Canon 80D RAW files available for download. DR improved.

[JohanCruyff]

One question:

If Canon's Low-ISO Dynamic Ranges [totally or almost] catches up with Sony's one, will we all be able to complain ONLY about the presence of the AA-filter on most Canon cameras?
Photography forums will not make sense anymore!
P.S. In any case, with or without the AA-filter, never forget that Canon is doomed!

 

[Sharlin]

One question:

If Canon's Low-ISO Dynamic Ranges [totally or almost] catches up with Sony's one, will we all be able to complain ONLY about the presence of the AA-filter on most Canon cameras?

Oh, low-ISO DR is sooo last year. Given the D5's definitely non-ISO-invariant DR results, now it's again fashionable to complain about high ISO performance instead!

 

[ritholtz]

One question:

If Canon's Low-ISO Dynamic Ranges [totally or almost] catches up with Sony's one, will we all be able to complain ONLY about the presence of the AA-filter on most Canon cameras?

Oh, low-ISO DR is sooo last year. Given the D5's definitely non-ISO-invariant DR results, now it's again fashionable to complain about high ISO performance instead!

Wondering how dxomark going to score 80d. Probably 80d might have better DR than D5.. If it is true, it is going to be funny when someone starts a thread telling how Canon 80d is better than D5 based on DXOmark numbers.

 

[LetTheRightLensIn]

EDIT: well others say it might clip at 16000+ which would be way higher than before and might be enough to make a difference so hang on while I recalcuate.


Hmm well it seems to be mixed. After all this time they did jump DR noticeably, but after all this and if this is the best they can do it's a bit disappointing in that it's still a ways off from Exmor. It closes the gap but probably even less than half the way.

Very roughly it seems like it might have about 2.88 Std Dev read noise for ISO100. To normalize to what DxO does to compare everything fairly regardless of MP count you then apply normalization factor of sqrt(8MP/24.2MP) and multiply that times the 2.88 read noise. Then you also take the high signal-low signal, in the past the high for IOS has often been around 13400 or so for the 14bit cameras and 1024 or 2056 for the low, for this 80D the low is 512, not sure the high but it doesn't change things super much if you are off so just use though numbers. (EDIT: some claim they think they see it peaking at 16383 which is much higher than before and enough to make some difference. That is an oddly high number but I'll use it and hopefully it's correct.) Anyway take that difference and divide by the prior results and then take Log2 of that and there you go and you get 13.2 in this case. It seems the Canon 750D Rebel 24.2 MP got around 12 so it seems around 1 to a little over 1 stop better than their prior best APS-C (and 1.8 compared to the old 50D and prior stuff) and 1-1.2 stop better than 5D2/5D3.

OTOH top Sony/Nikon/etc. APS-C seem to hit 14.5ish or so so you are talking 1 stop better for Canon but still 1.3 stops behind so it's really kind of a shame if this turns out to be all they can manage and it will be like this for another 10 years. It's enough to help for some scenes but still a little short for lots of dappled lighting in forest scenes which Exmor can JUST pull off but this might still leave a little questionable, although better.

As far as banding it appears as if there might be totally zero so that is good.

Hopefully the 1DX 2 or at least 5D4 will have even better sensor tech than tech than the 80D.

I'm really starting to think they should just give up and buy from SOny now that SOny split their sensor division appears willing to sell as much as they can to anyone as soon as they can.

How awesome would a 42.MP high DR camera that can handle oversampled 4k video be combined with a DSLR body from canon for high speed AF, fps, etc. and Canon UI?

Anyway it seems to be the best Canon has ever done and it's a good deal better than say a 50D sensor since it goes from tons of banding to zero (most likely) and maybe jumps almost 1.8 stops better DR (and since the banding is way better, more than 1.8 better in real world usability). OTOH it still seems some ways, another decade at Canon pace, behind catching up to Exmor of today and even a few years ago.

But they may have a better version in the 1DX2/5D4 so who knows for that stuff. And the measurements here are a little rough, it could easily end up +/- 0.33 difference in reality.

 

[nhz]

Hmm well it seems to be mixed. After all this time they did jump DR noticeably, but after all this and if this is the best they can do it's a bit disappointing in that it's still a ways off from Exmor. It closes the gap but probably even less than half the way.
...
OTOH top Sony/Nikon/etc. APS-C seem to hit 14.5ish or so so you are talking 1 stop better for Canon but still 1.3 stops behind so it's really kind of a shame if this turns out to be all they can manage and it will be like this for another 10 years. It's enough to help for some scenes but still a little short for lots of dappled lighting in forest scenes which Exmor can JUST pull off but this might still leave a little questionable, although better.
...
Anyway it seems to be the best Canon has ever done and it's a good deal better than say a 50D sensor since it goes from tons of banding to zero (most likely) and maybe jumps almost 1.8 stops better DR (and since the banding is way better, more than 1.8 better in real world usability). OTOH it still seems some ways, another decade at Canon pace, behind catching up to Exmor of today and even a few years ago.

Yes, about one stop improvement compared to their best APS-C sensor and about 2 stops compared to the APS-C sensors from almost ten years ago ... and still way behind Nikon. The first DPReview tests with pushing shadows by 5-6 stops show about a stop improvement over 70D, and 1.5-2 stops behind Nikon D7200.

I'm using a 450D and I'm not happy with its DR, which according to DXO was the same as my 300D from 2003 (and in my impression in practice even a bit worse, maybe due to different chroma noise pattern / banding). I haven't upgraded in all those years because there was very little improvement in DR / noise performance which for me is the main bottleneck for image quality. Over 95% of my images are taken at 100-800 ISO (action shots mostly 400/800). The 80D shows some obvious improvement, but it this all they can do after almost zero progress in over ten years?

Of course there is more to image quality than just DR in the 100-800 ISO range, and the DXO DR ratings can be debated but they look pretty realistic to me. If I want a better sensor some of the options with their cost are listed below; the Canon options don't look like value for money. If they could offer a similar sensor improvement in a Rebel or SL2 it would look better, but that could take some time. Of course there are other factors in chosing a camera, the new AF system of the 80D etc. should be nice but for me image (sensor) quality is at the top of the list.

Canon 450D DR 10.8 cost 0
Canon 750D DR 12.0 win 1.2 stop, cost €600
Canon 80D DR 13.2 win 2.4 stops, cost €1300
Nikon D5500 DR 14.0 win 3.2 stops, cost €600
Nikon D7200 DR 14.6 win 3.8 stops, cost €1070
Nikon D500 DR xx.x win 4 stops(?), cost €2370

I really want a tilt screen, otherwise I would already have purchased the Nikon D7200. The 750D/760D and D5500 are not a real option for me because they lack focus adjustment and have small viewfinders. I'm even considering the D500 and if its low ISO DR proves as good as the D7200 it might be a very attractive option despite the high price (which might come down a bit once they are in stock?).

 

[ritholtz]

EDIT: well others say it might clip at 16000+ which would be way higher than before and might be enough to make a difference so hang on while I recalcuate.


Hmm well it seems to be mixed. After all this time they did jump DR noticeably, but after all this and if this is the best they can do it's a bit disappointing in that it's still a ways off from Exmor. It closes the gap but probably even less than half the way.

Very roughly it seems like it might have about 2.88 Std Dev read noise for ISO100. To normalize to what DxO does to compare everything fairly regardless of MP count you then apply normalization factor of sqrt(8MP/24.2MP) and multiply that times the 2.88 read noise. Then you also take the high signal-low signal, in the past the high for IOS has often been around 13400 or so for the 14bit cameras and 1024 or 2056 for the low, for this 80D the low is 512, not sure the high but it doesn't change things super much if you are off so just use though numbers. (EDIT: some claim they think they see it peaking at 16383 which is much higher than before and enough to make some difference. That is an oddly high number but I'll use it and hopefully it's correct.) Anyway take that difference and divide by the prior results and then take Log2 of that and there you go and you get 13.2 in this case. It seems the Canon 750D Rebel 24.2 MP got around 12 so it seems around 1 to a little over 1 stop better than their prior best APS-C (and 1.8 compared to the old 50D and prior stuff) and 1-1.2 stop better than 5D2/5D3.

OTOH top Sony/Nikon/etc. APS-C seem to hit 14.5ish or so so you are talking 1 stop better for Canon but still 1.3 stops behind so it's really kind of a shame if this turns out to be all they can manage and it will be like this for another 10 years. It's enough to help for some scenes but still a little short for lots of dappled lighting in forest scenes which Exmor can JUST pull off but this might still leave a little questionable, although better.

As far as banding it appears as if there might be totally zero so that is good.

Hopefully the 1DX 2 or at least 5D4 will have even better sensor tech than tech than the 80D.

I'm really starting to think they should just give up and buy from SOny now that SOny split their sensor division appears willing to sell as much as they can to anyone as soon as they can.

How awesome would a 42.MP high DR camera that can handle oversampled 4k video be combined with a DSLR body from canon for high speed AF, fps, etc. and Canon UI?

Anyway it seems to be the best Canon has ever done and it's a good deal better than say a 50D sensor since it goes from tons of banding to zero (most likely) and maybe jumps almost 1.8 stops better DR (and since the banding is way better, more than 1.8 better in real world usability). OTOH it still seems some ways, another decade at Canon pace, behind catching up to Exmor of today and even a few years ago.

But they may have a better version in the 1DX2/5D4 so who knows for that stuff. And the measurements here are a little rough, it could easily end up +/- 0.33 difference in reality.

It is only Nikons. Latest Sony (a6300) has less DR than 80d. d7200 and d5500 are the ones with highest DR. Between them, d7200 turns gray and dark patches into violet or brown with anything above +3ev push in DPR tests. D5500 is the one actually holds colors also with these extra pushes. D5000 series cameras are amazing. Nikons even have better iq in video. Hopefully Canon improved this in 80d.

Thanks

 

[bmwzimmer]

I really want a tilt screen, otherwise I would already have purchased the Nikon D7200. The 750D/760D and D5500 are not a real option for me because they lack focus adjustment and have small viewfinders. I'm even considering the D500 and if its low ISO DR proves as good as the D7200 it might be a very attractive option despite the high price (which might come down a bit once they are in stock?).
[/quote]

I'd wait a few months after the D500 is released for a D510 with all the bugs worked out.

 

[scyrene]

I'm using a 450D and I'm not happy with its DR, which according to DXO was the same as my 300D from 2003 (and in my impression in practice even a bit worse, maybe due to different chroma noise pattern / banding). I haven't upgraded in all those years because there was very little improvement in DR / noise performance which for me is the main bottleneck for image quality. Over 95% of my images are taken at 100-800 ISO (action shots mostly 400/800). The 80D shows some obvious improvement, but it this all they can do after almost zero progress in over ten years?

Of course there is more to image quality than just DR in the 100-800 ISO range, and the DXO DR ratings can be debated but they look pretty realistic to me. If I want a better sensor some of the options with their cost are listed below; the Canon options don't look like value for money. If they could offer a similar sensor improvement in a Rebel or SL2 it would look better, but that could take some time. Of course there are other factors in chosing a camera, the new AF system of the 80D etc. should be nice but for me image (sensor) quality is at the top of the list.

I really want a tilt screen, otherwise I would already have purchased the Nikon D7200. The 750D/760D and D5500 are not a real option for me because they lack focus adjustment and have small viewfinders. I'm even considering the D500 and if its low ISO DR proves as good as the D7200 it might be a very attractive option despite the high price (which might come down a bit once they are in stock?).

I don't mean to be rude, and of course cost is a limiting factor here, but are you really saying no Canon camera since the 450D is better for action than that one is? I'd have thought AF alone was at least as important as DR, but I don't know what you're shooting precisely...

And tbh if Nikon provide what you want, why not jump ship? Your camera is eight years old. Almost any current model from any producer will be a big step up in most areas.

 

[nhz]

I'm using a 450D and I'm not happy with its DR, which according to DXO was the same as my 300D from 2003 (and in my impression in practice even a bit worse, maybe due to different chroma noise pattern / banding). I haven't upgraded in all those years because there was very little improvement in DR / noise performance which for me is the main bottleneck for image quality. Over 95% of my images are taken at 100-800 ISO (action shots mostly 400/800). The 80D shows some obvious improvement, but it this all they can do after almost zero progress in over ten years?

Of course there is more to image quality than just DR in the 100-800 ISO range, and the DXO DR ratings can be debated but they look pretty realistic to me. If I want a better sensor some of the options with their cost are listed below; the Canon options don't look like value for money. If they could offer a similar sensor improvement in a Rebel or SL2 it would look better, but that could take some time. Of course there are other factors in chosing a camera, the new AF system of the 80D etc. should be nice but for me image (sensor) quality is at the top of the list.

I really want a tilt screen, otherwise I would already have purchased the Nikon D7200. The 750D/760D and D5500 are not a real option for me because they lack focus adjustment and have small viewfinders. I'm even considering the D500 and if its low ISO DR proves as good as the D7200 it might be a very attractive option despite the high price (which might come down a bit once they are in stock?).

I don't mean to be rude, and of course cost is a limiting factor here, but are you really saying no Canon camera since the 450D is better for action than that one is? I'd have thought AF alone was at least as important as DR, but I don't know what you're shooting precisely...

And tbh if Nikon provide what you want, why not jump ship? Your camera is eight years old. Almost any current model from any producer will be a big step up in most areas.

I don't shoot just 'action' like sports photography etc., in that case I would of course have purchased a very different camera like 7D2 or 1DX or some Nikon equivalent. I shoot landscape/cityscape and nature/wildlife, and for me 'wildlife' is mostly flying dragonflies which is a specific type of 'action' photography. I prefer to use one camera for those subjects (in addition to the infrared camera that I also use ...), so I'm looking for the best compromise for those kind of subjects. When buying a new camera I want one that I can use for years.

All Canon APS-C DSLRs from 2015 or earlier have very similar DR/noise quality at low-medium ISO compared to my 450D, it would be hard to notice except in carefully controlled conditions. The 750D/760D is a small step up that one would notice, but it's still only one stop. And besides better sensor I also want focus adjustment and a better viewfinder. The 80D is the first that is significantly improved IMHO.

I'm currently using MF instead of AF for flying dragonflies, because AF is simply way too slow and I think that even with a 7D2 it would be difficult sometimes because dragonflies are very fast and erratic flyers. The 7D2 is also a heavy and expensive camera (in Europe) and it doesn't have a tilt screen which I really want, so for me it's not an option at all. Canon FF is not an option either because those are relatively big/heavy (or a bit crippled like the 6D), don't have a tilt screen, have lower pixel density (bad for wildlife, except the even heavier 5DS) and offer by definition only one stop in IQ gain over APS-C with similar sensor technology.

The Nikons mentioned above all are more attractive to me than the Canons, but of course in addition to the cost of the body there is the cost of upgrading your lenses which is a big factor. So I can't buy just e.g. a D7200 body and be done, if that was the case it would be an easy decision ...

 

[scyrene]

I'm using a 450D and I'm not happy with its DR, which according to DXO was the same as my 300D from 2003 (and in my impression in practice even a bit worse, maybe due to different chroma noise pattern / banding). I haven't upgraded in all those years because there was very little improvement in DR / noise performance which for me is the main bottleneck for image quality. Over 95% of my images are taken at 100-800 ISO (action shots mostly 400/800). The 80D shows some obvious improvement, but it this all they can do after almost zero progress in over ten years?

Of course there is more to image quality than just DR in the 100-800 ISO range, and the DXO DR ratings can be debated but they look pretty realistic to me. If I want a better sensor some of the options with their cost are listed below; the Canon options don't look like value for money. If they could offer a similar sensor improvement in a Rebel or SL2 it would look better, but that could take some time. Of course there are other factors in chosing a camera, the new AF system of the 80D etc. should be nice but for me image (sensor) quality is at the top of the list.

I really want a tilt screen, otherwise I would already have purchased the Nikon D7200. The 750D/760D and D5500 are not a real option for me because they lack focus adjustment and have small viewfinders. I'm even considering the D500 and if its low ISO DR proves as good as the D7200 it might be a very attractive option despite the high price (which might come down a bit once they are in stock?).

I don't mean to be rude, and of course cost is a limiting factor here, but are you really saying no Canon camera since the 450D is better for action than that one is? I'd have thought AF alone was at least as important as DR, but I don't know what you're shooting precisely...

And tbh if Nikon provide what you want, why not jump ship? Your camera is eight years old. Almost any current model from any producer will be a big step up in most areas.

I don't shoot just 'action' like sports photography etc., in that case I would of course have purchased a very different camera like 7D2 or 1DX or some Nikon equivalent. I shoot landscape/cityscape and nature/wildlife, and for me 'wildlife' is mostly flying dragonflies which is a specific type of 'action' photography. I prefer to use one camera for those subjects (in addition to the infrared camera that I also use ...), so I'm looking for the best compromise for those kind of subjects. When buying a new camera I want one that I can use for years.

All Canon APS-C DSLRs from 2015 or earlier have very similar DR/noise quality at low-medium ISO compared to my 450D, it would be hard to notice except in carefully controlled conditions. The 750D/760D is a small step up that one would notice, but it's still only one stop. And besides better sensor I also want focus adjustment and a better viewfinder. The 80D is the first that is significantly improved IMHO.

I'm currently using MF instead of AF for flying dragonflies, because AF is simply way too slow and I think that even with a 7D2 it would be difficult sometimes because dragonflies are very fast and erratic flyers. The 7D2 is also a heavy and expensive camera (in Europe) and it doesn't have a tilt screen which I really want, so for me it's not an option at all. Canon FF is not an option either because those are relatively big/heavy (or a bit crippled like the 6D), don't have a tilt screen, have lower pixel density (bad for wildlife, except the even heavier 5DS) and offer by definition only one stop in IQ gain over APS-C with similar sensor technology.

The Nikons mentioned above all are more attractive to me than the Canons, but of course in addition to the cost of the body there is the cost of upgrading your lenses which is a big factor. So I can't buy just e.g. a D7200 body and be done, if that was the case it would be an easy decision ...

Ah, okay. Dragonfly photography is specialised, and I have respect for anyone who can do it - I've all but given up on them! I don't think it's fair to say full frame isn't good enough for wildlife due to the pixel density - many dedicated wildlife photographers who can afford them use either a 1Dx or 5DIII (or other brand equivalents). Of course, small and light, high pixel density, and a flip screen - I can see why you've been reluctant to upgrade. Still, I think you'd find there would be a big step up in quality from the 450D (to, say, the 70D) despite what spec sheets might suggest.

Incidentally, the cost of upgrading lenses is surely the same whether sticking with Canon or changing to Nikon - more or less? You can sell most lenses and get a lot of the value back. That's been my experience, anyway. Bodies, not so much.

 

[nhz]

Ah, okay. Dragonfly photography is specialised, and I have respect for anyone who can do it - I've all but given up on them! I don't think it's fair to say full frame isn't good enough for wildlife due to the pixel density - many dedicated wildlife photographers who can afford them use either a 1Dx or 5DIII (or other brand equivalents). Of course, small and light, high pixel density, and a flip screen - I can see why you've been reluctant to upgrade. Still, I think you'd find there would be a big step up in quality from the 450D (to, say, the 70D) despite what spec sheets might suggest.

Incidentally, the cost of upgrading lenses is surely the same whether sticking with Canon or changing to Nikon - more or less? You can sell most lenses and get a lot of the value back. That's been my experience, anyway. Bodies, not so much.

For wildlife photography in low light conditions a FF camera is in general still the better choice. And it depends on how one travels, if you travel by car and can bring all the gear you need near your photography spot it's easier to use heavy equipment. I travel by bike and walking so I want light gear, also because dragonflies move fast. A FF camera with by definition much bigger lens (for same magnification) would not work for this because it is far more difficult to hold and quickly point/track. Apart from the sensor this is one reason for looking at Nikon, their 4/300PF lens is much lighter than my Canon 4/300IS and even if I'm chosing a relatively heavy Nikon body like D500 the combo is still lighter than my current Rebel with the Canon 300mm.

I have tried some of the newer Rebels and looked at files from the 70D, and IMHO the image quality in the 200-800 ISO range is pretty close, about what the DXO numbers suggest.

The second hand lens market in my country doesn't work very well IMHO, you take a 20-25% loss on most lenses which adds up. Ebay doesn't work for lens sales here, and selling internationally is very difficult because our purchase prices include VAT; in general that would only work for professionals who can deduct the VAT when purchasing their lenses. I would prefer to stay with Canon because I like their lenses better, but the half-hearted improvement in 80D image quality and its high initial price doesn't make that very attractive.

 

[scyrene]

Ah, okay. Dragonfly photography is specialised, and I have respect for anyone who can do it - I've all but given up on them! I don't think it's fair to say full frame isn't good enough for wildlife due to the pixel density - many dedicated wildlife photographers who can afford them use either a 1Dx or 5DIII (or other brand equivalents). Of course, small and light, high pixel density, and a flip screen - I can see why you've been reluctant to upgrade. Still, I think you'd find there would be a big step up in quality from the 450D (to, say, the 70D) despite what spec sheets might suggest.

Incidentally, the cost of upgrading lenses is surely the same whether sticking with Canon or changing to Nikon - more or less? You can sell most lenses and get a lot of the value back. That's been my experience, anyway. Bodies, not so much.

For wildlife photography in low light conditions a FF camera is in general still the better choice. And it depends on how one travels, if you travel by car and can bring all the gear you need near your photography spot it's easier to use heavy equipment. I travel by bike and walking so I want light gear, also because dragonflies move fast. A FF camera with by definition much bigger lens (for same magnification) would not work for this because it is far more difficult to hold and quickly point/track. Apart from the sensor this is one reason for looking at Nikon, their 4/300PF lens is much lighter than my Canon 4/300IS and even if I'm chosing a relatively heavy Nikon body like D500 the combo is still lighter than my current Rebel with the Canon 300mm.

I have tried some of the newer Rebels and looked at files from the 70D, and IMHO the image quality in the 200-800 ISO range is pretty close, about what the DXO numbers suggest.

The second hand lens market in my country doesn't work very well IMHO, you take a 20-25% loss on most lenses which adds up. Ebay doesn't work for lens sales here, and selling internationally is very difficult because our purchase prices include VAT; in general that would only work for professionals who can deduct the VAT when purchasing their lenses. I would prefer to stay with Canon because I like their lenses better, but the half-hearted improvement in 80D image quality and its high initial price doesn't make that very attractive.

That's fair enough. I don't fully agree with your conclusions (for instance, I don't think the difference between the 80D's sensor and that of equivalent Nikons is going to make a practical difference to almost any photographs), but regardless I hope you find a camera that works for you, whoever makes it

Incidentally, I'm a walking/public transport nature photography person too - it definitely shapes what is possible (although I went with big and heavy gear anyway)

 

[nhz]

That's fair enough. I don't fully agree with your conclusions (for instance, I don't think the difference between the 80D's sensor and that of equivalent Nikons is going to make a practical difference to almost any photographs), but regardless I hope you find a camera that works for you, whoever makes it

I'm not sure the difference with the equivalent Nikons is important in practice either, it's more that I would like some more headroom just in case (and I would regret my purchase if next year they have another camera with 1-2 stops more DR ...). And it also depends on the RAW converter used, PP etc. For the dragonflies the 80D sensor should be fine (assuming it performs well at 400-1600 ISO, we should know that soon enough), for landscape images I would really prefer DR of 14-15 stops as it would make life much easier.

 

[honza_lin]

I did an ACR RAW conversion of the 80D (left) and D7200 (right) at ISO 100.

Pushed 5 stops, equal WB values, no sharpening, everything else left as default.

Someone over on DPR that knows vastly more about sensor and data tech than I do, seems to think the measured DR is lower for the 80D because of the way Canon handles the file data, but in actual use, is probably pretty close to Nikon's best. It will be interesting to see how DXO measures the 80D if it is found that the actual files from it offer the same level of DR as Nikon in real-world use.

Upload you please original raw file this photos? .. Canon 80D and Nikon D7200. Thank you very much

 

[K-amps]

One question:

If Canon's Low-ISO Dynamic Ranges [totally or almost] catches up with Sony's one, will we all be able to complain ONLY about the presence of the AA-filter on most Canon cameras?

Oh, low-ISO DR is sooo last year. Given the D5's definitely non-ISO-invariant DR results, now it's again fashionable to complain about high ISO performance instead!

Wondering how dxomark going to score 80d. Probably 80d might have better DR than D5.. If it is true, it is going to be funny when someone starts a thread telling how Canon 80d is better than D5 based on DXOmark numbers.

I suspect DxO gives a lot of weightage to color depth, which in Canon sensors is still is behind Sonikon... so DxO will be more than happy not to tweak their algorithm to keep the 80D behind. ;D