and be able to extract the whole information.

That would be awesome, wouldn't you agree?
--Florent

Flickr gallery
  Reply
#47
Quote:Thxbb;

 

I think what Jabez02 meant was if a 36MP user was getting 12PMP of resolution out of thirty six (which is easy to achieve: reasonable zooms get 15) you would expect far fewer PMPs our of a 16MP sensor. And indeed the very best M43 lenses on M43 sensors seem to get 8-10.

 

None of that says that you *need* more than 8-10PMp; I used to have an 8MP camera once and I imagine the lens/camera system was probably getting 4PMP and it seemed fine. The point is just that even holding fixed the greater importance of blur, the difficulty of designing lenses etc. the more more MP generally the more PMP in the result, even if often not much more.

 

But of course I agree few need that; there are other reasons for choosing between systems than resolution.
 

Looks like I misunderstood what he meant. In this case, I agree.
--Florent

Flickr gallery
  Reply
#48
I don't mean the HDR effects more that you have details in shadow and in clouds. Actually the DR of the eye is not that high but it is adapting very fast. So it is not the eye but the scene which is the limit.

 

After using the GR I think I will not buy a system camera again. The quality of the images is so much better, it so small and fast to use. And you can much easier introduce new technologies like curved sensors etc. too for cameras with fixed lenses. And for this new high resolution cameras you need new lenses anyway.

 

But the best is that I can give it to my family if I buy a new one. ;-)

  Reply
#49
Quote:dbm below explains it better than me. Our brain is just very good at adapting the exposure on the fly which leads to a high DR perception while the eye itself has a DR close to a camera sensor.

 

There are situations where high DR is very valuable and there is no way around it. Try shooting a sunset on the beach where the background is very bright while the foreground being quite dark. Good luck with that.
You mean like this?

[Image: 06E7E74BF25F4227AC06876B8FA5C521.jpg]

(shot in JPG)

Quote:Currently, the only way to shoot such scenes is to either use ND grad filters or HDR.

If we had sensors able to capture the whole range of information in these situation, we would have the ability to just take one shot and be able to extract the whole information.

That would be awesome, wouldn't you agree?
  Reply
#50
Quote:You mean like this?

[Image: 06E7E74BF25F4227AC06876B8FA5C521.jpg]

(shot in JPG)
 

Not quite. More like this:

 

[Image: end-of-world-compare.jpg]

 

Or this:

 

[Image: NDGradExample.jpg]

 

 

Or this:

 

[Image: grad-nd.jpg]

 

 

In all these cases, one could also use a single exposure and postprocess to obtain a somewhat similar result. However, there will be either quite a bit of noise present in the under-exposed area of the image or some blown highlight.

If the over-exposed or under-exposed area is somewhat small like in your example, it's okay. However, in my examples above, these are large areas of the image which is a different ball game.

A sensor providing much higher DR than what is currently available would allow to take a single shot and be able to obtain the same results as with ND grad filters.

--Florent

Flickr gallery
  Reply


Forum Jump:


Users browsing this thread: 3 Guest(s)