Pages: 1 2 3 [4]   Go Down

Author Topic: When and where will DSLR image quality plataeu?  (Read 18438 times)

Fine_Art

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1172
Re: When and where will DSLR image quality plataeu?
« Reply #60 on: November 08, 2014, 12:58:49 pm »

Thanks Jim,
Before correcting my above post, is there anything else needing fixing?
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: When and where will DSLR image quality plataeu?
« Reply #61 on: November 08, 2014, 01:55:28 pm »

Thanks Jim,
Before correcting my above post, is there anything else needing fixing?

I question the premise. While the focal length does indeed play into the angular resolution, the linear resolution in the sensor plane for a diffraction-limited lens is dependent on only the wavelength of the light and the f-stop.

Here's a table I created a few years ago, before I started thinking of this problem in terms of MTF multiplication:

http://blog.kasson.com/?p=432

I haven't looked at it in a while; I hope there are no errors.

Here's a classical image system engineering look at the problem:

http://blog.kasson.com/?p=5742

The reference system in the "System Q" metric is monochromatic, unfortunately. The concept does approximately extend to Bayer sensors, but it gets a little sloppy, since Bayer resolution varies with direction, demosaicing algorithm, CFA filter characteristics, as well as wavelength.

Jim

Fine_Art

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1172
Re: When and where will DSLR image quality plataeu?
« Reply #62 on: November 08, 2014, 05:42:22 pm »

I question the premise. While the focal length does indeed play into the angular resolution, the linear resolution in the sensor plane for a diffraction-limited lens is dependent on only the wavelength of the light and the f-stop.

Here's a table I created a few years ago, before I started thinking of this problem in terms of MTF multiplication:

http://blog.kasson.com/?p=432

I haven't looked at it in a while; I hope there are no errors.

Here's a classical image system engineering look at the problem:

http://blog.kasson.com/?p=5742

The reference system in the "System Q" metric is monochromatic, unfortunately. The concept does approximately extend to Bayer sensors, but it gets a little sloppy, since Bayer resolution varies with direction, demosaicing algorithm, CFA filter characteristics, as well as wavelength.

Jim



The f stop is a handy way of bypassing dealing with the fl and the angle of view. Your table has the pixels across at diffraction which says the same thing.

I think the premise does work because that is how we work as photographers. Personally my two biggest styles of photography are landscape then wildlife. For landscape I don't worry about the details. For wildlife I very much worry about the angle of view and the pixel level resolution. To get undisturbed wildlife shots I use either a 300 f4 on aps-c (85mm front element) or a 1200fl 222mm (10"-1.75" secondary) telescope. That lens has angular resolution that will work on my ff or my finer apsc for 1800fl equivalent from the crop. I can also use a 2x tele with both. I am not blowing $12k on a 600mm lens.

Given most photographers have one goto long lens for wildlife, the angle of view is central to them placing themselves to fill the frame with their quarry. Mess that up you miss it all.

Going back to my original post the premise is what you are shooting dictates the field of view you want. We then translate that into sticking a lens of the proper FL on the camera. For the people that use long FLs they already have to start thinking about the lens with tele, matching their pixel density. For others using wide angles, it will never become an issue.

On the other hand it does not work to go to a scene then pick an f2.8 lens or an f5.6 lens. I bet your table is correct. I do not work that way so IMO that premise is wrong.

Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: When and where will DSLR image quality plataeu?
« Reply #63 on: November 08, 2014, 06:00:17 pm »

The f stop is a handy way of bypassing dealing with the fl and the angle of view. Your table has the pixels across at diffraction which says the same thing.

I think the premise does work because that is how we work as photographers. Personally my two biggest styles of photography are landscape then wildlife. For landscape I don't worry about the details. For wildlife I very much worry about the angle of view and the pixel level resolution. To get undisturbed wildlife shots I use either a 300 f4 on aps-c (85mm front element) or a 1200fl 222mm (10"-1.75" secondary) telescope. That lens has angular resolution that will work on my ff or my finer apsc for 1800fl equivalent from the crop. I can also use a 2x tele with both. I am not blowing $12k on a 600mm lens.

Given most photographers have one goto long lens for wildlife, the angle of view is central to them placing themselves to fill the frame with their quarry. Mess that up you miss it all.

Going back to my original post the premise is what you are shooting dictates the field of view you want. We then translate that into sticking a lens of the proper FL on the camera. For the people that use long FLs they already have to start thinking about the lens with tele, matching their pixel density. For others using wide angles, it will never become an issue.

On the other hand it does not work to go to a scene then pick an f2.8 lens or an f5.6 lens. I bet your table is correct. I do not work that way so IMO that premise is wrong.



To each his own. It's the results that count. If it works for you, you're doing the right thing.

Jim

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: When and where will DSLR image quality plataeu?
« Reply #64 on: November 13, 2014, 08:38:08 am »

Apart from Fuji, the world of APS-C lenses is a sad desert of 18-something zooms with elements made from the bottoms of Coke bottles, and a few better lenses, most of which are 10+ years old, designed when it wasn't clear full frame was going to be viable.
I don't know Nikon well, but my Canon EF-S 17-55 f/2.8 USM IS is decent. I hear that the 15-85 is optically good as well. The EF-S 60mm macro is said to be good. The new EF-S 11-18 appears to be a solid performer at a low cost.

What is really missing (in my view) is top-notch normal primes with large aperture (and ideally IS). My guess is that we will never see those, since the customers willing to pay for (and carry) those kind of lenses can easily opt for FF.

I have gotten the impression that there is little reason to do e.g. an EF-S 70-200 since the reduction in size/weight/price would be minimal for a given performance, vs just using the excellent EF versions?

-h
Logged

dwswager

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1375
Re: When and where will DSLR image quality plataeu?
« Reply #65 on: November 13, 2014, 02:08:20 pm »

I don't know Nikon well, but my Canon EF-S 17-55 f/2.8 USM IS is decent. I hear that the 15-85 is optically good as well. The EF-S 60mm macro is said to be good. The new EF-S 11-18 appears to be a solid performer at a low cost.

What is really missing (in my view) is top-notch normal primes with large aperture (and ideally IS). My guess is that we will never see those, since the customers willing to pay for (and carry) those kind of lenses can easily opt for FF.

I have gotten the impression that there is little reason to do e.g. an EF-S 70-200 since the reduction in size/weight/price would be minimal for a given performance, vs just using the excellent EF versions?

-h

Many factors including ones you cited prevent Nikon or Canon from doing too much in DX lenses.  I have owned 2 DX bodies D300 and D7100 and I have ever only purchased 1 DX lens - 18-200mm f/3.5-4.5 VR II which is surprisingly good, distortions and all.  The DX Nikkor 35mm f/1.8 looks like a steal, but I already have a 35mm f/2D.  There is nothing preventing you from using FX size lenses with the benefit that as FX cameras come down in price you can use them on both.  On the wide end, it is just too much cheaper and easier to do fast lenses for FX than equivalent FOV on DX. 
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: When and where will DSLR image quality plataeu?
« Reply #66 on: November 14, 2014, 12:57:54 am »

There is nothing preventing you from using FX size lenses with the benefit that as FX cameras come down in price you can use them on both.
But designing a lens specifically for crop sensors evidently allows for smaller, lighter and cheaper lenses (in some cases). So should one buy lenses for what one needs today, or for what one might need in the future?
Quote
On the wide end, it is just too much cheaper and easier to do fast lenses for FX than equivalent FOV on DX. 
Crop wide-angles tends to be cheaper than FF wide-angles. The EF-S 10-18mm f/4.5 - f/5.6 with IS costs $299 and gets good reviews.

http://www.the-digital-picture.com/Reviews/Canon-EF-S-10-18mm-f-4.5-5.6-IS-STM-Lens.aspx

An equivalent FF lens would have to be 16-29mm f/7.2-f/9 IS. Such a lens does not exists AFAIK, but if it did, would it cost $299? I doubt it. The Canon 16-35mm f/4.L IS costs about 4x as much. Granted, it is a very different lens.

We could talk about the Sigma 18-35mm f/1.8 vs 24-70mm f/2.8.

-h
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: When and where will DSLR image quality plataeu?
« Reply #67 on: November 14, 2014, 01:20:58 am »

Please do!

Best regards
Erik


We could talk about the Sigma 18-35mm f/1.8 vs 24-70mm f/2.8.

Logged
Erik Kaffehr
 

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: When and where will DSLR image quality plataeu?
« Reply #68 on: November 14, 2014, 03:46:46 am »

Please do!

Best regards
Erik

http://www.photozone.de/canon-eos/872-sigma1835f18_canon?start=1
"The Sigma 18-35mm f/1.8 DC HSM | A produced nothing short of stunning resolution figures in the MTF lab.
...
this Sigma lens is also as expensive, as big and as heavy as its full format counterparts"
Tested on a (15MP, 3168-line) Canon 50D. MTF50 : 2500 LW/PH @ 18mm and 24mm f/1.8 in image centre. >2000 LW/PH at centre/border/extreme at all apertures and focal lengths tested.

http://www.photozone.de/canon_eos_ff/773-canon2470f28mk2ff?start=1
Tested on a (21.1MP, 3744-line) Canon 5Dmk2. MTF50 : 3500 LW/PH @ 24mm and 40mm f/2.8 in image centre. >2300 LW/PH at centre/border/extreme at all apertures and focal lengths tested.

If you want the same image quality, light-gathering capability/DOF etc, it seems that similar weight, size, price etc is needed in order for a product to be feasible (constrained by >f/0.5 not being practically available at any crop factor). The wildcard, it seems, is that manufacturers _choose_ to make products based on what they believe will sell, and so far they have often prioritized lower price/size/weight on the smaller formats, and more light/DOF-flexibility for larger formats.

-h
« Last Edit: November 14, 2014, 03:52:32 am by hjulenissen »
Logged
Pages: 1 2 3 [4]   Go Up