Best low light FF Canon DSLR

Started 4 months ago | Discussions
MAC Forum Pro • Posts: 14,168
Re: Best low light FF Canon DSLR

J A C S wrote:

steks wrote:

Is there somewhere info to be found to figure out which camera is most light sensitive? I am thinking of buying a secondhand 5D/5Dii of 6D to use indoors with a 50mm prime. For other situations my 77D is more than enough for my use.

Of those you listed, the 6D, definitely. The newer ones do not offer any/much low light improvements.

and I got a great 6d value, $900 refurb

I like the colors of my older model

 MAC's gear list:MAC's gear list
Canon EOS 6D Canon EOS Rebel SL1 Canon EOS 7D Mark II Canon EOS Rebel T7i Canon EF 70-200mm f/2.8L USM +10 more
KLO82 Senior Member • Posts: 1,208
Re: Best low light FF Canon DSLR

If you believe Canon's claim:

1DX II > 6D II > 5D IV > 6D > 5DS/5DSR

 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
OP steks New Member • Posts: 11
Re: Best low light FF Canon DSLR

Thanks for all the reactions, to keep things short, is it fair to say that there is difference in low lIght IQ, but it’s wrong te expect a high improvement when using an comparable aperture.

Lets compare a 30mm f/1.4 on my 77D cropsenor with a 50mm stopped down to f/2.2 on a 6D (30x1.6=48 and 1.4x1.6=2.2) am I correct when saying, that because of the double in size photosites on the 6D (24,2/20,2x1.6=1.9), you can expect one stop extra sensitivity? Furthermore you have the abiltity to use faster primes, lets say a 50mm f/1.4 which also gives you an extra stop?

About an TTL flash, Thanks for the advice, i have one and used it quite a lot on my 400D when I only had a kit lens. I will experiment with it again, but there are a lot of situations where a flash won’t be the best or easiest solution.

J A C S
J A C S Forum Pro • Posts: 14,957
Re: Best low light FF Canon DSLR

steks wrote:

Thanks for all the reactions, to keep things short, is it fair to say that there is difference in low lIght IQ, but it’s wrong te expect a high improvement when using an comparable aperture.

The 6D has a very noticeable improvement over the 5D2 and the 5D in high ISO noise.

Lets compare a 30mm f/1.4 on my 77D cropsenor with a 50mm stopped down to f/2.2 on a 6D (30x1.6=48 and 1.4x1.6=2.2) am I correct when saying, that because of the double in size photosites on the 6D (24,2/20,2x1.6=1.9), you can expect one stop extra sensitivity?

The size of the photosites is a secondary factor. In that case, you get more or less the same noise but a noticeably better resolution (some of it will be lost in the noise).

Furthermore you have the abiltity to use faster primes, lets say a 50mm f/1.4 which also gives you an extra stop?

Actually, this is the factor giving you less noise, at the expense of a shallower DOF.

Dr_Jon Veteran Member • Posts: 6,131
Re: Best low light FF Canon DSLR

KLO82 wrote:

If you believe Canon's claim:

1DX II > 6D II > 5D IV > 6D > 5DS/5DSR

Which claim (really asking)?

Also would someone like to say what they mean by "low light", as low-ISO shots differ from high ISO and people do both in low light (although more the latter).

Oh and at low ISO 5Dsr beats the 6s on a image scale, but not per pixel...

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Canon%20EOS%206D,Canon%20EOS%206D%20Mark%20II
At high ISO it's a wash...

 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
Dr_Jon Veteran Member • Posts: 6,131
Re: Best low light FF Canon DSLR
1

steks wrote:

Thanks for all the reactions, to keep things short, is it fair to say that there is difference in low lIght IQ, but it’s wrong te expect a high improvement when using an comparable aperture.

Lets compare a 30mm f/1.4 on my 77D cropsenor with a 50mm stopped down to f/2.2 on a 6D (30x1.6=48 and 1.4x1.6=2.2) am I correct when saying, that because of the double in size photosites on the 6D (24,2/20,2x1.6=1.9), you can expect one stop extra sensitivity? Furthermore you have the abiltity to use faster primes, lets say a 50mm f/1.4 which also gives you an extra stop?

About an TTL flash, Thanks for the advice, i have one and used it quite a lot on my 400D when I only had a kit lens. I will experiment with it again, but there are a lot of situations where a flash won’t be the best or easiest solution.

It's complicated as the older Canon's performance varies quite a bit with ISO. The newer ones are better for that, but the DPAF ones lose a little to sensors with one pixel per pixel (if you see what I mean). The big thing to help at higher ISOs is to have a dual-gain sensor, which Canon haven't done yet. A BSI sensor helps a teeny bit, but Canon don't have them either (but doesn't really matter at APS/FF sizes).

Generally what you need to consider:

* The f-stop is light per unit area. So at the same f-stop bigger sensors receive more light in the same time period. (Or conversely the smaller sensors' data is spread more widely at the final viewing size, which is usually sensor size independent.) Pixel size doesn't matter much as you get more pixels for the same area of the final image (it does if you pixel-peep, but that means zooming into the image more and is meaningless).

* Note faster lenses only help at a particular sensor size if:
(i) They enable you to reduce the ISO (otherwise the camera just alters the shutter speed to keep the light gathered the same).
(ii) You can live with the depth of field you end up with.

* Image noise is mostly noise in the incoming light, if you capture twice the light the percentage of that which is noise goes down by 1.4x.

* The camera noise tends to even out at ISO 800 and above, so not a factor unless you have a dual-gain sensor. It causes the step in the Sony graph here:
http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%206D,Nikon%20D850,Sony%20ILCE-7M3
(BTW the bouncing of the 6D graph is because it only does whole ISO stops in hardware, if you want +/- 1/3 stops it shoots at a whole ISO - i.e. 100, 200, 400, 800... - and pushes or pulls the digital image 1/3 of a stop.)

* Other really minor factors are how permissive the colour filters over the pixels are (mostly they capture a bit either side of the nominal light, so let maybe 40% of the incoming light through, not the 33% you might expect) plus the Quantum Efficiency (QE), namely what percentage of incoming photons get converted to electrons, this varies with colour and most ways of calculating it from DXO (or similar) data have so many error sources I'd ignore them and assume it's about 50% and so you don't care when comparing cameras. Oh and how well the microlenses and light pipes aim the light into the pixel, which again like the previous two I wouldn't worry about.

So in summary at higher ISOs you mostly care about about sensor size or if a dual gain sensor is available (not for Canon). At lower ISOs you want one of the newer Canons with the analogue to digital converters on the sensor, as that reduces the camera's low ISO noise significantly, so for FF 5D4 or R, but not RP as that's the 6D2's sensor.

Oh and don't forget whether it can AF well at low light levels...

I think the 5D4 is probably numerically the best Canon DSLR for low light overall, although the 6D's noise can look more pleasing (very subjective). But a Sony A7rIII is better.

Or just ignore all that and use either Neat Image 8 or DXO Photolab's Prime noise reduction....

 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
quiquae Senior Member • Posts: 1,533
Re: Best low light FF Canon DSLR

Waldemar wrote:

I think you´d need something like a Nikon D750 to get visibly (in print!) cleaner high ISO shots than from your 77D. Canon´s current APS-C sensor is pretty good.

I don’t have the 77D, but I do have the M5 which uses a very similar sensor, and I disagree. While the modern APS-C sensors do have dynamic ranges as wide as older full frame ones at low ISO, you should see at least a full stop of difference between the 77D and the seven year old 6D at ISO 1600 and above.

When I had the 6D, I usually had the upper limit of auto-ISO set to 3200, but would change to 6400 at the first sign of low light, stretched it to 12800 routinely, and even 25600 was still quite usable if necessary. With the 77D/M5 sensor, ISO 1600 is as far as I’m willing to let auto-ISO go; 3200 is a compromise, and 6400 gets selected only in a pinch. That’s still a full stop better than the previous generation sensor on e.g. Rebel T4i, but well behind what the 6D could do.

 quiquae's gear list:quiquae's gear list
Canon PowerShot SD400 Canon EOS M5 Canon EOS R Canon EF 100mm F2.8L Macro IS USM Canon EF 70-200mm F4L IS USM +10 more
Dr_Jon Veteran Member • Posts: 6,131
Re: Best low light FF Canon DSLR

J A C S wrote:

steks wrote:

Thanks for all the reactions, to keep things short, is it fair to say that there is difference in low lIght IQ, but it’s wrong te expect a high improvement when using an comparable aperture.

The 6D has a very noticeable improvement over the 5D2 and the 5D in high ISO noise.

Lets compare a 30mm f/1.4 on my 77D cropsenor with a 50mm stopped down to f/2.2 on a 6D (30x1.6=48 and 1.4x1.6=2.2) am I correct when saying, that because of the double in size photosites on the 6D (24,2/20,2x1.6=1.9), you can expect one stop extra sensitivity?

The size of the photosites is a secondary factor. In that case, you get more or less the same noise but a noticeably better resolution (some of it will be lost in the noise).

Furthermore you have the abiltity to use faster primes, lets say a 50mm f/1.4 which also gives you an extra stop?

Actually, this is the factor giving you less noise, at the expense of a shallower DOF.

I said this below, but I'll point it out again in case it's helpful. A f/1.4 lens gives you the same light as a f/2 lens as the camera will just halve the time the shutter is open to give a correct exposure. It only helps if you can reduce the ISO, which changes how much light you collect to give a correct exposure.

So:

Light level Ev 10 (Twilight)

FF, f/2, ISO 200, shutter speed will be 1/500th, light captured will be X
FF, f/1.4, ISO 200, shutter speed will be 1/1000th, light captured will be X
APS, f/1.4, ISO 200, shutter speed will be 1/1000th, light captured will be X/2.56 (-1.36 stops)
FF, f/1.4, ISO 100, shutter speed will be 1/500th, light captured will be 2X

 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
GreatWhiteWing Contributing Member • Posts: 925
Re: Best low light FF Canon DSLR

One way to compare is right here on dpreview. Go to sample images and use comparison tool. Can select up to four cameras and compare a sample image at various ISO's. Give you a pretty good idea of the relative performance of various sensors.

 GreatWhiteWing's gear list:GreatWhiteWing's gear list
Canon EOS 7D Mark II Canon EF-S 10-22mm F3.5-4.5 USM Canon EF 24-70mm F4L IS USM Tamron SP 1.4X Pro Teleconverter Tamron SP 70-200 F2.8 G2 +2 more
Dr_Jon Veteran Member • Posts: 6,131
Re: Best low light FF Canon DSLR

GreatWhiteWing wrote:

One way to compare is right here on dpreview. Go to sample images and use comparison tool. Can select up to four cameras and compare a sample image at various ISO's. Give you a pretty good idea of the relative performance of various sensors.

The problem (I think) is in order to see a decent comparison you need to use the tool that makes the images about a comparable size, and I'm really not sure now well that works for downsizing noise, plus I think it doesn't downsize the smallest one, just the others to match. (Or get the full-res images and re-size them yourself.)

Okay, so I threw a 16MP sensor in (bottom right), to make sure everything else got resized (note it is at a lower ISO, so ignore the Fuji):
https://www.dpreview.com/reviews/image-comparison?attr18=lowlight&attr13_0=canon_eos6d&attr13_1=canon_eos5dmkiv&attr13_2=canon_eosr&attr13_3=fujifilm_xpro1&attr15_0=raw&attr15_1=raw&attr15_2=raw&attr15_3=raw&attr16_0=12800&attr16_1=12800&attr16_2=12800&attr16_3=6400&normalization=compare&widget=1&x=-0.8976828885400314&y=0.10899887514060742

Note detail level on 5D4 (all at 16MP)_:
https://www.dpreview.com/reviews/image-comparison?attr18=lowlight&attr13_0=canon_eos6d&attr13_1=canon_eos5dmkiv&attr13_2=canon_eosr&attr13_3=fujifilm_xpro1&attr15_0=raw&attr15_1=raw&attr15_2=raw&attr15_3=raw&attr16_0=6400&attr16_1=6400&attr16_2=6400&attr16_3=6400&normalization=compare&widget=1&x=-0.42521821036106755&y=-0.6988751406074243

 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
KLO82 Senior Member • Posts: 1,208
Re: Best low light FF Canon DSLR

Dr_Jon wrote:

KLO82 wrote:

If you believe Canon's claim:

1DX II > 6D II > 5D IV > 6D > 5DS/5DSR

Which claim (really asking)?

Also would someone like to say what they mean by "low light", as low-ISO shots differ from high ISO and people do both in low light (although more the latter).

Oh and at low ISO 5Dsr beats the 6s on a image scale, but not per pixel...

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Canon%20EOS%206D,Canon%20EOS%206D%20Mark%20II
At high ISO it's a wash...

 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
J A C S
J A C S Forum Pro • Posts: 14,957
Re: Best low light FF Canon DSLR

Dr_Jon wrote:

J A C S wrote:

steks wrote:

Thanks for all the reactions, to keep things short, is it fair to say that there is difference in low lIght IQ, but it’s wrong te expect a high improvement when using an comparable aperture.

The 6D has a very noticeable improvement over the 5D2 and the 5D in high ISO noise.

Lets compare a 30mm f/1.4 on my 77D cropsenor with a 50mm stopped down to f/2.2 on a 6D (30x1.6=48 and 1.4x1.6=2.2) am I correct when saying, that because of the double in size photosites on the 6D (24,2/20,2x1.6=1.9), you can expect one stop extra sensitivity?

The size of the photosites is a secondary factor. In that case, you get more or less the same noise but a noticeably better resolution (some of it will be lost in the noise).

Furthermore you have the abiltity to use faster primes, lets say a 50mm f/1.4 which also gives you an extra stop?

Actually, this is the factor giving you less noise, at the expense of a shallower DOF.

I said this below, but I'll point it out again in case it's helpful. A f/1.4 lens gives you the same light as a f/2 lens as the camera will just halve the time the shutter is open to give a correct exposure.

Of course, the OP and I were talking about light per unit of time.

It only helps if you can reduce the ISO, which changes how much light you collect to give a correct exposure.

The ISO is mostly irrelevant, and the "correct exposure" does not mean much in this context. With a faster lens, you can actually have a different exposure.

Mako2011
MOD Mako2011 Forum Pro • Posts: 26,124
Light

Dr_Jon wrote:

I said this below, but I'll point it out again in case it's helpful. A f/1.4 lens gives you the same light as a f/2 lens as the camera will just halve the time the shutter is open to give a correct exposure. It only helps if you can reduce the ISO, which changes how much light you collect to give a correct exposure.

Just so folks are clear.  Changing ISO alone has no effect on how much light you can collect. Only a chance in shutter speed and/or aperture and/or scene luminance changes how much light you collect

-- 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)

OP steks New Member • Posts: 11
Re: Best low light FF Canon DSLR

Dr_Jon wrote:

steks wrote:

Thanks for all the reactions, to keep things short, is it fair to say that there is difference in low lIght IQ, but it’s wrong te expect a high improvement when using an comparable aperture.

Lets compare a 30mm f/1.4 on my 77D cropsenor with a 50mm stopped down to f/2.2 on a 6D (30x1.6=48 and 1.4x1.6=2.2) am I correct when saying, that because of the double in size photosites on the 6D (24,2/20,2x1.6=1.9), you can expect one stop extra sensitivity? Furthermore you have the abiltity to use faster primes, lets say a 50mm f/1.4 which also gives you an extra stop?

About an TTL flash, Thanks for the advice, i have one and used it quite a lot on my 400D when I only had a kit lens. I will experiment with it again, but there are a lot of situations where a flash won’t be the best or easiest solution.

It's complicated as the older Canon's performance varies quite a bit with ISO. The newer ones are better for that, but the DPAF ones lose a little to sensors with one pixel per pixel (if you see what I mean). The big thing to help at higher ISOs is to have a dual-gain sensor, which Canon haven't done yet. A BSI sensor helps a teeny bit, but Canon don't have them either (but doesn't really matter at APS/FF sizes).

Generally what you need to consider:

* The f-stop is light per unit area. So at the same f-stop bigger sensors receive more light in the same time period. (Or conversely the smaller sensors' data is spread more widely at the final viewing size, which is usually sensor size independent.) Pixel size doesn't matter much as you get more pixels for the same area of the final image (it does if you pixel-peep, but that means zooming into the image more and is meaningless).

* Note faster lenses only help at a particular sensor size if:
(i) They enable you to reduce the ISO (otherwise the camera just alters the shutter speed to keep the light gathered the same).
(ii) You can live with the depth of field you end up with.

* Image noise is mostly noise in the incoming light, if you capture twice the light the percentage of that which is noise goes down by 1.4x.

* The camera noise tends to even out at ISO 800 and above, so not a factor unless you have a dual-gain sensor. It causes the step in the Sony graph here:
http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%206D,Nikon%20D850,Sony%20ILCE-7M3
(BTW the bouncing of the 6D graph is because it only does whole ISO stops in hardware, if you want +/- 1/3 stops it shoots at a whole ISO - i.e. 100, 200, 400, 800... - and pushes or pulls the digital image 1/3 of a stop.)

* Other really minor factors are how permissive the colour filters over the pixels are (mostly they capture a bit either side of the nominal light, so let maybe 40% of the incoming light through, not the 33% you might expect) plus the Quantum Efficiency (QE), namely what percentage of incoming photons get converted to electrons, this varies with colour and most ways of calculating it from DXO (or similar) data have so many error sources I'd ignore them and assume it's about 50% and so you don't care when comparing cameras. Oh and how well the microlenses and light pipes aim the light into the pixel, which again like the previous two I wouldn't worry about.

So in summary at higher ISOs you mostly care about about sensor size or if a dual gain sensor is available (not for Canon). At lower ISOs you want one of the newer Canons with the analogue to digital converters on the sensor, as that reduces the camera's low ISO noise significantly, so for FF 5D4 or R, but not RP as that's the 6D2's sensor.

Oh and don't forget whether it can AF well at low light levels...

I think the 5D4 is probably numerically the best Canon DSLR for low light overall, although the 6D's noise can look more pleasing (very subjective). But a Sony A7rIII is better.

Or just ignore all that and use either Neat Image 8 or DXO Photolab's Prime noise reduction....

Pff, what did i get myself into.

Dr_Jon Veteran Member • Posts: 6,131
Re: Best low light FF Canon DSLR

KLO82 wrote:

Dr_Jon wrote:

KLO82 wrote:

If you believe Canon's claim:

1DX II > 6D II > 5D IV > 6D > 5DS/5DSR

Which claim (really asking)?

Also would someone like to say what they mean by "low light", as low-ISO shots differ from high ISO and people do both in low light (although more the latter).

Oh and at low ISO 5Dsr beats the 6s on a image scale, but not per pixel...

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Canon%20EOS%206D,Canon%20EOS%206D%20Mark%20II
At high ISO it's a wash...

Thanks, although without saying what the quality is like at those ISOs I'm not sure I see it as all that definitive.

 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
Dr_Jon Veteran Member • Posts: 6,131
Re: Light
1

Mako2011 wrote:

Dr_Jon wrote:

I said this below, but I'll point it out again in case it's helpful. A f/1.4 lens gives you the same light as a f/2 lens as the camera will just halve the time the shutter is open to give a correct exposure. It only helps if you can reduce the ISO, which changes how much light you collect to give a correct exposure.

Just so folks are clear. Changing ISO alone has no effect on how much light you can collect. Only a chance in shutter speed and/or aperture and/or scene luminance changes how much light you collect

I'd disagree about the point here - what ISO changes is the upper limit to how much light you can collect, as it amplifies the signal coming from the pixel and so will clip at a lower level. Oh and I should have said:
"It only helps if you can reduce the ISO, which changes how much light you can collect."

There are two things that screw you over:

(1) Collecting more light than a pixel can hold - get white.

(2) Collecting less light than the pixel can hold but setting the ISO amplifier so it amplifies it off the top of the range of the ADC that digitises the signal - get white.

Neither gets you a useful image.

(Somewhat simplified) example of case #2...
Let's say the pixel can collect enough electrons we can make 3V if it's full. We have a digitiser that has an input range of 0-3V and divides that into 16k steps. If at ISO 100 the amplifier just passes the signal through we only get the top reading if the pixel is full. At ISO 200 it doubles the voltage before the digitiser, so anything over a 1/2 full pixel will give you pure white (you collected more light, but the bright bits were off the top of the range after amplification).

So, another (also simplified) example:

* If at ISO 100, and f/2 the camera selected 1/500th as the correct shutter speed it would work that out to put what it saw as the brightest parts of the scene at 60% (made-up number, varies a lot with manufacturer, but they have a number) of the sensor's capacity (to allow highlight headroom).

* If you then shot at f/1.4 in an automatic mode (with the same ISO) the camera would halve the shutter open time so you still get it's 60% full pixel for white. You get the same amount of light collected.

* If you shot at f /1.4 in manual and kept everything the same you'd get 2x the light and overflow the pixel in the brighter areas (like pouring two 60% full buckets into another same-size bucket), so they'd be all white.

ISO example:

* If you shot at ISO 200 f/2 and the camera (or your light meter) said the scene required 1/1000th that would fill the sensor 30% for white (using my 60% made-up number). That's because the ISO amplifier will double the signal before the digitiser, giving 60% of its range (which is what the camera wants).

* Changing to a f/1.4 lens (keeping ISO and shutter speed) will fill the pixel to 60% and then the ISO amplifier doubles it to 120% and you lose the highlights as they are off the top of the digitiser's range (strictly they were off the top of the amplifier's range, so it would limit them).

* BUT (still at f/1.4, with the same shutter speed) changing the ISO to 100 fills the pixel to 60%, as previously, but doesn't amplify it and gives you 2x the light in your data.

* However if you're already at the lowest ISO you can't get more light from a faster lens, just a higher shutter speed and less DoF. (The camera will just do the 60% thing, or you will.)

Somewhat over-simplified, but anything more will lose everyone (if I'm not there already)....and it's basically what happens.

P.S. Note you can over-expose (e.g. ETTR) to reduce the highlight room the camera will use for a "correct exposure" (e.g. replacing 60% with 90%), although the image you see on the camera's screen and the JPEG won't be that useful (as the camera made them assuming 60% was the bright bits). However you get some more light and so less noise, but highlights can get lost (as small areas can be over what the camera took at the bright bits, which is why it went with 60%).

 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
Dr_Jon Veteran Member • Posts: 6,131
Sorry...

steks wrote:

Dr_Jon wrote:

steks wrote:

Thanks for all the reactions, to keep things short, is it fair to say that there is difference in low lIght IQ, but it’s wrong te expect a high improvement when using an comparable aperture.

Lets compare a 30mm f/1.4 on my 77D cropsenor with a 50mm stopped down to f/2.2 on a 6D (30x1.6=48 and 1.4x1.6=2.2) am I correct when saying, that because of the double in size photosites on the 6D (24,2/20,2x1.6=1.9), you can expect one stop extra sensitivity? Furthermore you have the abiltity to use faster primes, lets say a 50mm f/1.4 which also gives you an extra stop?

About an TTL flash, Thanks for the advice, i have one and used it quite a lot on my 400D when I only had a kit lens. I will experiment with it again, but there are a lot of situations where a flash won’t be the best or easiest solution.

It's complicated as the older Canon's performance varies quite a bit with ISO. The newer ones are better for that, but the DPAF ones lose a little to sensors with one pixel per pixel (if you see what I mean). The big thing to help at higher ISOs is to have a dual-gain sensor, which Canon haven't done yet. A BSI sensor helps a teeny bit, but Canon don't have them either (but doesn't really matter at APS/FF sizes).

Generally what you need to consider:

* The f-stop is light per unit area. So at the same f-stop bigger sensors receive more light in the same time period. (Or conversely the smaller sensors' data is spread more widely at the final viewing size, which is usually sensor size independent.) Pixel size doesn't matter much as you get more pixels for the same area of the final image (it does if you pixel-peep, but that means zooming into the image more and is meaningless).

* Note faster lenses only help at a particular sensor size if:
(i) They enable you to reduce the ISO (otherwise the camera just alters the shutter speed to keep the light gathered the same).
(ii) You can live with the depth of field you end up with.

* Image noise is mostly noise in the incoming light, if you capture twice the light the percentage of that which is noise goes down by 1.4x.

* The camera noise tends to even out at ISO 800 and above, so not a factor unless you have a dual-gain sensor. It causes the step in the Sony graph here:
http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%206D,Nikon%20D850,Sony%20ILCE-7M3
(BTW the bouncing of the 6D graph is because it only does whole ISO stops in hardware, if you want +/- 1/3 stops it shoots at a whole ISO - i.e. 100, 200, 400, 800... - and pushes or pulls the digital image 1/3 of a stop.)

* Other really minor factors are how permissive the colour filters over the pixels are (mostly they capture a bit either side of the nominal light, so let maybe 40% of the incoming light through, not the 33% you might expect) plus the Quantum Efficiency (QE), namely what percentage of incoming photons get converted to electrons, this varies with colour and most ways of calculating it from DXO (or similar) data have so many error sources I'd ignore them and assume it's about 50% and so you don't care when comparing cameras. Oh and how well the microlenses and light pipes aim the light into the pixel, which again like the previous two I wouldn't worry about.

So in summary at higher ISOs you mostly care about about sensor size or if a dual gain sensor is available (not for Canon). At lower ISOs you want one of the newer Canons with the analogue to digital converters on the sensor, as that reduces the camera's low ISO noise significantly, so for FF 5D4 or R, but not RP as that's the 6D2's sensor.

Oh and don't forget whether it can AF well at low light levels...

I think the 5D4 is probably numerically the best Canon DSLR for low light overall, although the 6D's noise can look more pleasing (very subjective). But a Sony A7rIII is better.

Or just ignore all that and use either Neat Image 8 or DXO Photolab's Prime noise reduction....

Pff, what did i get myself into.

Sorry, shorter version - IMHO I'd say the 5D4 is best, the 6D is also worth a look.

Sony have some better cameras (the buzz-word is dual-gain) which can take EF lenses via adaptors, but they don't work quite as well as on native cameras (the lenses).

Oh and the FF cameras don't take EF-S lenses if you have them.

Sorry if the long version was a bit.... well... long...

(Edit) There's an even longer one now, about fast lenses and ISO, sorry again.

 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: 26,124
Never

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

I said this below, but I'll point it out again in case it's helpful. A f/1.4 lens gives you the same light as a f/2 lens as the camera will just halve the time the shutter is open to give a correct exposure. It only helps if you can reduce the ISO, which changes how much light you collect to give a correct exposure.

Just so folks are clear. Changing ISO alone has no effect on how much light you can collect. Only a chance in shutter speed and/or aperture and/or scene luminance changes how much light you collect

I'd disagree about the point here - what ISO changes is the upper limit to how much light you can collect, as it amplifies the signal coming from the pixel and so will clip at a lower level.

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

-- 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: 6,131
Re: Never
1

Mako2011 wrote:

Dr_Jon wrote:

Mako2011 wrote:

Dr_Jon wrote:

I said this below, but I'll point it out again in case it's helpful. A f/1.4 lens gives you the same light as a f/2 lens as the camera will just halve the time the shutter is open to give a correct exposure. It only helps if you can reduce the ISO, which changes how much light you collect to give a correct exposure.

Just so folks are clear. Changing ISO alone has no effect on how much light you can collect. Only a chance in shutter speed and/or aperture and/or scene luminance changes how much light you collect

I'd disagree about the point here - what ISO changes is the upper limit to how much light you can collect, as it amplifies the signal coming from the pixel and so will clip at a lower level.

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, as while ISO doesn't change the amount of light the sensor collects it does change whether you can use that for an image. Every time you crank the ISO up another part of the sensor's capacity becomes hidden and you can't measure it and can't use it to make an image. Hence while the pixel's Full Well Capacity remains the same every time you crank the ISO up a stop you can't use the top half of of the previous capacity, hence the Full Well Capacity as seen by the photographer halves.

So yes, if a f/2 lens fills the sensor to 30% a f/1.4 lens (everything else the same) will fill it to 60%, so you get twice as much light captured. But at ISO 200 (assuming ISO 100 = no amplification) they will be reported as 60% and 100% (as the 60% will be clipped by the amplifier). AT ISO 400 they will both be 100% and neither will provide image information.

So, with no ISO amplification you get to use 100% of the pixel's capacity to get image data.
At 2x (e.g. ISO 200) you can only use the bottom 50% of the capacity, anything above that will report as 100%.

ISO - Part of pixel capacity can use

100 - 100%

200 - 50% (anything over 50% will be clipped)

400 - 25%

800 - 12.5%

etc.

This is important as the camera will try to set a "correct exposure" to use a decent amount of the pixel's capacity (I chose 60% as an easy-to-work-with example). Every time you double the ISO that will be 60% of a smaller, usable, part of the ISO 100 capacity. Hence the shutter speed increases to cut the amount of light collected.

For example here are the measured Full Well Capacities for the 6D:

As you said the pixel's Full Well Capacity will always stay the same (probably the ISO 100 value), but my point is that doesn't matter as you can't use that at higher ISOs as every extra ISO stop increase amplifies the top half of it (approximately, they can play with the highlight headroom over increasing ISO) off the top of what can be measured.

P.S. the high read-noise (the noise due to the camera, which is usually swamped by the noise in the light) at lower ISOs is due to noise picked up on the trip between the ISO amplifier (on chip) and the digitiser (off-chip in the 6D). The 5D4 sensor has an on-chip digitiser (well, digitisers) and so its low ISO noise is much less and doesn't fall much as ISO increases.

P.P.S. (For completeness) The reason for the read (camera) noise falling at is at every stop of ISO increase the signal (and the noise) prior to the ISO amplifier is doubled by the amplifier (note that noise is of the order of a couple of electrons). You then add a big bunch of noise going from the ISO amplifier on it's trip off-chip (for a 6D). However that is fixed and with each doubling of the data it becomes a less and less significant part of it.

If you capture 10,000 electrons in a pixel then, at the input to the ISO amplifier, you will have, on average, 100 electrons of noise from the light and a bit less than 2 from the sensor + trip so far (note noise doesn't just add, so 121 and a bit total). If you don't amplify it you will still have 10,000 and add another 25 (-ish, on average) electrons worth of noise on the trip to the digitiser. At ISO 400 it gets amplified 4x to 40,000 electrons, but you still only add 25 of noise, so a much smaller part.

 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: 22,185
Re: Best low light FF Canon DSLR
1

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.

Keyboard shortcuts:
FForum MMy threads