Where are we on sensor resolution?

[Dinarius]

In 13 years of digital, I've used Canon 1Ds Mark lll (21Mp), Hasselblad H3Dll 39ms (multi-shot), Sony RX 100 Mark I, Canon 5DMk IV.

In terms of colour straight out of the box, the Hassie blows the others clean out of the water. (I do a lot of art repro)

In terms of the "elasticity" of the file - dynamic range and (in particular) how the file behaves in Curves, I rank them as follows:

Hassie
Canon 1Ds MkIV
Sony RX 100

....and a distant fourth...

Canon 5DMk IV

My point is this:

39MP on a sensor the size of the Hassie is nothing, and you can see it in the files. It's like looking at Fuji Velvia of old.

Squeezing more and more pixels onto the same size sensors hasn't worked for me with my most recent purchase - the 5D MkIV.

Dynamic ranger and colour just aren't there compared to the other cameras, although looked at in isolation, the Canon files are fine.

I haven't tried the new Nikon (a lot of pixels on the same size sensor) and I've heard great reports about the Sony A7r iii ("Medium format is dead", etc.) but, I think that something fundamental needs to happen with sensor technology before we make serious progress. A total redesign from the ground up.

D.

[Jack Hogan]

In 13 years of digital, I've used Canon 1Ds Mark lll (21Mp), Hasselblad H3Dll 39ms (multi-shot), Sony RX 100 Mark I, Canon 5DMk IV.

In terms of colour straight out of the box, the Hassie blows the others clean out of the water. (I do a lot of art repro)

Hi Dinarius,

Unique  perhaps - but 'accurate' for repro out of the box, you've got to be kidding.  Have you seen SSFs for its CFA (or absent those its xRGB matrix)?
If you like the colors it produces better, kudos to blad's software and visual psychologists.  Not much to do with the sensor though.

In terms of the "elasticity" of the file - dynamic range and (in particular) how the file behaves in Curves, I rank them as follows:

Hassie
Canon 1Ds MkIV
Sony RX 100

....and a distant fourth...

Canon 5DMk IV

My point is this:

39MP on a sensor the size of the Hassie is nothing, and you can see it in the files. It's like looking at Fuji Velvia of old.

Squeezing more and more pixels onto the same size sensors hasn't worked for me with my most recent purchase - the 5D MkIV.

Dynamic ranger and colour just aren't there compared to the other cameras, although looked at in isolation, the Canon files are fine.

Based on what you are saying I think what you are noticing are the benefits of a larger sensor and less magnification on the final image.  Not much to do with sensor tech though.

Jack

[Dinarius]

Hi Dinarius,

Unique  perhaps - but 'accurate' for repro out of the box, you've got to be kidding.  Have you seen SSFs for its CFA (or absent those its xRGB matrix)?
If you like the colors it produces better, kudos to blad's software and visual psychologists.  Not much to do with the sensor though.

Based on what you are saying I think what you are noticing are the benefits of a larger sensor and less magnification on the final image.  Not much to do with sensor tech though.

Jack

1. I should have said that the Hassie colour is superior as a starting point. I profile using Phocus to calibrate for each setup. The difference between what I'm getting at exposure and what it changes to after profiling is less than with any other camera I've used.

2. Not sure exactly what you mean by your second point. By elasticity, I was referring to how dynamic range behaves in Curves. What has that go to do with "magnification of the final image"?

I process everything in Linear mode, both in Phocus and C1. Even then the 5D Mk IV dynamic range is easily tested.

Thanks.

D.

[Jack Hogan]

2. Not sure exactly what you mean by your second point. By elasticity, I was referring to how dynamic range behaves in Curves. What has that go to do with "magnification of the final image"?

It was more of a general comment: a larger sensor typically allows more flexibility (elasticity:-) because one often has more options for trade offs to make in post.  For instance when many people say that a larger format has better tones when viewing an image displayed at a certain size, they typically really refer to the fact that the tones they see appear to have more detail.  And they show more detail because the larger format needs to be magnified less than a smaller format in order to be shown at that size.  See for instance here.

Jack

[Dinarius]

It was more of a general comment: a larger sensor typically allows more flexibility (elasticity:-) because one often has more options for trade offs to make in post.  For instance when many people say that a larger format has better tones when viewing an image displayed at a certain size, they typically really refer to the fact that the tones they see appear to have more detail.  And they show more detail because the larger format needs to be magnified less than a smaller format in order to be shown at that size.  See for instance here.

Jack

Jack,

I take your point about detail.

In my view, one of the most revealing aspects of how a file behaves is how it responds to use of the Highlight slider in C1.

In my experience, fewer pixels on a given size sensor behave better - there's just more to recover. I'm not talking about blown pixels. I'm talking about, for example, a landscape with puffy white clouds; you push up the shadows and the mid-tones and this frequently results in the need to recover some of those white clouds. (Yes, I know there are a dozen ways to skin the cat - grad filters etc.) Fewer pixels brings back more of the cloud detail.

And as I said above; this is all most telling in Curves. You can drag fewer (and therefore bigger) pixels in all directions before they say enough!

However, my caveat is that I'm only comparing the cameras I listed. Other high-pixel count cameras such as the D850 and the Sony may indeed be better than the Canon.

D.

[Dan Wells]

At least in terms of dynamic range, which is one of the important factors leading to more flexible files, the Sony-based sensors are well ahead of anything Canon makes right now. Sony of course uses their own sensors, but Nikon and Fuji use their own custom versions of Sony sensors with various advantages and disadvantages over the base models, and smaller players like Pentax and Hasselblad buy stock Sony sensors. Even Phase One is a Sony sensor user, although they (like Fuji) like to mess with the color filtration. Who knows if Canon will close the gap, or if they're focused on other things, but at this point, the best 24 MP APS-C files (Fuji, with some others close) have more dynamic range than at least most Canon FF files...

[Dinarius]

At least in terms of dynamic range, which is one of the important factors leading to more flexible files, the Sony-based sensors are well ahead of anything Canon makes right now. Sony of course uses their own sensors, but Nikon and Fuji use their own custom versions of Sony sensors with various advantages and disadvantages over the base models, and smaller players like Pentax and Hasselblad buy stock Sony sensors. Even Phase One is a Sony sensor user, although they (like Fuji) like to mess with the color filtration. Who knows if Canon will close the gap, or if they're focused on other things, but at this point, the best 24 MP APS-C files (Fuji, with some others close) have more dynamic range than at least most Canon FF files...

That concurs with what I’ve read and been told, as well as with the experience I’ve already described.

If I wasn’t locked into a clutch of Canon lenses (funny how they all sell you a camera that can practically make your dinner for little more than the price of a lens that is designed and built by computers. I’m guessing that cameras are practically loss leaders in the hope of recovering money with the lenses. Thank heavens for Sigma!) I wouldn’t be buying one now. That said, I know I’m knit-picking.

D.

[ErikKaffehr]

Hi,

In my experience, C1 is cheating a lot with highlights. Default C1 processing brightens highlights, so the C1 user is forced to expose highlights under saturation.

Lightroom/ACR go a bit excessive in the other direction, applying highlight recovery by default.

Regarding the main issue, sensor resolution, I would think that we could go quite a bit more towards higher resolution, but results may not look good when 'pixel peeping' at actual pixels. For correct detail, we would want to have small pixels with very little contrast.

But, those images would look soft at actual pixels, even they would result in superior prints.

Best regards
Erik

Jack,

I take your point about detail.

In my view, one of the most revealing aspects of how a file behaves is how it responds to use of the Highlight slider in C1.

In my experience, fewer pixels on a given size sensor behave better - there's just more to recover. I'm not talking about blown pixels. I'm talking about, for example, a landscape with puffy white clouds; you push up the shadows and the mid-tones and this frequently results in the need to recover some of those white clouds. (Yes, I know there are a dozen ways to skin the cat - grad filters etc.) Fewer pixels brings back more of the cloud detail.

And as I said above; this is all most telling in Curves. You can drag fewer (and therefore bigger) pixels in all directions before they say enough!

However, my caveat is that I'm only comparing the cameras I listed. Other high-pixel count cameras such as the D850 and the Sony may indeed be better than the Canon.

D.