Down-sampled high MP sensors provide more spacial resolution.

[dancook]

I have seen examples of how the D810 is sharper than the D750 and the 5DSR is sharper than a 5DM3 - I thought maybe it's all down to the lack of filter

I have seen an article on producing super high-res images,a 92MP stacked image downsampled to the camera's sensor MP for comparison to a straight forward shot, has more spacial resolution (sharper again).

Then even Ken Rockwell

A higher-resolution camera set to record down at only 12MP gives a sharper picture than a native 12MP camera because it eliminates the softening effects of Bayer interpolation.

In English, cameras lie about their resolution. Cameras really only have one-third to one-half their rated resolution because their sensors detect only one color (red, or green, or blue) at each pixel location. In order to create a color image, the data for each individual color (red or green or blue) is interpolated (smeared) across several pixel locations so we have a red, a green and a blue value for every pixel location.

This smearing is called Bayer interpolation, since Mr. Bayer defined how the red, green and blue sensors are laid out across most sensors. Since we only can detect one color at each pixel location, we have to smear the other colors around so we have full color for each pixel. If we look at an image at its native resolution, all cameras lose sharpness because of this.

When we reduce the size of the image either later in software or right there in the camera, if we reduce it enough then we have complete red, green and blue information for every one of the fewer pixel locations.

In English, when we reduce image size, the resulting image will be sharper pixel-to-pixel than when it was at the full native resolution

This is why you'll see that a 50MP camera shot at its 12MP setting is much sharper than a 12MP camera shot at 12MP.

So why do I see people labouring the point that a high MP camera is a waste of money? that the images when downsampled for use on our monitors will look NO different from those of a lower MP camera.

I'm just trying to understand the whole 'picture' as it were..

 

[seaodyssey]

Possibly because they use the same sharening parameters on every picture.
They don't sharpen for the camera, nor for the picture, nor the output.
Lots of people only sharpen once.

Don't ask for how to do it as I'm still getting my head around it.

Pete

 

[Kodiak Qc]

•
I have 3 bodies with different pixel counts: 12+MP, 24+MP and 36+MP. Since
my work is in the "publishing" aspect of the business, it is clear that all 3 make
sense. I would not shoot with 36+MP for web of screen end use, just like I would
not use 12+MP for posters or centrefolds!

As for sharpening, I never render sharpened files.

 

[dancook]

I'm not suggesting additional sharpening, just the suggestion there is more detail in the 50MP image which when down sampled maintains more spatial resolution.

Here is a downsampled 92MP image for a more distinct comparison.

http://petapixel.com/2015/02/21/a-practical-guide-to-creating-superresolution-photos-with-photoshop/

 

[simonblue]

Their for me two ways of looking at it.

1/ If your an amateur photography, does the camera you using suit you and your need,and turn out the photos your happy with,regardless MP count ?
2/Your a pro, again does the camera you use serve your needs,and are your client happy with your work ?

 

[gramps]

Looks interesting ... but also a bit of a faff!

 

[dancook]

Looks interesting ... but also a bit of a faff!

Creating a supersize resolution image from stacking images is a faff, but it was only a way to help explain how a higher MP camera could provide more detail even when down-sampled.

I'm more interested in peoples thoughts on a 50MP camera providing more detail than a 24MP camera even when downsampled to 24MP. So even beyond the ability to crop and print large - there is an advantage to high MP images even when viewing them to fit on our monitors.

Whether or not this helps me understand how close full frame can ever come to medium format, nothing to do with printing large either, I don't know.

 

[HoppyUK]

<snip>
So why do I see people labouring the point that a high MP camera is a waste of money? that the images when downsampled for use on our monitors will look NO different from those of a lower MP camera.

I'm just trying to understand the whole 'picture' as it were..

Those people are theorists who don't fully understand the theory You will always gain sharpness when down-sampling a high resolution image because of the way the MTF characteristics cascade (Modulation Transfer Function), amongst other things. For example, if you have a 24mp sensor and a lens that will deliver sharpness to the same standard (both resolution and contrast levels, however you want to define those!) when you combine the two you end up with substantially less than 24mp of detail.

There are a lot of things missing from Ken Rockwell's piece. Eg, he's working at above the diffraction threshold that caps sharpness below its potential and we don't know the effect of AA filtering (never published), other than it reduces sharpness and lower mp sensors require much stronger AA filtering than higher mp ones. Also, Bayer interpolation only synthesises colours, not brightness. There are losses though, that are perhaps not explained simply by the presence of an AA filter. We don't know, but try a Sigma Foveon camera and the extra detail is quite dramatic.

At the end of the day though, more pixels will always give you enhanced resolution, higher contrast, and improved sharpness, even if nothing else changes - all other things being equal. But since they never actually are precisely equal, accurate comparisons are very difficult and the gains are relatively small (not a linear increase, or pro-rata, or anything near) and hard to attribute. The biggest contributor to sharpness though, is simply format size rather than pixels (within reason). Physically larger sensors allow the lens to perform at a higher MTF contrast level and sharpness goes up. That's easy to see and to prove, hence FF beats APS-C, which beats M4/3, and so on down.

The other thing about larger sensors, that is perhaps more important now that even smartphones are capable of at least 'good enough' sharpness in good light, is their ISO performance in lower light. That's where smaller formats still struggle, though technology is closing the gap.

 

[Pookeyhead]

I have seen examples of how the D810 is sharper than the D750 and the 5DSR is sharper than a 5DM3 - I thought maybe it's all down to the lack of filter

I have seen an article on producing super high-res images,a 92MP stacked image downsampled to the camera's sensor MP for comparison to a straight forward shot, has more spacial resolution (sharper again).

Then even Ken Rockwell



So why do I see people labouring the point that a high MP camera is a waste of money? that the images when downsampled for use on our monitors will look NO different from those of a lower MP camera.

I'm just trying to understand the whole 'picture' as it were..

I've already covered this comprehensively, with demonstration images.

https://www.talkphotography.co.uk/threads/resolution.606495/#post-7107720


The Canon 50MP camera does not look better when down-sampled because it's 50MP.. but because its the only full frame camera in the line up apart from the A7 and I can't really tell them apart that easily except at 100% and that makes no sense, and doesn't agree with my findings. The A7 shot doesn't look to be perfectly focused either to me.

Sharpening appears to have been applied unequally to these images too. He says he's applied none, but he's clearly shooting in JPEG because he's talking about camera processing settings. To really test this you need to compare like for like, which means shooting raw files with absolutely no image processing. He only mentions sharpening being set for the Canon images.

Skewed, subjective test to generate page hits which is Rockwell's usual MO.

Rockwell is an idiot.

The other thing about larger sensors, that is perhaps more important now that even smartphones are capable of at least 'good enough' sharpness in good light, is their ISO performance in lower light. That's where smaller formats still struggle, though technology is closing the gap.

The same technology applied to bigger sensors though, means that gap remains equal.

 

[HoppyUK]

Further to my post above re cascading MTF factors, here are some comparison lens test figures I did some time ago that show the effect of increased pixel density on sharpness. Everything is identical, except the camera. Lens is a Sigma Art 35mm at f/4 and the cameras are a Canon 5D2 (21mp FF), Canon 7D (47mp-equivalent on FF) and Nikon-1 V1 (74mp-equivalent on FF). At 48-lpmm, MTF measured 55% on the 5D2, 67% on the 7D, and 71% on the V1. All those cameras have AA filters, and more recently I've compared the 74mp-equiv of the Nikon V1 to the Canon 5DSR that has 50mp on FF, but with the AA filter cancelled - and the MTF figures were identical.

David's point that improved sensor technology applies to all formats is true, but I believe that in practise it is not being applied equally. Smartphones are getting the lion's share of investment and the overall image quality they're capable of (and this is just my view, nothing scientific) seems to be way better than their tiny sensors should seem to be capable of! And at the other end of the scale, medium format digital (low investment priority) was using the very old CCD tech that was massively outclassed on ISO and noise performance by much smaller/newer CMOS sensors.

 

[SFTPhotography]

•
!

As for sharpening, I never render sharpened files.

Why not?

 

[Kodiak Qc]

Why not?

•
Clients hate that! …simply.
It is not the client really but his AD since he is in charge of the artistic
rendition on the file.

When the shot was taken, it had to be sharp… as the AD was possibly
present and accepted it. If he was not, at this level, you dare not have
a OoF SOOC take and your goose is cooked in the business.

There is just no way out: if ones wants to be working in this trade, never
be in a position where you have to find excuses… any excuses.

 

[Pookeyhead]

I'm not suggesting additional sharpening, just the suggestion there is more detail in the 50MP image which when down sampled maintains more spatial resolution.

Here is a downsampled 92MP image for a more distinct comparison.

http://petapixel.com/2015/02/21/a-practical-guide-to-creating-superresolution-photos-with-photoshop/

View attachment 51178

There are differences between those two that are nothing to do with resolution. One has a great deal more noise and what looks like compression artefacts to me.

 

[SFTPhotography]

•
Clients hate that! …simply.
It is not the client really but his AD since he is in charge of the artistic
rendition on the file.

When the shot was taken, it had to be sharp… as the AD was possibly
present and accepted it. If he was not, at this level, you dare not have
a OoF SOOC take and your goose is cooked in the business.

There is just no way out: if ones wants to be working in this trade, never
be in a position where you have to find excuses… any excuses.

I'm not sure I follow. You'll never present OOF images or even work on them but sharpening properly is surely a wise step.

Perhaps your client will do their own which is fair enough?

 

[Pookeyhead]

I'm not sure I follow. You'll never present OOF images or even work on them but sharpening properly is surely a wise step.

Perhaps your client will do their own which is fair enough?

All digital images are sharpened. It's set in the metadata, and there will always be sharpening unless you actually set it to 0 yourself, which no one does.

 

[Kodiak Qc]

I'm not sure I follow.

Of course you do!

You'll never present OOF images

Nope, that's out of the equation.

or even work on them…

I said I never render a sharpened image… but a sensor is mineral, not organic!
…so some work has to be done but never sharpening.

but sharpening properly is surely a wise step.

… not if you want to keep your friends!

Perhaps your client will do their own which is fair enough?

That's why the AD is there with his graphic team.

 

[Kodiak Qc]

It's set in the metadata, and there will always be sharpening unless you actually set it to 0 yourself, which no one does.

This is why it is important to have the correct lens profile*** from the manufacturer.
Once the proper AF fine tuning is save in the camera, it is in the RAW file and that
will be "read" by my converter and the lens profile added before I preview. All these
plus the required post-production will be reflected in the rendered end file format.

*** EDIT: and camera profile

 

[Pookeyhead]

This is why it is important to have the correct lens profile*** from the manufacturer.
Once the proper AF fine tuning is save in the camera, it is in the RAW file and that
will be "read" by my converter and the lens profile added before I preview. All these
plus the required post-production will be reflected in the rendered end file format.

*** EDIT: and camera profile

Regardless... unless you actually manually set that sharpness slider to 0, you are sharpening your images because sharpening will be applied automatically. Fact.

As for AF fine tuning, I've never once adjusted any of my cameras/lenses for that (except to once to try and establish if my first D800 had the left AF focusing fault). My images are perfectly sharp.