Approve the Cookies
This website uses cookies to improve your user experience. By using this site, you agree to our use of cookies and our Privacy Policy.
OK
Index  •   • New posts  •   • RTAT  •   • 'Best of'  •   • Gallery  •   • Gear  •   • Reviews
Guest
New posts  •   • RTAT  •   • 'Best of'  •   • Gallery  •   • Gear  •   • Reviews
Register to forums    Log in

 
FORUMS Canon Cameras, Lenses & Accessories Canon EOS Digital Cameras 
Thread started 28 Jul 2014 (Monday) 15:27
Search threadPrev/next
sponsored links
(this ad will go away when you log in as a registered member)

Why Canon, when Nikon...

 
Mornnb
Goldmember
1,646 posts
Gallery: 6 photos
Likes: 23
Joined Aug 2012
Location: Sydney
     
Aug 15, 2014 19:12 |  #316

The point is that the dynamic range difference is not the fabrication process or that the sony photodiodes let in more light, they do not. It's to do with the A/D converters and amplification systems.
Canon's sensors before you get to the A/D converter stage actually pick up more light and less noise than the Sony Exmor, largely due to the size and efficiency of the individual photodiodes, and this is likely also why Canon tends to have better high ISO performance. It's Canon's analog circuitry and A/D converters which introduce the noise at low ISO.


Canon 5D Mark III - Leica M240
EF 16-35mm F/4 IS L - EF 14mm f/2.8 L II - - EF 17mm TS-E L - EF 24-70mm f/2.8 L II - EF 70-200mm IS II f/2.8 L - Sigma 35mm f/1.4 Art - Sigma 85mm f/1.4 EX
Voigtlander 15mm III - 28mm Elmarit-M ASPH - 35mm f/1.4 Summilux-M FLE - 50mm f/1.4 Summilux-M ASPH
500px (external link)

  
  LOG IN TO REPLY
sponsored links
(this ad will go away when you log in as a registered member)
AJSJones
Goldmember
Avatar
2,647 posts
Gallery: 6 photos
Likes: 92
Joined Dec 2001
Location: California
     
Aug 15, 2014 19:40 |  #317

Charlie wrote in post #17098858 (external link)
on the flip side, you can should really ETTR even with the sony sensor. No need to be sloppy just for the sake of rubbing it in ;)

Indeed, it's true for any sensor when you wish to maximize captured DR. Setting the highest (desired) luminance values anywhere below saturation will automatically decrease the captured DR. If your highest value is captured 2 stops below saturation, you've just thrown away 2 stops of potential DR. To achieve the DR's mentioned above, you need to ETTR till you're almost saturating/fullwell/cl​ipping, even if you subsequently need to process to the left. (Don't base this exposure on the in-camera (jpeg) histogram unless you know what you are doing!:D)


My picture galleries (external link)

  
  LOG IN TO REPLY
Mornnb
Goldmember
1,646 posts
Gallery: 6 photos
Likes: 23
Joined Aug 2012
Location: Sydney
     
Aug 15, 2014 19:47 |  #318

AJSJones wrote in post #17098915 (external link)
To achieve the DR's mentioned above, you need to ETTR till you're almost saturating/fullwell/cl​ipping, even if you subsequently need to process to the left.

There are disadvantages to this however, because the camera loses a lot of tonal information as you get close to the highlights. One has to balance colour quality of the highlights with noise in the shadows... Generally raising the shadows produces much better results than dropping the highlights.


Canon 5D Mark III - Leica M240
EF 16-35mm F/4 IS L - EF 14mm f/2.8 L II - - EF 17mm TS-E L - EF 24-70mm f/2.8 L II - EF 70-200mm IS II f/2.8 L - Sigma 35mm f/1.4 Art - Sigma 85mm f/1.4 EX
Voigtlander 15mm III - 28mm Elmarit-M ASPH - 35mm f/1.4 Summilux-M FLE - 50mm f/1.4 Summilux-M ASPH
500px (external link)

  
  LOG IN TO REPLY
AJSJones
Goldmember
Avatar
2,647 posts
Gallery: 6 photos
Likes: 92
Joined Dec 2001
Location: California
     
Aug 15, 2014 19:54 |  #319

What? Another compromise in photography? Will they never end? :D


My picture galleries (external link)

  
  LOG IN TO REPLY
davesrose
Title Fairy still hasn't visited me!
3,809 posts
Likes: 555
Joined Apr 2007
Location: Atlanta, GA
     
Aug 15, 2014 21:30 |  #320

Mornnb wrote in post #17098928 (external link)
There are disadvantages to this however, because the camera loses a lot of tonal information as you get close to the highlights. One has to balance colour quality of the highlights with noise in the shadows... Generally raising the shadows produces much better results than dropping the highlights.

For effective dynamic range, it's more important to reach the saturation point. Let's take a look at bit depth. Suppose there's a perfect sensor/AD system that can fill all 16,384 tonal levels that a 14bpc system is capable of. If Sony ADC and Canon sensor tech mated and you had some perfect 0-16384 tonal range;)

If you threw out the first bit, you'd be throwing out only 2 tonal values. If you threw out the last bit, you'd be throwing out 8192 tonal values.

I think in this thread, we've debated the SNR of of the sensors quite a bit.:D I'll add my two cents about the actual brand trends. It's apparent Sony/Nikon sensors have better noise handling and resolution. If you factor the extra resolution, what advantage with Canon's sensor sensitivity has becomes less significant. It's certainly evident Sony is providing pressure for Canon to keep updating their DSLRs...and that's a good thing. We can either look at the glass is half empty or glass is half full. When I first got my 5Dc, the Nikon vs Canon camps were pretty much the opposite of what they are now (that Nikon had better AF/ Canon had better sensor). I don't see why their strengths won't keep switching back and forth: it's been going on for 70 years. I'm interested in seeing if Canon does introduce a sensor with the foveon type sensor patents they have. I suspect the September announcements will be for the 7DII, while the 5DmkIII will still be updated next year.


Canon 5D mk IV
EF 135mm 2.0L, EF 70-200mm 2.8L IS II, EF 24-70 2.8L II, EF 50mm 1.4, EF 100mm 2.8L Macro, EF 16-35mm 4L IS, Sigma 150-600mm C, 580EX, 600EX-RT, MeFoto Globetrotter tripod, grips, Black Rapid RS-7, CAMS plate and strap system, Lowepro Flipside 500 AW, and a few other things...
smugmug (external link)

  
  LOG IN TO REPLY
Mornnb
Goldmember
1,646 posts
Gallery: 6 photos
Likes: 23
Joined Aug 2012
Location: Sydney
     
Aug 16, 2014 16:43 |  #321

Ok. Now I pose this question.
Canon at the moment is making the best lenses. The 24-70mm II on a 22MP 5D Mark III is producing more detailed photos than Nikon's 24-70mm on their super high res 36MP sensor, this is entirely related to the resolving power of the lens. Same for Canon's 70-200mm offering. And the TS-E's, the 17mm and 24mm.
Why Nikon, when you can get a Sony A7R and adapt it to get that 36MP Exmor sensor behind Canon's generally superior lenses? Best of both worlds.


Canon 5D Mark III - Leica M240
EF 16-35mm F/4 IS L - EF 14mm f/2.8 L II - - EF 17mm TS-E L - EF 24-70mm f/2.8 L II - EF 70-200mm IS II f/2.8 L - Sigma 35mm f/1.4 Art - Sigma 85mm f/1.4 EX
Voigtlander 15mm III - 28mm Elmarit-M ASPH - 35mm f/1.4 Summilux-M FLE - 50mm f/1.4 Summilux-M ASPH
500px (external link)

  
  LOG IN TO REPLY
Glenn ­ NK
Goldmember
Avatar
4,630 posts
Likes: 2
Joined Oct 2006
Location: Victoria, BC
     
Aug 16, 2014 17:20 |  #322

Mornnb wrote in post #17098928 (external link)
There are disadvantages to this however, because the camera loses a lot of tonal information as you get close to the highlights. One has to balance colour quality of the highlights with noise in the shadows... Generally raising the shadows produces much better results than dropping the highlights.

Actually it's the other way around. The brightest stop carries twice the level of information of the next brightest stop and so on.

Tonal information is only lost when one or more of the channels are clipped.


When did voluptuous become voluminous?

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 05:28 |  #323

Charlie wrote in post #17092107 (external link)
4 stops... that's 200, 400, 800, 1600.... that's equivalent to ISO 1600 for a landscape shot. I wouldnt be comfortable with that.

No it isn't.

ISO 1600 means the full image has been amplified 4 stops.

Having a sensor with the extra DR from the start means you have no amplification involved - any noise from boosting the shadows are just affecting the shadows but not the other parts.

Next thing - if Hogloff is using 30 second exposures - what option would he have if taking multiple exposures? Doing one exposure with 240 or 360 seconds? Or doing one with 30 seconds at ISO 100 and then another with 30 seconds at ISO 1600?

You think taking the shadows from a separate 30 seconds exposure at ISO 1600 will be better?


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 05:34 |  #324

davesrose wrote in post #17092325 (external link)
Please read my posts: I've stated quite a few times in this thread that bumping up the shadows +4 stops and upscaling the 5D shots up to 36MP is not an apples to apples comparison.

Note that if you print photos, the print will perform a similar normalizing scaling. In the end, the photos are captured with same-size sensor and are expected to be printed to same size paper. So the 36MP photo will print with smaller pixels where the noise of each pixel will be less seen.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 05:48 |  #325

davesrose wrote in post #17092568 (external link)
Fred's article is quite clear: the 5D image was scaled up to 36MP while the D800 image wasn't interpolated.

DR is the total tones of contrast from the largest intensity vs blackest black. On the sensor level, it is the saturation point vs SNR...again, white to steps of black.

No. DR is the difference between the noise hiss in your loudspeakers when you don't play musing to the highest volume you can play before they clip.

While total tones is how small steps the volume can be increased in.

Your Bluray or DVD may contain 24-bit audio while your CD contains 16-bit audio. A difference in the quality of weak sounds. But your stereo will still hiss it's noise at the same level and you can still play both DVD and CD to the same level before they clip.

The issue with the CD is that it handles weak sounds badly.
Just as a camera sensor have issues with shadows.

So DR is not number of steps from white to black. But number of stops from white to black. While tonality is number of steps within a stop.

It can be quite academic to try to compare the difference in DR between the sensor DR vs the recorded RAW DR. In the end of the day, though, the highest DR of a digital image is up to 14 stops of light: 14bpc file size. That's the maximum DR of any DSLR recorded medium. The math is quite specific: it's a log that goes like this in tonality: 2,4,8,16,32,64,128,256​,512,1024,2048,4096,81​92.

But that is irrelevant if the last stops are noise. You can't record an orchestra and be happy with it if you have 100 people busy talking while you record the sound. Having a 14-bit ADC doesn't mean you get 14-bit quality.

If the Sony sensors are able to fill more values in all stops, then there may be some more wiggle room with the darker blacks. However, in my real world post processing, I have yet to push my shadows up 4+stops.

My 5D2 doesn't require me to be even close to 4 stops before the image gets unusable. It's more like 9-something stops I get. So hardly any push at all gives banding.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 06:08 |  #326

davesrose wrote in post #17097522 (external link)
No I'm not. My math is quite apparent: scaling a 22MP image up to 36MP, you're randomly adding 40% more pixels. You know, I work with graphic designers and printers. They try to avoid blowing up an image at all costs. A computer algorithm tries to average X number of pixels and then adds more of a similar color. If it averages one pixel of noise, it may amplify it quite a bit.

I've already stated that the noise handling of the D800 is apparently better then the 5D (both a factor of sensor AND resolution). You can keep showing examples of non-HDR photographs with noise in the shadows. That doesn't address what my point is: that I have yet to see examples of how well the Sony sensors handle clipping....one of the primary factors of DR.

You know, the 36MP image is created from smaller pixels. Scale it down to 22MP and it will become better.

Upscaling doesn't just average a single pixel of noise. You have a kernel that decides the weights used among neighbour pixels.

And in the end, you have three options:
- view on monitor. Then both images gets scaled down. But the 36MP image gets scaled down more. So it stays ahead.
- print small. Same as above.
- print large. Then both images needs to be upscaled. But the 36MP image gets less upscaled. So it stays ahead.

It doesn't matter how you turn this, the 22MP image of the 5D3 is way worse than the 36MP image of the D800E when it comes to the shadow parts of the image. It's just a question if you will allow yourself to be openminded enough to accept it.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 06:23 |  #327

davesrose wrote in post #17097556 (external link)
When you're pushing up shadows, you're looking more at noise handling.

Two different things happens.

1) The noise gets more visible.

2) Things hidden below the noise floor with one sensor shows up with another sensor.

Note that your computer-generated images does not contain noise, and have no noise floor. The only noise involved is the noise functions - like Perlin - that are used to create procedural structure - color, intensity, perturbation, ... - to surfaces.

I would argue that ML dual ISO is not adding DR. It's splitting the difference between two ISOs to get better sensitivity throughout the inherent contrast range of the image. It is not raising the saturation point of the sensel though. This is a fine technical point to be sure;)

ML is a work-around for a kind of "defect" in Canon sensor readout. The ML functionality wouldn't work these wonders with the Sony Exmor sensor.

The Canon has issues with banding destroying the last bits of the AD conversion.
Increasing ISO increases the amplification of the signal before it gets digitized. So while the amplification adds noise (as we know from higher ISO), it moves the shadow data above the "damaged" lowest bits.

So the goal with ML isn't to get more tonality steps in the shadow stops, but to get the shadow stops away from the low-bits readout failure.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 06:30 |  #328

davesrose wrote in post #17097650 (external link)
I have yet to see any upscaling software approach the detail of a source that's native to that resolution. At Siggraph, I have seen developmental software that does a fair job with blowing up an image to not be as fuzzy. Professional printers still prefer native resolution sources. In the video realm, plenty of folks where opting in buying expensive video upscalers to try to get HD quality out of DVD. Even with the best of them, there's a night and day difference between SD vs HD resolutions on a HDTV.

People want BD or 4K video instead of DVD video or VHS video. But that is irrelevant to this debate. 4k video can resolve more features than DVD. But you can't compare the noise of one pixel from a 4k video with one pixel from a DVD video. And you can't compare one pixel from the 36MP D800E with one pixel from a 22MP 5D3. You need to scale! You need to scale! You NEED to scale!

Scale up or scale down doesn't matter. Scale up and you lose quality in the 22->36MP conversion. Scale down and you reduce noise in the 36MP->22MP conversion. In the end, the D800E still wins. Wins, wins, wins. Despite starting with almost twice as large pixels on the sensor, it still have more noise per pixel in the shadows of a ISO 100 capture. The larger pixels can't help what the ADC conversion destroys.

The only important thing here is that when you compare, then the same features captured in the photo needs to fill the same area of your monitor or of a print. So scaling MUST be performed!


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 06:36 |  #329

AJSJones wrote in post #17097829 (external link)
But DR is DEFINED and MEASURED by mathematically evaluating the signal to noise ratio in the darkest shadows and calculating, from the data coming from each sensel, the luminance value at which signal rises above the noise and can therefore present real image information. You cannot even specify a DR without first measuring the noise level in the shadows. If you do not use the terms this way, please define sensor DR in terms of how you do see it.

This is also the danger with the mathematical view.

The "normal" way would be to measure the average output of the sensor when capturing no light. Then slowly increase the light and compute the average output until that average can be seen to increase. That would represent the darkest level the sensor can detect.

The problem here is that this gives Canon sensors quite good DR figures, because the actual sensitivity of the sensors are good - as we know from high-ISO use.

It's just that normal dynamic-range measurements are based on the noise floor being random noise so a view of this noise floor should not show any pattern. This fails badly with a number of Canon sensors, where there are noise patterns moving several stops into the captured data range. So the usable DR ends up being much worse than the measured DR.

Here (external link) is a complete discussion of the concepts as used in describing and measuring sensor performance that are used in the digital photography world:D In particular see Fig 8a for noise sources, each of which affect the sensor DR. Here (external link) is a thorough analysis of the 5D3 for a blow by blow analysis of banding and improvement over 5D2.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
pwm2
"Sorry for being a noob"
Avatar
8,626 posts
Likes: 3
Joined May 2007
Location: Sweden
     
Aug 17, 2014 06:45 |  #330

davesrose wrote in post #17097860 (external link)
DR is defined by BOTH the saturation point of the sensel and the noise floor. The Sony sensors may have a better noise floor (whether with sensor and/or A/D converter). I still have yet to see clear data on how they differ at the saturation point. In absolute terms, DR in imaging is about capturing the amount of luminance that's present in a scene. Digital cameras are improving, but still have a ways to go to fully capture the DR the human eye can percieve: within in single shot.

You normally measure the noise floor relative to the saturation point, and then present in dB, stops or similar. Then you may play with software in the camera to decide how much dynamic range to reserve above your 18% grayscale level. But what is debated here is the total usable range from clipping and down to the noise floor (and way above the average noise floor when considering banding noise).

Next thing - don't believe our eyes can see so huge dynamic range. Lots of our dynamic range comes from our eyes adapting as we move our view to different parts of the scene. So the iris changes size and chemical processes changes the sensitivity. So lots of stops are just virtual HDR from multiple exposures and our brain combining it into a "make believe" huge dynamic range.

Same thing with our ears - we have muscles that can tighten the ear drum to adjust the sensitivity. The biological sensors doesn't have magical super-range.


5DMk2 + BG-E6 | 40D + BG-E2N | 350D + BG-E3 + RC-1 | Elan 7E | Minolta Dimage 7U | (Gear thread)
10-22 | 16-35/2.8 L II | 20-35 | 24-105 L IS | 28-135 IS | 40/2.8 | 50/1.8 II | 70-200/2.8 L IS | 100/2.8 L IS | 100-400 L IS | Sigma 18-200DC
Speedlite 420EZ | Speedlite 580EX | EF 1.4x II | EF 2x II

  
  LOG IN TO REPLY
sponsored links
(this ad will go away when you log in as a registered member)

57,177 views & 0 likes for this thread
Why Canon, when Nikon...
FORUMS Canon Cameras, Lenses & Accessories Canon EOS Digital Cameras 
AAA
x 1600
y 1600

Jump to forum...   •  Rules   •  Index   •  New posts   •  RTAT   •  'Best of'   •  Gallery   •  Gear   •  Reviews   •  Member list   •  Polls   •  Image rules   •  Search   •  Password reset

Not a member yet?
Register to forums
Registered members may log in to forums and access all the features: full search, image upload, follow forums, own gear list and ratings, likes, more forums, private messaging, thread follow, notifications, own gallery, all settings, view hosted photos, own reviews, see more and do more... and all is free. Don't be a stranger - register now and start posting!


COOKIES DISCLAIMER: This website uses cookies to improve your user experience. By using this site, you agree to our use of cookies and to our privacy policy.
Privacy policy and cookie usage info.


POWERED BY AMASS forum software 2.1forum software
version 2.1 /
code and design
by Pekka Saarinen ©
for photography-on-the.net

Latest registered member is atomanth
795 guests, 216 members online
Simultaneous users record so far is 15144, that happened on Nov 22, 2018

Photography-on-the.net Digital Photography Forums is the website for photographers and all who love great photos, camera and post processing techniques, gear talk, discussion and sharing. Professionals, hobbyists, newbies and those who don't even own a camera -- all are welcome regardless of skill, favourite brand, gear, gender or age. Registering and usage is free.