November 26, 2014, 05:35:54 AM

Author Topic: 1DX and 5D3 RAW files  (Read 42060 times)

Northstar

  • 1D X
  • *******
  • Posts: 1503
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #30 on: May 21, 2013, 12:03:49 PM »
Certainly, let me use an analogy which might help.  Lets say you are recording a violinist onto a tape recorder using a high quality tape.  When the violinist plays softly you have to amplify the signal to increase the volume.  As you do so you also increase the noise from the tape.  Maybe the violinist is playing so soft that their sound level falls to the point where it's difficult to tell if it's a violin or tape hiss (noise).  Depending on which tape you use (high quality or cheap quality) and what equipment you are using (radio shack tape deck or Yamaha digital tape deck)  you will have more or less noise when recording that violinist at the same recording level.

This also holds true for image sensors.  When the brightness is low such as in a shadow or low light situation the photon levels are so low that they are mixed with noise.  Depending on the sensor and supporting backend electronics there may be more or less noise.  Amplification, same as with that tape deck, amplifies not only the photons but also the noise level.  Higher quality components (better sensor, larger pixels, better amplifiers) can all contribute to less noise compared to the signal and more DR in the shadows.

You absolutely cannot compare sensor IQ without also considering the supporting electronics used in processing the signal.  Even if the cameras use the exact same sensor, different electronics on the back end will affect the IQ.  There are a lot of different points to consider.  What causes IQ differences in 5D3 and 1DX may be different in other models depending on the generation of support electronics also used.

Remember also that the image as it hits the sensor is still analog.  It's not converted to digital until after any amplification has already occurred.

Failure of the amplifier to pull photons from the background noise.

Please explain what you mean by this statement.

this makes sense eastwind...and i think somewhere in the supporting electronics/advanced circuits/design lies the specific answer to bdun's question.

one question though...if "better quality components/circuitry" supporting each photosite is so critical, then why does my old nikon d7000 (for $1000) have significantly better DR under iso 400 compared to a $6700 1dx when the "supporting parts" are surely of lesser quality.....does it all fall on design at that point?
Sport Shooter

1dX and 5d3... 24-70 2.8ii, 70-200 2.8ii, 1.4xiii and 2xiii, 85, 40mm, 300 2.8L IS....430ex

canon rumors FORUM

Re: 1DX and 5D3 RAW files
« Reply #30 on: May 21, 2013, 12:03:49 PM »

East Wind Photography

  • 1D Mark IV
  • ******
  • Posts: 776
  • EWP
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #31 on: May 21, 2013, 12:26:24 PM »
Oh man, there are a lot of moving targets to make that comparison.  ;)  It's likely a combination of factors.  However the best solution to better IQ ultimately lies in better light sensitivity per pixel and less background noise.  Likely a better sensor design from the start requiring less photon amplification and processing after the image is acquired. 

Often a solution may be better suited for one particular situation but to bring it to market has to work in a variety of situations and someone makes a decision to live with less quality in order to support a wider range of operating conditions.  We see that in many different products including cars where we can get incredible gas mileage but you can't deviate from sea level.  Drive up a mountain and your car stalls out.  They have to be made to withstand anything the consumer can throw at it and keep working.  the nikon may very well have been designed with ISO 400 in mind to keep costs down whereas the 1DX may have been designed with higher ISO in mind.

Certainly, let me use an analogy which might help.  Lets say you are recording a violinist onto a tape recorder using a high quality tape.  When the violinist plays softly you have to amplify the signal to increase the volume.  As you do so you also increase the noise from the tape.  Maybe the violinist is playing so soft that their sound level falls to the point where it's difficult to tell if it's a violin or tape hiss (noise).  Depending on which tape you use (high quality or cheap quality) and what equipment you are using (radio shack tape deck or Yamaha digital tape deck)  you will have more or less noise when recording that violinist at the same recording level.

This also holds true for image sensors.  When the brightness is low such as in a shadow or low light situation the photon levels are so low that they are mixed with noise.  Depending on the sensor and supporting backend electronics there may be more or less noise.  Amplification, same as with that tape deck, amplifies not only the photons but also the noise level.  Higher quality components (better sensor, larger pixels, better amplifiers) can all contribute to less noise compared to the signal and more DR in the shadows.

You absolutely cannot compare sensor IQ without also considering the supporting electronics used in processing the signal.  Even if the cameras use the exact same sensor, different electronics on the back end will affect the IQ.  There are a lot of different points to consider.  What causes IQ differences in 5D3 and 1DX may be different in other models depending on the generation of support electronics also used.

Remember also that the image as it hits the sensor is still analog.  It's not converted to digital until after any amplification has already occurred.

Failure of the amplifier to pull photons from the background noise.

Please explain what you mean by this statement.

this makes sense eastwind...and i think somewhere in the supporting electronics/advanced circuits/design lies the specific answer to bdun's question.

one question though...if "better quality components/circuitry" supporting each photosite is so critical, then why does my old nikon d7000 (for $1000) have significantly better DR under iso 400 compared to a $6700 1dx when the "supporting parts" are surely of lesser quality.....does it all fall on design at that point?

Quasimodo

  • 1D Mark IV
  • ******
  • Posts: 963
  • Easily intrigued :)
    • View Profile
    • 500px.com
Re: 1DX and 5D3 RAW files
« Reply #32 on: May 21, 2013, 12:30:24 PM »
For those interested, here is from someone who reverse engineered the CR2 raw's:

From a data container perspective there's a lot to learn.

Haven't had the time to go through everything to answer OP's question whether there really is more raw data (read: more POWER ;-)) in the 1Dx raws or not.

Please enjoy the read:
http://lclevy.free.fr/cr2/

Thanks! Looking forward to read it.
1Dx, 5x600 EX RT, ST-E3Canon:16-35L II,  24-105L , 70-200L IS II, 135L, 100L, 2x III TC, EF 25II, 40 F2.8 STM, Sigma 35 F1.4 Art, Sigma 50 F1.4 Art, Sigma 85 F1.4, Sigma 150-500.
Canon A-1, 199A, FD: 24/2.8, 35/2.0, 100/2.8, Vivitar 400/5.6 Mamiya RZ67 pro ii, 50,110,180
www.500px.com/gerhard1972

Meh

  • 5D Mark III
  • ******
  • Posts: 701
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #33 on: May 21, 2013, 12:41:20 PM »
Certainly, let me use an analogy which might help.  Lets say you are recording a violinist onto a tape recorder using a high quality tape.  When the violinist plays softly you have to amplify the signal to increase the volume.  As you do so you also increase the noise from the tape.  Maybe the violinist is playing so soft that their sound level falls to the point where it's difficult to tell if it's a violin or tape hiss (noise).  Depending on which tape you use (high quality or cheap quality) and what equipment you are using (radio shack tape deck or Yamaha digital tape deck)  you will have more or less noise when recording that violinist at the same recording level.

This also holds true for image sensors.  When the brightness is low such as in a shadow or low light situation the photon levels are so low that they are mixed with noise.  Depending on the sensor and supporting backend electronics there may be more or less noise.  Amplification, same as with that tape deck, amplifies not only the photons but also the noise level.  Higher quality components (better sensor, larger pixels, better amplifiers) can all contribute to less noise compared to the signal and more DR in the shadows.

You absolutely cannot compare sensor IQ without also considering the supporting electronics used in processing the signal.  Even if the cameras use the exact same sensor, different electronics on the back end will affect the IQ.  There are a lot of different points to consider.  What causes IQ differences in 5D3 and 1DX may be different in other models depending on the generation of support electronics also used.

Remember also that the image as it hits the sensor is still analog.  It's not converted to digital until after any amplification has already occurred.

Failure of the amplifier to pull photons from the background noise.

Please explain what you mean by this statement.

Not even close to explaining your statement which was (and I'll add some emphasis)... "FAILURE of the amplifier to PULL photons from the background noise".  All you've done is tell us with your analogy that for low signal levels the SNR would be high.  Please enlighten us how even the best amplifier in the world, even a hypothetically perfect amplifier, will "PULL photons from the background noise".

neuroanatomist

  • CR GEEK
  • **********
  • Posts: 14962
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #34 on: May 21, 2013, 12:52:03 PM »
Failure of the amplifier to pull photons from the background noise.
Not even close to explaining your statement which was (and I'll add some emphasis)... "FAILURE of the amplifier to PULL photons from the background noise".  All you've done is tell us with your analogy that for low signal levels the SNR would be high.  Please enlighten us how even the best amplifier in the world, even a hypothetically perfect amplifier, will "PULL photons from the background noise".

But wait, he stated,

I  am quite done with this.

When an individual's metacognition is insufficiently developed to understand when s/he has moved on from a concept, others may question that individuals understanding of more complex issues.
EOS 1D X, EOS M, and lots of lenses
______________________________
Flickr | TDP Profile/Gear List

East Wind Photography

  • 1D Mark IV
  • ******
  • Posts: 776
  • EWP
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #35 on: May 21, 2013, 06:29:01 PM »
No actually trying to say that a poor (cheap) amplifier can introduce more noise than a higher quality one such that when it's amplified, the photons can be pulled further from the noise.  All of the noise you see in an image is not due to what you are photographing, it's noise from the amplifiers within the chip including that introduced by heat.

The best amplifier in the world will not produce any noise and as such will be able to pull the photons perfectly from a noiseless background.

So it's likely the higher end cameras have higher quality amplifiers (equals less noise) and in the case of the 1DX has a chip that is capable of recording more photons per pixel.  Probably one of the reasons they stayed with the 18MP sensor so they could get better high ISO performance.

Certainly, let me use an analogy which might help.  Lets say you are recording a violinist onto a tape recorder using a high quality tape.  When the violinist plays softly you have to amplify the signal to increase the volume.  As you do so you also increase the noise from the tape.  Maybe the violinist is playing so soft that their sound level falls to the point where it's difficult to tell if it's a violin or tape hiss (noise).  Depending on which tape you use (high quality or cheap quality) and what equipment you are using (radio shack tape deck or Yamaha digital tape deck)  you will have more or less noise when recording that violinist at the same recording level.

This also holds true for image sensors.  When the brightness is low such as in a shadow or low light situation the photon levels are so low that they are mixed with noise.  Depending on the sensor and supporting backend electronics there may be more or less noise.  Amplification, same as with that tape deck, amplifies not only the photons but also the noise level.  Higher quality components (better sensor, larger pixels, better amplifiers) can all contribute to less noise compared to the signal and more DR in the shadows.

You absolutely cannot compare sensor IQ without also considering the supporting electronics used in processing the signal.  Even if the cameras use the exact same sensor, different electronics on the back end will affect the IQ.  There are a lot of different points to consider.  What causes IQ differences in 5D3 and 1DX may be different in other models depending on the generation of support electronics also used.

Remember also that the image as it hits the sensor is still analog.  It's not converted to digital until after any amplification has already occurred.

Failure of the amplifier to pull photons from the background noise.

Please explain what you mean by this statement.

Not even close to explaining your statement which was (and I'll add some emphasis)... "FAILURE of the amplifier to PULL photons from the background noise".  All you've done is tell us with your analogy that for low signal levels the SNR would be high.  Please enlighten us how even the best amplifier in the world, even a hypothetically perfect amplifier, will "PULL photons from the background noise".

East Wind Photography

  • 1D Mark IV
  • ******
  • Posts: 776
  • EWP
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #36 on: May 21, 2013, 06:30:47 PM »
I was done with the insults  ;)  I limit myself to one per month whether someone deserves it or not!

Failure of the amplifier to pull photons from the background noise.
Not even close to explaining your statement which was (and I'll add some emphasis)... "FAILURE of the amplifier to PULL photons from the background noise".  All you've done is tell us with your analogy that for low signal levels the SNR would be high.  Please enlighten us how even the best amplifier in the world, even a hypothetically perfect amplifier, will "PULL photons from the background noise".

But wait, he stated,

I  am quite done with this.

When an individual's metacognition is insufficiently developed to understand when s/he has moved on from a concept, others may question that individuals understanding of more complex issues.

canon rumors FORUM

Re: 1DX and 5D3 RAW files
« Reply #36 on: May 21, 2013, 06:30:47 PM »

Daniel Flather

  • 1D Mark IV
  • ******
  • Posts: 871
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #37 on: May 21, 2013, 06:38:03 PM »
I'm sure someone knows, just no one on these forums.  ;)

"This is off our first record, most people don't own it"
| 5D3 | 8-15L | 24L II | 35L | 50L | 85L II | 100/2.8 | 200/2L | EOS M | 22 STM |

qwerty

  • PowerShot G1 X II
  • ***
  • Posts: 49
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #38 on: May 21, 2013, 09:28:46 PM »
Which begs the question... http://en.wikipedia.org/wiki/Begging_the_question

Seriously though, can anyone give an example of an image, shot with identical exposure and lens of the same scene at the same time, and with the same post-processing, that illustrates the OP's point (1)?

DxO (don't start...) shows the 1dx having a slight but real advantage in terms of both resolution-normalized and per-pixel performance, but the levels are less than what I personally would be able to notice without analysis or a very contrived example.

Have you considered the possibility that people shooting with a 1dx are just more adept (on average) at getting the most out of their gear?  A priori, I think this is the most likely explanation for what the OP observes.  A pro with a 1dx will take a better shot and post-process it better than a doofus with a 5d iii like me :) The reason you see more in the processed image is because there is more there to start with. I expect if you switched the cameras so that the pro had the 5d iii, you would find that the 5d iii magically started producing much better raw images (that could be pushed further).



(1) Note that, according to official camera discussion forum rules, if the obvious conclusion drawn from the illustrative image disagrees with anyone's biases, any conclusions drawn from it will not be considered acceptable unless the artistic merit of said image would make everyone from an Old Master to the most fringe avante-garde artist weep tears of ecstasy.
« Last Edit: May 21, 2013, 11:28:24 PM by qwerty »

neuroanatomist

  • CR GEEK
  • **********
  • Posts: 14962
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #39 on: May 21, 2013, 11:08:30 PM »
Have you considered the possibility that people shooting with a 1dx are just more adept (on average) at getting the most out of their gear?  A priori, I think this is the most likely explanation for what the OP observes.  A pro with a 1dx will take a better shot and post-process it better than a doofus with a 5d iii like me :) The reason you see more in the processed image is because there is more there to start with. I expect if you switched the cameras so that the pro had the 5d iii, you would find that the 5d iii magically started producing much better raw images (that could be pushed further).

Well, the OP has a 5DIII and a pair of 1D X bodies, and is therefore speaking from personal experience of shooting similar scenes (gymnasium sports with f/2.8 supertele primes, needing fast shutter speeds in the typically poor lighting of such venues) with both cameras personally.  He's processing his own RAW files from the two cameras, and the needs of the shots often demand careful work in post (if you've shot basketball, volleyball, etc., in a gym, you're familiar with the need for high ISO, the crappy color of the inadequate-for-photography lighting, etc.). He's commented in several threads on the difference in how far he can push files from the two cameras.

So in this case, I don't think your explanation is the likely one.
EOS 1D X, EOS M, and lots of lenses
______________________________
Flickr | TDP Profile/Gear List

qwerty

  • PowerShot G1 X II
  • ***
  • Posts: 49
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #40 on: May 21, 2013, 11:30:41 PM »
Have you considered the possibility that people shooting with a 1dx are just more adept (on average) at getting the most out of their gear?  A priori, I think this is the most likely explanation for what the OP observes.  A pro with a 1dx will take a better shot and post-process it better than a doofus with a 5d iii like me :) The reason you see more in the processed image is because there is more there to start with. I expect if you switched the cameras so that the pro had the 5d iii, you would find that the 5d iii magically started producing much better raw images (that could be pushed further).

Well, the OP has a 5DIII and a pair of 1D X bodies, and is therefore speaking from personal experience of shooting similar scenes (gymnasium sports with f/2.8 supertele primes, needing fast shutter speeds in the typically poor lighting of such venues) with both cameras personally.  He's processing his own RAW files from the two cameras, and the needs of the shots often demand careful work in post (if you've shot basketball, volleyball, etc., in a gym, you're familiar with the need for high ISO, the crappy color of the inadequate-for-photography lighting, etc.). He's commented in several threads on the difference in how far he can push files from the two cameras.

So in this case, I don't think your explanation is the likely one.

I stand corrected. My bad for not noticing his signature. I have not seen his previous posts, but will take a look.

privatebydesign

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2780
  • Ermintrude says "moo"
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #41 on: May 21, 2013, 11:44:04 PM »
The majority of my experiences are with my 1Ds MkIII's and customer (I print for other photographers) 5D MkII files, and I feel that this is an equally valid, if not more so (and the OP agreed with me), comparison due to the aforementioned same sensor but different output.

I know from hands on experience that 5D MkII RAW files are not as malleable as 1Ds MkIII RAW files.

This is not a question of DR, noise, banding, FPN or any one "issue" it is a comment on the actual finished output of the image file, the 1 series cameras just have more flexibility in them when worked by the same person.

Here is an example of a torture test where a 5D MkII would get very different results, unfortunately I don't have direct comparison shots but try this with a 5D MkII.

Image is a 1Ds MkIII shot at 1600iso and underexposed by two stops. This was then lifted in post by two stops to give an effective 6400iso at an EV of 10ish, or just past sunset light. I did noise reduction to a level many would feel excessive but I wanted to see what detail was left, as you can see from the 100% crop the individual hairs are still rendered well. Note there is zero banding, no noise and no FPN. Detail and DR have been badly impacted but the image is 100% usable at a variety of sizes.
Never look down on anybody, unless you are helping them up.

bdunbar79

  • Canon EF 300mm f/2.8L IS II
  • *******
  • Posts: 2601
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #42 on: May 22, 2013, 12:18:48 PM »
Thanks everyone for the continued discussions.  Really interesting stuff so far.  I've been thinking about this and I don't think it is a DR issue.  It seems, like pbd is saying, the colors, contrast, other things.  And I owned a 1Ds3 and I like working with those RAW files more than the 5D2 as well.  Not that the 5D2 didn't produce great images (it did!) but it was the "maleability" of the RAW files as pbd says.

Good news is that I think I do have my RAW files from both the 1Ds3 and 5D3 on a trip last summer.  I am going to look at those. 

Neuro's right.  I didn't take my 5D3 into high ISO situations not because it's not good at high ISO, it's just that objectively the skin tones were much easier to correct with the 1Dx than the 5D3, noise aside.  And just going by the sensor scores/performance reviews, this doesn't make a whole lot of sense.
2 x 1DX
Big Ten, GLIAC, NCAC

canon rumors FORUM

Re: 1DX and 5D3 RAW files
« Reply #42 on: May 22, 2013, 12:18:48 PM »

vscd

  • EOS M2
  • ****
  • Posts: 185
  • 5DC
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #43 on: May 22, 2013, 01:44:43 PM »
To bring some new points to the discussion... the 1Ds Mark III and the 5D Mark II have both the same sensor, but a different CPU. The Digic III and Digic IV may differ a lot in signalprocessing. Which doesn't mean that one or another is better. The same goes to 1DX and 5D Mark III, of course they both share a Digic V+, but the 1DX has two of em.

As you all know from the Intel or AMD Market, a CPU has a lot of steps and internal improvement from one series to another, even if they share the same name (f.e. Core2).

Last but not least, the internal firmwarealghorithm may differ from cam to cam. So somethimes the raws are getting better with a firmwareupdate, no one knows why ;)
5DC, 24-85, 85 1.2L II, 80-200 2.8L, 100 2.8L IS, 14 2.8, 35 1.4, 75-300 IS, 40STM

RGF

  • 1D X
  • *******
  • Posts: 1300
  • How you relate to the issue, is the issue.
    • View Profile
Re: 1DX and 5D3 RAW files
« Reply #44 on: May 23, 2013, 10:06:45 AM »
Certainly, let me use an analogy which might help.  Lets say you are recording a violinist onto a tape recorder using a high quality tape.  When the violinist plays softly you have to amplify the signal to increase the volume.  As you do so you also increase the noise from the tape.  Maybe the violinist is playing so soft that their sound level falls to the point where it's difficult to tell if it's a violin or tape hiss (noise).  Depending on which tape you use (high quality or cheap quality) and what equipment you are using (radio shack tape deck or Yamaha digital tape deck)  you will have more or less noise when recording that violinist at the same recording level.

This also holds true for image sensors.  When the brightness is low such as in a shadow or low light situation the photon levels are so low that they are mixed with noise.  Depending on the sensor and supporting backend electronics there may be more or less noise.  Amplification, same as with that tape deck, amplifies not only the photons but also the noise level.  Higher quality components (better sensor, larger pixels, better amplifiers) can all contribute to less noise compared to the signal and more DR in the shadows.

You absolutely cannot compare sensor IQ without also considering the supporting electronics used in processing the signal.  Even if the cameras use the exact same sensor, different electronics on the back end will affect the IQ.  There are a lot of different points to consider.  What causes IQ differences in 5D3 and 1DX may be different in other models depending on the generation of support electronics also used.

Remember also that the image as it hits the sensor is still analog.  It's not converted to digital until after any amplification has already occurred.

Failure of the amplifier to pull photons from the background noise.

Please explain what you mean by this statement.

EWP

Let me pose the following, it is a question that I have wondering about for a while.

If I take the native signal from the sensor as capture let's it would be a proper exposure at ISO 800.  At ISO 400 the electronics would amplify the signal "half" as much.   At ISO 100 the amplification would be an "eighth.  Let me take the image from ISO 800 and from ISO 100.  I then increase the exposure of the ISO 100 image by 3 stops in PS.

Naively they should give very similar results, but electronic gain gives a better picture.  Why?  Does the electronic amplification provide a more continuous signal?

And to take this one more step further, how does statistics of counting random events which is described by a poison distribution enter into the discussion (or is the number of photons generally so large that signal noise except at very high ISO values does not matter)?

canon rumors FORUM

Re: 1DX and 5D3 RAW files
« Reply #44 on: May 23, 2013, 10:06:45 AM »