Best low light FF Canon DSLR

Started 4 months ago | Discussions
Dr_Jon Veteran Member • Posts: 6,108
Re: Best low light FF Canon DSLR

John Sheehy wrote:

Dr_Jon wrote:

It depends on what you mean, if you mean the best DR then above ISO 1000 they are basically all the same.

I hate to sound like a broken record, but why do you consider DR at high ISOs?

Just because they can be measured, does not mean that they reflect high ISO noise. They do not; they only correlate weakly, because manufacturers tend to keep highlight headroom in a general range in most cameras.

PDR is not a useful reference for high ISO noise. It has nothing directly to do with high ISO noise. The DPReview comparison tool is more useful, because it shows what noise actually looks like, with the same exposure.

But either you get different sized images which are hard to compare, all but one downscaled (that may make the noise in three look better or worse depending on how they do it) or all 4 downscaled a lot (ditto).

BTW I gave Studio scene examples here (in this thread):
https://www.dpreview.com/forums/post/62828972
(Note I cheated and used COMP but threw in a lower-res sensor so all the ones they might be interested in got scaled to 16MP, as I thought that might be interesting.)

Roger Clark has a good article on the 6D, but not more recent cameras to compare it to, plus not an easy read for many people.

http://www.clarkvision.com/reviews/evaluation-canon-6d/index.html

P.S. my "Summary" was 5D4 followed by 6D, or possibly an A7III/A7rIII, would you differ?

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
Now...

Dr_Jon wrote:

Mako2011 wrote:

It never changes the amount of light you can collect. As an example....blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode. Notice that regardless of what ISO setting you select...the sensor never captures more or less light. How the resulting image looks may change...but the amount of light captured by the sensor never ever does. The ISO setting has no effect on the Full Well Capacity nor the min charge an individual pixel can hold

I think my view is the more useful one here,

The original view/statement, "ISO affected amount of light collected", was not accurate though and could lead many astray.

..as while ISO doesn't change the amount of light the sensor collects

And that's a big 180 from your original premise. And leads to much more useful theory. Glad you see the "Light". It helps others.

And don't forget, sometimes a change in ISO results in nothing more than the camera processing engine raising or lowering the brightness setting...a pure digital only manipulation that could be done just as well had one used Light Room to change "ISO"

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Dr_Jon Veteran Member • Posts: 6,108
Re: Now...

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

It never changes the amount of light you can collect. As an example....blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode. Notice that regardless of what ISO setting you select...the sensor never captures more or less light. How the resulting image looks may change...but the amount of light captured by the sensor never ever does. The ISO setting has no effect on the Full Well Capacity nor the min charge an individual pixel can hold

I think my view is the more useful one here,

The original view/statement, "ISO affected amount of light collected", was not accurate though and could lead many astray.

..as while ISO doesn't change the amount of light the sensor collects

And that's a big 180 from your original premise. And leads to much more useful theory. Glad you see the "Light". It helps others.

And don't forget, sometimes a change in ISO results in nothing more than the camera processing engine raising or lowering the brightness setting...a pure digital only manipulation that could be done just as well had one used Light Room to change "ISO"

No, sorry, I'm sticking with my original premise. As ISO increases you capture less light in the Real World. The fact that the sensor captures the same amount of light is irrelevant as there is no way to access that data. So it's no use to anyone.

It is true that for +/- 1/3 stops a few cameras will do it digitally (I talked about that in a post here) but that's pretty irrelevant when it comes to making much of a difference. (Also at very low or very very high ISOs that may happen too, but really away from what people are likely to be doing in low light.)

I'm going with: "At most useful ISOs doubling ISO halves the amount of light you can capture", since it's what happens... suggesting anything else isn't sensible as it's in such rare special cases.

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
turn...
1

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

It never changes the amount of light you can collect. As an example....blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode. Notice that regardless of what ISO setting you select...the sensor never captures more or less light. How the resulting image looks may change...but the amount of light captured by the sensor never ever does. The ISO setting has no effect on the Full Well Capacity nor the min charge an individual pixel can hold

I think my view is the more useful one here,

The original view/statement, "ISO affected amount of light collected", was not accurate though and could lead many astray.

..as while ISO doesn't change the amount of light the sensor collects

And that's a big 180 from your original premise. And leads to much more useful theory. Glad you see the "Light". It helps others.

And don't forget, sometimes a change in ISO results in nothing more than the camera processing engine raising or lowering the brightness setting...a pure digital only manipulation that could be done just as well had one used Light Room to change "ISO"

No, sorry, I'm sticking with my original premise.

So above you say "ISO doesn't change the amount of light the sensor collects"...

and now you are saying "ISO does change the amount of light the sensor collect"?

As ISO increases you capture less light in the Real World.

No....A factually inaccurate statement. Try it. Blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode (Shoot RAW). Notice that regardless of what ISO setting you select...the sensor never captures more or less light. You can open the files in RawDigger and confirm the amount of light captured did not changes you increase ISO setting

The fact that the sensor captures the same amount of light is irrelevant as there is no way to access that data. So it's no use to anyone.

Not true....Many of use often shoot RAW in an ISO-less ETTR workflow (often using UniWB) from time to time. Pretty easy to access all the data

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Dr_Jon Veteran Member • Posts: 6,108
Re: turn...

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

It never changes the amount of light you can collect. As an example....blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode. Notice that regardless of what ISO setting you select...the sensor never captures more or less light. How the resulting image looks may change...but the amount of light captured by the sensor never ever does. The ISO setting has no effect on the Full Well Capacity nor the min charge an individual pixel can hold

I think my view is the more useful one here,

The original view/statement, "ISO affected amount of light collected", was not accurate though and could lead many astray.

..as while ISO doesn't change the amount of light the sensor collects

And that's a big 180 from your original premise. And leads to much more useful theory. Glad you see the "Light". It helps others.

And don't forget, sometimes a change in ISO results in nothing more than the camera processing engine raising or lowering the brightness setting...a pure digital only manipulation that could be done just as well had one used Light Room to change "ISO"

No, sorry, I'm sticking with my original premise.

So above you say "ISO doesn't change the amount of light the sensor collects"...

and now you are saying "ISO does change the amount of light the sensor collect"?

I'm sorry, what don't you get that while the sensor captures the the same amount of light at higher ISO there is no way to access the data that so no-one cares?

As ISO increases you capture less light in the Real World.

No....A factually inaccurate statement. Try it. Blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode (Shoot RAW). Notice that regardless of what ISO setting you select...the sensor never captures more or less light. You can open the files in RawDigger and confirm the amount of light captured did not changes you increase ISO setting

Really not, as the ISO amplifier blows the data off the top of the range AND IT ISN'T IN the Raw file!!! You are wrong here as the Raw file is captured after the ISO amplifier and so a bunch of stuff will be clipped and lost. See the 6D saturation numbers I gave. They are for Raw files!

The fact that the sensor captures the same amount of light is irrelevant as there is no way to access that data. So it's no use to anyone.

Not true....Many of use often shoot RAW in an ISO-less ETTR workflow (often using UniWB) from time to time. Pretty easy to access all the data

Err, most of the canon camera mentioned are crap for shooting ISO-less as they aren't ISO-less except at ISOs 800 or 1600 and above (I went into detail on that). Plus you weren't talking about ISO-invariance before (where you don't get an image to review)?

Shooting ISO-invariant for any ISO below 800 will just be a noise-fest on off-sensor ADC Canon cameras. That isn't what the OP wanted.

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
still not

Dr_Jon wrote:

Mako2011 wrote:

So above you say "ISO doesn't change the amount of light the sensor collects"...

and now you are saying "ISO does change the amount of light the sensor collect"?

I'm sorry, what don't you get that while the sensor captures the the same amount of light at higher ISO

Now you are back to agreeing that ISO does not affect the amount of light the sensor captures/records

As ISO increases you capture less light in the Real World.

No....A factually inaccurate statement. Try it. Blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode (Shoot RAW). Notice that regardless of what ISO setting you select...the sensor never captures more or less light. You can open the files in RawDigger and confirm the amount of light captured did not changes you increase ISO setting

Really not, as the ISO amplifier blows the data off the top of the range AND IT ISN'T IN the Raw file!!!

Only time you "blow out the data" (exceed full well capacity) and can't use it is if you set shutter speed and/or aperture so full well capacity gets exceeded. ISO doesn't not play a part in what the camera sensor actually captures/records

Not true....Many of use often shoot RAW in an ISO-less ETTR workflow (often using UniWB) from time to time. Pretty easy to access all the data

Shooting ISO-invariant for any ISO below 800 will just be a noise-fest on off-sensor ADC Canon cameras. That isn't what the OP wanted.

But again....it shows ISO does not have any affect on how much light the sensor actually captures

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Dr_Jon Veteran Member • Posts: 6,108
Re: still not

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

So above you say "ISO doesn't change the amount of light the sensor collects"...

and now you are saying "ISO does change the amount of light the sensor collect"?

I'm sorry, what don't you get that while the sensor captures the the same amount of light at higher ISO

Now you are back to agreeing that ISO does not affect the amount of light the sensor captures/records

Again. Really. You can capture it but as there is no way whatever to see it it's pointless.

As ISO increases you capture less light in the Real World.

No....A factually inaccurate statement. Try it. Blank wall lit by constant artificial light. Set camera on tripod at f2 and 1/60s in MANUAL EXPOSURE mode (Shoot RAW). Notice that regardless of what ISO setting you select...the sensor never captures more or less light. You can open the files in RawDigger and confirm the amount of light captured did not changes you increase ISO setting

Really not, as the ISO amplifier blows the data off the top of the range AND IT ISN'T IN the Raw file!!!

Only time you "blow out the data" (exceed full well capacity) and can't use it is if you set shutter speed and/or aperture so full well capacity gets exceeded. ISO doesn't not play a part in what the camera sensor actually captures/records

BUT as ISO increases you cannot, in any way, see what was captured so it is no use to photography.

Not true....Many of use often shoot RAW in an ISO-less ETTR workflow (often using UniWB) from time to time. Pretty easy to access all the data

Shooting ISO-invariant for any ISO below 800 will just be a noise-fest on off-sensor ADC Canon cameras. That isn't what the OP wanted.

But again....it shows ISO does not have any affect on how much light the sensor actually captures

Okay, we're getting nowhere, it's no use to the OP and I don't see anything I've said that I don't think is correct, so how about we all go get a coffee. BTW do you disagree with the cameras I suggested for them?

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
Yes

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

So above you say "ISO doesn't change the amount of light the sensor collects"...

and now you are saying "ISO does change the amount of light the sensor collect"?

I'm sorry, what don't you get that while the sensor captures the the same amount of light at higher ISO

Now you are back to agreeing that ISO does not affect the amount of light the sensor captures/records

Again. Really. You can capture it

Yes...ISO does not affect what the sensor captured.

but as there is no way whatever to see it it's pointless.

YMMV

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Dr_Jon Veteran Member • Posts: 6,108
Re: Yes

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

So above you say "ISO doesn't change the amount of light the sensor collects"...

and now you are saying "ISO does change the amount of light the sensor collect"?

I'm sorry, what don't you get that while the sensor captures the the same amount of light at higher ISO

Now you are back to agreeing that ISO does not affect the amount of light the sensor captures/records

Again. Really. You can capture it

Yes...ISO does not affect what the sensor captured.

but as there is no way whatever to see it it's pointless.

YMMV

Okay, if you have a tunnelling electron microscopy you might be in with a shot, but you'd need to disassemble the chip while keeping it powered

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
Or

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Again. Really. You can capture it

Yes...ISO does not affect what the sensor captured.

but as there is no way whatever to see it it's pointless.

YMMV

Okay, if you have a tunnelling electron microscopy you might be in with a shot, but you'd need to disassemble the chip while keeping it powered

Or just a camera and Photoshop

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Dr_Jon Veteran Member • Posts: 6,108
Re: Or

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Again. Really. You can capture it

Yes...ISO does not affect what the sensor captured.

but as there is no way whatever to see it it's pointless.

YMMV

Okay, if you have a tunnelling electron microscopy you might be in with a shot, but you'd need to disassemble the chip while keeping it powered

Or just a camera and Photoshop

No, as the data won't be in the Raw file... which is stored after the ISO amplifier and after the digitiser. Every time you crank the ISO up a stop it throws away the top half of the previous data.

Surely we're done?

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
depends

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Again. Really. You can capture it

Yes...ISO does not affect what the sensor captured.

but as there is no way whatever to see it it's pointless.

YMMV

Okay, if you have a tunnelling electron microscopy you might be in with a shot, but you'd need to disassemble the chip while keeping it powered

Or just a camera and Photoshop

No, as the data won't be in the Raw file...

Depends  .....regardless, the ISO setting won't affect how much light the sensor captures

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Dr_Jon Veteran Member • Posts: 6,108
Re: depends

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Again. Really. You can capture it

Yes...ISO does not affect what the sensor captured.

but as there is no way whatever to see it it's pointless.

YMMV

Okay, if you have a tunnelling electron microscopy you might be in with a shot, but you'd need to disassemble the chip while keeping it powered

Or just a camera and Photoshop

No, as the data won't be in the Raw file...

Depends .....regardless, the ISO setting won't affect how much light the sensor captures

But you' agree there is no way you can use most of that data as it is not available to you in any way, as it was clipped at the ISO amplifier?

Example - you can get 3V from a full pixel. The ADC converts 0-3V into 16k divisions.

At ISO 100 the ISO amplifier passes 0-3V to the ADC .
If the pixel was 1/8 full you get a count of 2k
If the pixel was 1/4 full you get a count of 4k
If the pixel was 1/2 full you get a count of 8k
If the pixel was 3/4 full you get a count of 12k

At ISO 200 it doubles the signal. 0-1.5V becomes 0-3V and 1.5-3V gets limited to 3V and
you have no idea what value was in a pixel that was more than half full.
If the pixel was 1/8 full you get a count of 4k
If the pixel was 1/4 full you get a count of 8k
If the pixel was 1/2 full you get a count of 16k
If the pixel was 3/4 full you get a count of 16k

At ISO 400 it quadruples the signal. 0-0.75V becomes 0-3V and 0.75-3V gets limited to 3V and you have no idea what value was in a pixel that was more than a quarter full.
If the pixel was 1/8 full you get a count of 8k
If the pixel was 1/4 full you get a count of 16k
If the pixel was 1/2 full you get a count of 16k
If the pixel was 3/4 full you get a count of 16k

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
Some can

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

No, as the data won't be in the Raw file...

Depends .....regardless, the ISO setting won't affect how much light the sensor captures

But you' agree there is no way you can use most of that data

Not necessarily. Depends on the situation and many variables. Regardless, the ISO setting won't affect how much light the sensor captures

John Sheehy Forum Pro • Posts: 22,151
Re: Best low light FF Canon DSLR
2

KLO82 wrote:

How did you derive your original statement about Canon ranking them for high ISO noise based on this?

Do you really think that the 5Ds needs 8x as much light as the 1DxII for the same quality at high ISOs?

Canon's top "non-expanded" ISO is not based directly on noise.

KLO82 Senior Member • Posts: 1,207
Re: Best low light FF Canon DSLR

John Sheehy wrote:

KLO82 wrote:

How did you derive your original statement about Canon ranking them for high ISO noise based on this?

Do you really think that the 5Ds needs 8x as much light as the 1DxII for the same quality at high ISOs?

Canon's top "non-expanded" ISO is not based directly on noise.

I think it is based on comparing pixel level noise of OOC jpegs. But even then, I have to admit that 5DS/ 5DSR cannot be that bad compared to the other cameras. I should have written "based on Canon's marketing claim" in the first place.

 KLO82's gear list:KLO82's gear list
Canon EOS 550D Canon EOS 5D Mark III Canon EF 50mm f/1.4 USM Canon EF 85mm f/1.8 USM Canon EF 24-105mm f/4L IS USM +1 more
Dr_Jon Veteran Member • Posts: 6,108
Re: Some can

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

No, as the data won't be in the Raw file...

Depends .....regardless, the ISO setting won't affect how much light the sensor captures

But you' agree there is no way you can use most of that data

Not necessarily. Depends on the situation and many variables. Regardless, the ISO setting won't affect how much light the sensor captures

I've said my piece, I think you've worked out what that is, so hopefully we're done...

 Dr_Jon's gear list:Dr_Jon's gear list
Nikon Coolpix 950 Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 +29 more
John Sheehy Forum Pro • Posts: 22,151
Re: Best low light FF Canon DSLR
2

KLO82 wrote:

John Sheehy wrote:

KLO82 wrote:

How did you derive your original statement about Canon ranking them for high ISO noise based on this?

Do you really think that the 5Ds needs 8x as much light as the 1DxII for the same quality at high ISOs?

Canon's top "non-expanded" ISO is not based directly on noise.

I think it is based on comparing pixel level noise of OOC jpegs. But even then, I have to admit that 5DS/ 5DSR cannot be that bad compared to the other cameras. I should have written "based on Canon's marketing claim" in the first place.

The read noise at high ISOs, as measured, only varies about 1.5 stops across all current FF cameras, and photon noise only varies by a much smaller amount.

There is no reason why the maximum ISO, if it were based on noise, to vary by much more than a stop across FF cameras.

Unfortunately, the DPR staff only does the studio comparison images up to the expanded ISOs, and the 5Ds stops at 12800.

I am not saying that the noise character of the 5Ds is perfect, but it could be that it competes very well against other FF bodies for small images pushed to ISO 1 million, but you would never know unless you did the comparison yourself. A camera that starts getting too noisy for pixel-level views on coarse monitors at ISO 12800 isn't necessarily going to be a poor performer if it is downsampled or smoothed, if it does not have low-frequency noises and banding noise of any significance. It is actually difficult to see how an image will survive from these high magnifications because the dot gamma effect of bright noise outliers can mask any low-frequency content.

Mako2011
MOD Mako2011 Forum Pro • Posts: 26,106
Just ...

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

No, as the data won't be in the Raw file...

Depends .....regardless, the ISO setting won't affect how much light the sensor captures

But you' agree there is no way you can use most of that data

Not necessarily. Depends on the situation and many variables. Regardless, the ISO setting won't affect how much light the sensor captures

I've said my piece, I think you've worked out what that is, so hopefully we're done...

I was just pointing out that the statement "if you can reduce the ISO, which changes how much light you collect to give a correct exposure" might imply to some the the ISO setting affects how much light the sensor collects at a given exposure and that isn't true.  That's all

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

J A C S
J A C S Forum Pro • Posts: 14,915
Re: Just ...

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

No, as the data won't be in the Raw file...

Depends .....regardless, the ISO setting won't affect how much light the sensor captures

But you' agree there is no way you can use most of that data

Not necessarily. Depends on the situation and many variables. Regardless, the ISO setting won't affect how much light the sensor captures

I've said my piece, I think you've worked out what that is, so hopefully we're done...

I was just pointing out that the statement "if you can reduce the ISO, which changes how much light you collect to give a correct exposure" might imply to some the the ISO setting affects how much light the sensor collects at a given exposure and that isn't true. That's all

In the context of noise, this is irrelevant. I higher ISO may clip highlights if the scene is contrasty enough but this will not increase the noise of what is left because of less light collected. On the other hand, a higher ISO improves the noise overall, which is the whole raison d'être (but this gets a bit out of topic).

Keyboard shortcuts:
FForum MMy threads