December 21, 2014, 08:00:37 AM

Author Topic: DxO reviews Sony A7s: king of low light photography?  (Read 24378 times)

thepancakeman

  • Canon 7D MK II
  • *****
  • Posts: 457
  • If at first you don't succeed, don't try skydiving
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #120 on: August 05, 2014, 12:27:05 PM »
I just add noise using Photoshop's "Add Noise" filter. That adds noise with a pixel-level frequency...or, in other words, per-pixel noise. It's a VERY minimal amount, you have to squint to see it, as it really isn't supposed to be obvious, and as such, is effectively meaningless in the midtones through highlights/whites.

The general point of this is to smooth out the harsh transitions that usually occur in the shadows due to low bit depth and quantization error during ADC. If you examine lifted shadows, from any camera (including, and maybe even particularly, a D800 or other camera with a Sony Exmor), you will very often notice a bit of posterization. Adding a very light amount of noise breaks that up, which helps improve gradient transitions and such in the shadows. It can also help artificially enhance detail that may otherwise look like smooth blobs due to noise reduction algorithms (this can especially be a problem if you did any noise reduction with masking, so you could apply NR more heavily in the shadows than in the midtones and highlights).

This just made this whole thread worth reading.  Brilliant!  (and thanks!)

canon rumors FORUM

Re: DxO reviews Sony A7s: king of low light photography?
« Reply #120 on: August 05, 2014, 12:27:05 PM »

Chuck Alaimo

  • 1D Mark IV
  • ******
  • Posts: 988
    • View Profile
    • Chuck Alaimo Photography
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #121 on: August 05, 2014, 12:37:28 PM »
This guy really doesn't get it... it's hopeless!

Hey, don't blame me - I'm not the one asserting that extra stops of DR and information in the image files is wasted, it is others. The only problem is that it would seem that all of those bits that they say you don't need are actually used by them anyway.

If people would stop trying to belittle and put down Sony's sensors because they deliver and offer more DR then it would be a whole lot easier. This is where the problems stem from: trying to assert that what comes off the Sony sensors is no better than the Canon's.

I'm pretty sure that if I tried, I could use Google and find threads on here where various people have waxed lyrical (and received support for) about Canon producing sensors with fewer larger pixels with more DR and better IQ - especially in low light. Well guess what, Sony has done that.

But instead of accepting that and congratulating Sony on doing it, people are arguing about how all of that extra DR and IQ is not necessary. What a load of horse sh*t.

wow....
Owns 5Dmkiii, 6D, 16-35mm, 24mm 1.4, 70-200mm 2.8, 50mm 1.4, 85 mm 1.8, 100mm 2.8 macro, 1-600RT, 2 430 EX's, 1 video light

msm

  • Rebel T5i
  • ****
  • Posts: 123
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #122 on: August 05, 2014, 12:53:14 PM »
... No. I have a LIMIT of 14 stops of dynamic range in a 14-bit file...

For each pixel yes.

jrista

  • Canon EF 400mm f/2.8L IS II
  • *********
  • Posts: 4814
  • EOL
    • View Profile
    • Nature Photography
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #123 on: August 05, 2014, 12:54:44 PM »
This guy really doesn't get it... it's hopeless!

Hey, don't blame me - I'm not the one asserting that extra stops of DR and information in the image files is wasted, it is others. The only problem is that it would seem that all of those bits that they say you don't need are actually used by them anyway.

If people would stop trying to belittle and put down Sony's sensors because they deliver and offer more DR then it would be a whole lot easier. This is where the problems stem from: trying to assert that what comes off the Sony sensors is no better than the Canon's.

I'm pretty sure that if I tried, I could use Google and find threads on here where various people have waxed lyrical (and received support for) about Canon producing sensors with fewer larger pixels with more DR and better IQ - especially in low light. Well guess what, Sony has done that.

But instead of accepting that and congratulating Sony on doing it, people are arguing about how all of that extra DR and IQ is not necessary. What a load of horse sh*t.

wow....

Yeah...he seems to think EVERYONE shoots at ISO 100.

I just had three 40x30" prints made of some of my photography. Bird and wildlife photos. All of them were at ISO 1600 and above, one was at ISO 12800. All of them had to be cropped a bit and rotated a little to get the right framing. Every single one of them was upsampled, and required heavy and carefully masked noise reduction in the smooth OOF  background areas to smooth them out, and careful masking and detail enhancement in the foreground detail areas. Even with a D800, we'd be talking about a 2x scale factor, from high ISO shots. There would be ABSOLUTELY ZERO benefit in IQ or DR if I'd used a D800 instead of a 5D III to take these shots, considering how they were printed.

neuroanatomist

  • CR GEEK
  • ************
  • Posts: 15227
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #124 on: August 05, 2014, 01:02:59 PM »
... No. I have a LIMIT of 14 stops of dynamic range in a 14-bit file...
For each pixel yes.

So, your point is... ??

Hopefully not that downsampling a 14-bit file can result in >14-bits of DR.  That's dilbertthink.
EOS 1D X, EOS M, and lots of lenses
______________________________
Flickr | TDP Profile/Gear List

Sporgon

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2079
  • 5% of gear used 95% of the time
    • View Profile
    • www.buildingpanoramics.com
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #125 on: August 05, 2014, 01:27:01 PM »
Yeah...he seems to think EVERYONE shoots at ISO 100.

Yes but what he is saying has no relevance to ISO 100 either. To suggest that if paper only can reproduce x range of latitude so there is no point in working the data in 16 bit as opposed to an 8 bit JPEG is, quite honestly astonishing, more so given the fact that there are now so many good references to post processing on the web.

There really is no excuse for being so misinformed in this day and age.

msm

  • Rebel T5i
  • ****
  • Posts: 123
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #126 on: August 05, 2014, 01:34:10 PM »
... No. I have a LIMIT of 14 stops of dynamic range in a 14-bit file...
For each pixel yes.

So, your point is... ??

Hopefully not that downsampling a 14-bit file can result in >14-bits of DR.  That's dilbertthink.

Per downsampled pixel then yes. Assuming no noise, add for instance 4 14bit numbers and you get a 16bit number....

canon rumors FORUM

Re: DxO reviews Sony A7s: king of low light photography?
« Reply #126 on: August 05, 2014, 01:34:10 PM »

Keith_Reeder

  • EOS M2
  • ****
  • Posts: 223
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #127 on: August 05, 2014, 02:19:32 PM »
To suggest that if paper only can reproduce x range of latitude so there is no point in working the data in 16 bit as opposed to an 8 bit JPEG is, quite honestly astonishing

To be fair, he's saying that this is what he reckons Jrista is suggesting.


scyrene

  • Canon 6D
  • *****
  • Posts: 394
    • View Profile
    • My Flickr feed
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #128 on: August 05, 2014, 03:35:08 PM »
All this has been entertaining, if more than a little reminiscent of a few other discussions...

*Pulls pin out of grenade*

But to return to the original post, I'd say the 645z looks like 'king of low light' at present, based on what I've gleaned online. I'd still rather have a 1Dx (or wait for the next Canon full frame body) for the system benefit, as Neuro says. But in terms of low noise at high ISO I reckon the new Pentax wins - as you'd expect, with a bigger sensor area.

*Runs away*
5D mark III, 50D, 300D, EOS-M; Samyang 14mm f/2.8, 24-105L, MP-E, 85L II, 100L macro, 500L IS II, EF-M 18-55; 1.4xIII, 2x III + 2xII extenders; 600EX-RT; EF-M--EF adaptor.
Former lenses include: 70-200L f/4 non-IS, 200L 2.8, 400L 5.6

jrista

  • Canon EF 400mm f/2.8L IS II
  • *********
  • Posts: 4814
  • EOL
    • View Profile
    • Nature Photography
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #129 on: August 05, 2014, 03:39:19 PM »
To suggest that if paper only can reproduce x range of latitude so there is no point in working the data in 16 bit as opposed to an 8 bit JPEG is, quite honestly astonishing

To be fair, he's saying that this is what he reckons Jrista is suggesting.

He can't separate information precision from dynamic range. It's all the same thing to him, so he is missing the point.

neuroanatomist

  • CR GEEK
  • ************
  • Posts: 15227
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #130 on: August 05, 2014, 03:48:19 PM »
... No. I have a LIMIT of 14 stops of dynamic range in a 14-bit file...
For each pixel yes.

So, your point is... ??

Hopefully not that downsampling a 14-bit file can result in >14-bits of DR.  That's dilbertthink.

Per downsampled pixel then yes. Assuming no noise, add for instance 4 14bit numbers and you get a 16bit number....

Dilbertthink, indeed. A 16-bit number doesn't mean 16-bits of real data.  If the original data were captured at 14-bits, anything over that means pulling new data from some nether orifice and pretending it's real.  Unless the pixels were differentially read out to begin with, in which case you could achieve higher DR...at the cost of real resolution.  Do you think that upsampling adds real spatial information, too?   ::)
EOS 1D X, EOS M, and lots of lenses
______________________________
Flickr | TDP Profile/Gear List

jrista

  • Canon EF 400mm f/2.8L IS II
  • *********
  • Posts: 4814
  • EOL
    • View Profile
    • Nature Photography
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #131 on: August 05, 2014, 03:51:30 PM »
But to return to the original post, I'd say the 645z looks like 'king of low light' at present, based on what I've gleaned online. I'd still rather have a 1Dx (or wait for the next Canon full frame body) for the system benefit, as Neuro says. But in terms of low noise at high ISO I reckon the new Pentax wins - as you'd expect, with a bigger sensor area.

I agree, with one additional qualifier: Assuming identical framing. Framed the same, absolutely! Bigger sensor always wins, as far as noise is concerned, regardless of pixel size or count.

msm

  • Rebel T5i
  • ****
  • Posts: 123
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #132 on: August 05, 2014, 04:06:45 PM »
... No. I have a LIMIT of 14 stops of dynamic range in a 14-bit file...
For each pixel yes.

So, your point is... ??

Hopefully not that downsampling a 14-bit file can result in >14-bits of DR.  That's dilbertthink.

Per downsampled pixel then yes. Assuming no noise, add for instance 4 14bit numbers and you get a 16bit number....

Dilbertthink, indeed. A 16-bit number doesn't mean 16-bits of real data.  If the original data were captured at 14-bits, anything over that means pulling new data from some nether orifice and pretending it's real.  Unless the pixels were differentially read out to begin with, in which case you could achieve higher DR...at the cost of real resolution.  Do you think that upsampling adds real spatial information, too?   ::)

Nope it's an oversimplified example to illustrate a simple point: When you downscale an image you gain additional information per pixel in the downscaled image (assuming you use a sensible downscaling algorithm). Each pixel in the downscaled image will use information from multiple pixels in the original.  In practice it is really trivial to observe, take a somewhat noisy image and downscale, which is more noisy and thus has less dynamic range: a pixel in the original image or a pixel in the downscaled image? Obviously you don't gain editing latitude for the image as a whole (you loose information in the process), but each pixel individually gain DR.
« Last Edit: August 05, 2014, 04:29:02 PM by msm »

canon rumors FORUM

Re: DxO reviews Sony A7s: king of low light photography?
« Reply #132 on: August 05, 2014, 04:06:45 PM »

neuroanatomist

  • CR GEEK
  • ************
  • Posts: 15227
    • View Profile
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #133 on: August 05, 2014, 05:06:14 PM »
Nope it's an oversimplified example to illustrate a simple point: When you downscale an image you gain additional information per pixel in the downscaled image (assuming you use a sensible downscaling algorithm). Each pixel in the downscaled image will use information from multiple pixels in the original.  In practice it is really trivial to observe, take a somewhat noisy image and downscale, which is more noisy and thus has less dynamic range: a pixel in the original image or a pixel in the downscaled image? Obviously you don't gain editing latitude for the image as a whole (you loose it), but each pixel individually gain DR.

Gains DR to a point.  There's a ceiling, and that ceiling is the maximum DR at capture.  Sorry, but your suggestion that combining four pixels with 14-bits at capture and getting 16-bits of real data is ludicrous.  But you're probably making dilbert happy as you waft the stench of DxO's BS (aka Biased Scores) around the forums.  Nothing to be proud of, IMO.
EOS 1D X, EOS M, and lots of lenses
______________________________
Flickr | TDP Profile/Gear List

jrista

  • Canon EF 400mm f/2.8L IS II
  • *********
  • Posts: 4814
  • EOL
    • View Profile
    • Nature Photography
Re: DxO reviews Sony A7s: king of low light photography?
« Reply #134 on: August 05, 2014, 06:24:40 PM »
... No. I have a LIMIT of 14 stops of dynamic range in a 14-bit file...
For each pixel yes.

So, your point is... ??

Hopefully not that downsampling a 14-bit file can result in >14-bits of DR.  That's dilbertthink.

Per downsampled pixel then yes. Assuming no noise, add for instance 4 14bit numbers and you get a 16bit number....

Dilbertthink, indeed. A 16-bit number doesn't mean 16-bits of real data.  If the original data were captured at 14-bits, anything over that means pulling new data from some nether orifice and pretending it's real.  Unless the pixels were differentially read out to begin with, in which case you could achieve higher DR...at the cost of real resolution.  Do you think that upsampling adds real spatial information, too?   ::)

Nope it's an oversimplified example to illustrate a simple point: When you downscale an image you gain additional information per pixel in the downscaled image (assuming you use a sensible downscaling algorithm). Each pixel in the downscaled image will use information from multiple pixels in the original.  In practice it is really trivial to observe, take a somewhat noisy image and downscale, which is more noisy and thus has less dynamic range: a pixel in the original image or a pixel in the downscaled image? Obviously you don't gain editing latitude for the image as a whole (you loose information in the process), but each pixel individually gain DR.

If your downsampling, yes. In the context of the prior discussion, I frequently referred to upsampling. You aren't going to be gaining any dynamic range by upsampling, doesn't matter what camera your using.

canon rumors FORUM

Re: DxO reviews Sony A7s: king of low light photography?
« Reply #134 on: August 05, 2014, 06:24:40 PM »