July 21, 2018, 11:51:29 AM

Author Topic: Canon's First Full Frame Mirrorless Won't Have the Exact Same Sensors as the EOS 5D Mark IV. [CR2]  (Read 10324 times)

3kramd5

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2267

Patents don’t necessarily translate into production, but didn’t they already essentially catch up (5D4)?

they have done too many just to be pure research. if it was one patent sure. however there's been around 10-20 at least to do with stacked sensors in the last 6 months.

they are still around .25 to 1 EV behind EXMOR depending on how you measure DR.

while they are closer, there's still more than could be done, and also there's the possibility of them exceeding what Sony could put out of their sensors as well.  it's not as if Canon couldn't leapfrog the current technology.

Really there is only one way to measure DR. Some people arbitrarily define the lower SNR limit.

But DR is just one measure of sensor performance. Canon manages to be within  1/4eV or whatever and provide tracking via dual diode pixels across the frame. One could argue it’s vying with the A9 as the highest performing 135 format sensor in a consumer camera.

actually then the only way to measure is by DXO's measurements, which show a 1EV difference.

I should rephrase. There are multiple ways to measure it, but one consistent way to define it. Regardless of whether one dislikes SNR=1 as a basis, there are multiple dimensions of sensor performance, including DR but certainly not limited to it. Increasingly, sensor based functionality is increasing, so what a sensor facilitates becomes important.

canon rumors FORUM


rrcphoto

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2373

Patents don’t necessarily translate into production, but didn’t they already essentially catch up (5D4)?

they have done too many just to be pure research. if it was one patent sure. however there's been around 10-20 at least to do with stacked sensors in the last 6 months.

they are still around .25 to 1 EV behind EXMOR depending on how you measure DR.

while they are closer, there's still more than could be done, and also there's the possibility of them exceeding what Sony could put out of their sensors as well.  it's not as if Canon couldn't leapfrog the current technology.

Really there is only one way to measure DR. Some people arbitrarily define the lower SNR limit.

But DR is just one measure of sensor performance. Canon manages to be within  1/4eV or whatever and provide tracking via dual diode pixels across the frame. One could argue it’s vying with the A9 as the highest performing 135 format sensor in a consumer camera.

actually then the only way to measure is by DXO's measurements, which show a 1EV difference.

I should rephrase. There are multiple ways to measure it, but one consistent way to define it.

as much as i hate DXO's single number nonsense, the data does show the sensor statistical information.

what is unknown is their measurement strategy.  while there is only perhaps one way to define it, how it's measured can lead to dramatically different results with a canon sensor versus a sony sensor.

Until Canon implements second stage CDS, and BSI sensors (or conversely instead of BSI, implement it as stacked), there is no way they can approach the "theoretical efficiency" of what a full frame sensor should be able to perform at, or meet or exceed exmor.

My rationale for this entire train of thought isn't picking apart Canon sensors, but picking apart the notion that since they implemented ADC's that's all they can do to improve the sensors.  That's hogwash.


« Last Edit: July 13, 2018, 10:44:57 PM by rrcphoto »

3kramd5

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2267
Agreed. For what they claim to be (a trusted industry standard), their black box methods and dependent variables are unforgivable.

privatebydesign

  • Canon EF 800mm f/5.6L IS
  • ***********
  • Posts: 6933
  • Would you take advice from a cartoons stuffed toy?
No, depending on how you measure DR the 5D MkIV bests the very best Sony sensors by 0.2 stops of DR. The A9 is 0.01 stop 'better' than the 1DX MkII, that is within measurable error and the Sony is nearly two years younger than the 1DX MkII.

Get off the DR drivel, it isn't true anymore. The best DR sensor Sony have built is in the A7R, in 2013! Every camera they have built since then has had less DR than that A7R.

Nice reply.  Do you need a nap? or a soother?

the D850 uses an EXMOR sensor.

the A7III matches the D850 from ISO 100 onwards.


The A7III and D850 are both bested by the 2013 A7R. Where have Sony and Nikon advanced in the last 5 years? I fully agree that 5 years ago Canon were wanting in the DR department, now, not so much, all the best 135 format sensors are within .75 of a stop.

I trust photonstophotos.net more than DxO, their methodology is entirely open and repeatable.
Too often we lose sight of the fact that photography is about capturing light, if we have the ability to take control of that light then we grow exponentially as photographers. More often than not the image is not about lens speed, sensor size, DR, MP's or AF, it is about the light.

rrcphoto

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2373
No, depending on how you measure DR the 5D MkIV bests the very best Sony sensors by 0.2 stops of DR. The A9 is 0.01 stop 'better' than the 1DX MkII, that is within measurable error and the Sony is nearly two years younger than the 1DX MkII.

Get off the DR drivel, it isn't true anymore. The best DR sensor Sony have built is in the A7R, in 2013! Every camera they have built since then has had less DR than that A7R.

Nice reply.  Do you need a nap? or a soother?

the D850 uses an EXMOR sensor.

the A7III matches the D850 from ISO 100 onwards.


The A7III and D850 are both bested by the 2013 A7R. Where have Sony and Nikon advanced in the last 5 years? I fully agree that 5 years ago Canon were wanting in the DR department, now, not so much, all the best 135 format sensors are within .75 of a stop.

I trust photonstophotos.net more than DxO, their methodology is entirely open and repeatable.

PDR is not an accurate measure of sensor response, since it takes into account floating black levels.  it's a great test of attempting to measure visual differents aka photographic, but it's not the sensor.
Canon sensors don't do a post ADC CDS to remove the need for floating black levels, and can still get inaccurate data because of it.

« Last Edit: July 13, 2018, 10:54:52 PM by rrcphoto »

rrcphoto

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2373
Agreed. For what they claim to be (a trusted industry standard), their black box methods and dependent variables are unforgivable.

it is, but in this case the results make sense.

Canon is on as far as I can tell their 2nd to forth generation of ADC sensors depending on wether or not you choose to assume that the full frame sensors are a different generation, but loosely... and based upon pictorial proof only ..

750D, 6D Mark II - 1st generation ADC
80D, EOS-M5,6,50,5D Mark IV, 1DX Mark II - 2nd generation ADC.
full frame mirrorless? next generation? why not?

(That was also the answer to the sensor pics in the images from the last page).

there was a distinct sensor generation flip post 70D (so it had nothing to do with DPAF) that started with the 750D.

if i was rich i'd buy the chipworks reports just to see myself what was happening ;)

like I said, there is a ton more that Canon can and should do to improve their sensors, and why should they only be satisfied with meeting Sony sensor performance, what the heck is wrong with exceeding it.  Think about sitting back with the popcorn in DPR forums.
« Last Edit: July 13, 2018, 10:59:28 PM by rrcphoto »

privatebydesign

  • Canon EF 800mm f/5.6L IS
  • ***********
  • Posts: 6933
  • Would you take advice from a cartoons stuffed toy?
No, depending on how you measure DR the 5D MkIV bests the very best Sony sensors by 0.2 stops of DR. The A9 is 0.01 stop 'better' than the 1DX MkII, that is within measurable error and the Sony is nearly two years younger than the 1DX MkII.

Get off the DR drivel, it isn't true anymore. The best DR sensor Sony have built is in the A7R, in 2013! Every camera they have built since then has had less DR than that A7R.

Nice reply.  Do you need a nap? or a soother?

the D850 uses an EXMOR sensor.

the A7III matches the D850 from ISO 100 onwards.


The A7III and D850 are both bested by the 2013 A7R. Where have Sony and Nikon advanced in the last 5 years? I fully agree that 5 years ago Canon were wanting in the DR department, now, not so much, all the best 135 format sensors are within .75 of a stop.

I trust photonstophotos.net more than DxO, their methodology is entirely open and repeatable.

PDR is not an accurate measure of sensor response, since it takes into account floating black levels.  it's a great test of attempting to measure visual differents aka photographic, but it's not the sensor.
Canon sensors don't do a post ADC CDS to remove the need for floating black levels, and can still get inaccurate data because of it.

Bill Claff has been here on the forum and is happy to explain his rational and methodology. He is also happy to point out many of the errors that are made in assumptions about the various RAW files.

http://www.canonrumors.com/forum/index.php?topic=33208.msg680952#msg680952
Too often we lose sight of the fact that photography is about capturing light, if we have the ability to take control of that light then we grow exponentially as photographers. More often than not the image is not about lens speed, sensor size, DR, MP's or AF, it is about the light.

canon rumors FORUM


rrcphoto

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2373
 ;D
No, depending on how you measure DR the 5D MkIV bests the very best Sony sensors by 0.2 stops of DR. The A9 is 0.01 stop 'better' than the 1DX MkII, that is within measurable error and the Sony is nearly two years younger than the 1DX MkII.

Get off the DR drivel, it isn't true anymore. The best DR sensor Sony have built is in the A7R, in 2013! Every camera they have built since then has had less DR than that A7R.

Nice reply.  Do you need a nap? or a soother?

the D850 uses an EXMOR sensor.

the A7III matches the D850 from ISO 100 onwards.


The A7III and D850 are both bested by the 2013 A7R. Where have Sony and Nikon advanced in the last 5 years? I fully agree that 5 years ago Canon were wanting in the DR department, now, not so much, all the best 135 format sensors are within .75 of a stop.

I trust photonstophotos.net more than DxO, their methodology is entirely open and repeatable.

PDR is not an accurate measure of sensor response, since it takes into account floating black levels.  it's a great test of attempting to measure visual differents aka photographic, but it's not the sensor.
Canon sensors don't do a post ADC CDS to remove the need for floating black levels, and can still get inaccurate data because of it.

Bill Claff has been here on the forum and is happy to explain his rational and methodology. He is also happy to point out many of the errors that are made in assumptions about the various RAW files.

http://www.canonrumors.com/forum/index.php?topic=33208.msg680952#msg680952

I understand Bills rationale completely and in terms of photographically I agree with what is there, but it's not sensor DR.

Until Canon moves the post ADC CDS from floating black / masked pixels to the sensor level it can't be.

Bill's calculation takes into account the masked pixel areas, DXO I do believe does not. DXO in this case is more accurate for the sensor but not systematically.

EXMOR sensors aren't the be all and end all, and the absolute end of improvements in sensor design. The sensors can be improved further.

There's nothing saying that Canon's next generation of sensors couldn't exceed EXMOR.  it's highly unlikely - but it's possible that a technology leapfrog could occur.

Like I said now three times, the idea that once ADC is implemented in sensor that's it, that's all they can do is complete hogwash, which is the only reason why this is being discussed right now.

« Last Edit: July 13, 2018, 11:21:53 PM by rrcphoto »

3kramd5

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2267
Like I said now three times, the idea that once ADC is implemented in sensor that's it, that's all they can do is complete hogwash, which is the only reason why this is being discussed right now.

I don’t recall that belief being expressed, but fully admit to not reading every post.

Etienne

  • EOS 5DS R
  • ******
  • Posts: 1208

This is the first rumored Canon body in five years that has caught my interest:  FF mirrorless, DPAF, swivel screen (!!!!!), presumably awesome touchscreen implementation, presumably reasonable 4K implementation, presumably good DR and high ISO performance, rumored video-centric camera. This might keep me from buying an A7 mark III. But what about the mount and available native lenses? Canon may have to play catch-up with Sony on mirroless lenses ... and FAST!



ahsanford

  • Canon EF 800mm f/5.6L IS
  • ***********
  • Posts: 7233
  • USM > STM
Like I said now three times, the idea that once ADC is implemented in sensor that's it, that's all they can do is complete hogwash, which is the only reason why this is being discussed right now.

Design/architecture-wise quite possibly, but again I direct folks at the D800 / A7R era in 2012-2013 or so.  There's magic in them thar silicon and I don't care if it's first gen on-chip or 5th, it holds up really well.

- A

ahsanford

  • Canon EF 800mm f/5.6L IS
  • ***********
  • Posts: 7233
  • USM > STM
Like I said now three times, the idea that once ADC is implemented in sensor that's it, that's all they can do is complete hogwash, which is the only reason why this is being discussed right now.

I don’t recall that belief being expressed, but fully admit to not reading every post.

I did (sort of).  Guilty. 

See plots on page 2. 

I just don't think we should expect massive improvements.  No one has shown massive improvements in some time. 

This is a massive improvement in a replacement for a product over its predecessor.  It tripled the resolution and still leapt forward -- thank you, EXMOR, on-chip, etc.

This is a really nice improvement.  (Guess what happened between the old and new models -- hint, it's a trend.)

And here's what that first example has done since then, over 6 years time:  Not a huge step forward.

I'm not saying there aren't finer, more nuanced things that have improved -- what the noise actually looks like, how well you can manipulate the files, etc. -- but in broad strokes each manufacturer got a tasty bump from moving to on-chip and they haven't dramatically improved since then.

- A
« Last Edit: July 14, 2018, 12:35:51 AM by ahsanford »

Jethro

  • Canonflex
  • *
  • Posts: 1
Hi, it's my first post - although I've lurked for some time.

On the new sensor point, I don't pretend to understand the EXMORs and SNRs and ADCs, so I'm coming from a relative place of ignorance.  But, in terms of having a different FF sensor to the 5D4, is it not possible that it is a similar sensor (with the same MP count) but simply updated for a native mirrorless environment?  Is there anything that 'needs' to be done to an existing sensor designed for a SLR case to use it in a mirrorless case?  Or it simply could be that they are turning over the full frame 5d4-level sensor to incorporate the cumulative advances to date a couple of years earlier because the FF mirrorless release is a big event.

Like everyone else I'm sweating on the mount issue - as a 6D owner I'm getting ready to turn over (and potentially upgrade) the body, but I really don't want to turn over my L (and 3rd party) lenses.

canon rumors FORUM


privatebydesign

  • Canon EF 800mm f/5.6L IS
  • ***********
  • Posts: 6933
  • Would you take advice from a cartoons stuffed toy?
Hi, it's my first post - although I've lurked for some time.

On the new sensor point, I don't pretend to understand the EXMORs and SNRs and ADCs, so I'm coming from a relative place of ignorance.  But, in terms of having a different FF sensor to the 5D4, is it not possible that it is a similar sensor (with the same MP count) but simply updated for a native mirrorless environment?  Is there anything that 'needs' to be done to an existing sensor designed for a SLR case to use it in a mirrorless case?  Or it simply could be that they are turning over the full frame 5d4-level sensor to incorporate the cumulative advances to date a couple of years earlier because the FF mirrorless release is a big event.

Like everyone else I'm sweating on the mount issue - as a 6D owner I'm getting ready to turn over (and potentially upgrade) the body, but I really don't want to turn over my L (and 3rd party) lenses.

If the flange distance is the same then nothing should be needed to be changed sensor wise between SLR and MILC, if the MILC has a shorter flange distance then the sensor stack, which includes the microlens array on top of the sensor, would need to be changed to optimize the light path.
Too often we lose sight of the fact that photography is about capturing light, if we have the ability to take control of that light then we grow exponentially as photographers. More often than not the image is not about lens speed, sensor size, DR, MP's or AF, it is about the light.

privatebydesign

  • Canon EF 800mm f/5.6L IS
  • ***********
  • Posts: 6933
  • Would you take advice from a cartoons stuffed toy?
No, depending on how you measure DR the 5D MkIV bests the very best Sony sensors by 0.2 stops of DR. The A9 is 0.01 stop 'better' than the 1DX MkII, that is within measurable error and the Sony is nearly two years younger than the 1DX MkII.

Get off the DR drivel, it isn't true anymore. The best DR sensor Sony have built is in the A7R, in 2013! Every camera they have built since then has had less DR than that A7R.

Nice reply.  Do you need a nap? or a soother?

the D850 uses an EXMOR sensor.

the A7III matches the D850 from ISO 100 onwards.


The A7III and D850 are both bested by the 2013 A7R. Where have Sony and Nikon advanced in the last 5 years? I fully agree that 5 years ago Canon were wanting in the DR department, now, not so much, all the best 135 format sensors are within .75 of a stop.

I trust photonstophotos.net more than DxO, their methodology is entirely open and repeatable.

PDR is not an accurate measure of sensor response, since it takes into account floating black levels.  it's a great test of attempting to measure visual differents aka photographic, but it's not the sensor.
Canon sensors don't do a post ADC CDS to remove the need for floating black levels, and can still get inaccurate data because of it.

It might not be "an accurate measure of sensor response", whatever that means, but in my experience it is an accurate estimation of the differences I will see (or not) when processing RAW files from the various cameras. And that is the only criteria that is relevant and I have any interest in.

I used to be able to push and pull a Sony or Nikon RAW file a sh!t ton more than Canon RAW files, now, not so much.

Too often we lose sight of the fact that photography is about capturing light, if we have the ability to take control of that light then we grow exponentially as photographers. More often than not the image is not about lens speed, sensor size, DR, MP's or AF, it is about the light.

canon rumors FORUM