Well I'm not surprised the moon is hopelessly blown, it is illuminated by direct sunlight, while the rest of your scene is illuminated by moonlight or dusk. I'd be extremely surprised if any Sony sensor could do what you want here, seems like an unreasonable expectation. So I'm still waiting for a real example...
I cannot disagree with this however, I think we can agree with a wider DR, it would be closer to what Joel actually saw.
Moderator Journeys/Sports/Big Picture :: Need some help with dgrin?
Well I'm not surprised the moon is hopelessly blown, it is illuminated by direct sunlight, while the rest of your scene is illuminated by moonlight or dusk. I'd be extremely surprised if any Sony sensor could do what you want here, seems like an unreasonable expectation. So I'm still waiting for a real example...
A real example, lol? I could show you a billion sunrise/sunsets I've shot that would benefit by a wider dynamic range. But I'm pretty certain you'd reject them as well, just as you rejected David's examples. I think we're just going to have to agree to disagree on this one, Jack. It's all good.
I cannot disagree with this however, I think we can agree with a wider DR, it would be closer to what Joel actually saw.
Thanks, Ian. That's really the issue. A sensor with wider DR would allow an exposure somewhere between those two that would have given something closer to what I saw.
I mean no disrespect, I guess I would like to see a real-life comparison between Sony/Nikon and Canon sensors in this department. And not images of the inside of someone's office. Like, here's a sunset, here's the best I could make it look with Canon, here's the best I could make it look with Nikon.
I rejected David's example because he's using a known inferior raw converter which is bad at extracting DR. His example actually kind of supports my point, when he turns down the highlights and turns on Linear, suddenly the white sky becomes a beautiful blue with nice clouds, and the road turns black. I'm pretty sure LR could give you that and the foreground all properly exposed in one image.
-Jack
An "accurate" reproduction of a scene and a good photograph are often two different things.
Thanks, Ian. That's really the issue. A sensor with wider DR would allow an exposure somewhere between those two that would have given something closer to what I saw.
I still want to see how this new metering thing is gonna work. That could be interesting. Wonder who will be first with a real review?
Moderator Journeys/Sports/Big Picture :: Need some help with dgrin?
I mean no disrespect, I guess I would like to see a real-life comparison between Sony/Nikon and Canon sensors in this department. And not images of the inside of someone's office. Like, here's a sunset, here's the best I could make it look with Canon, here's the best I could make it look with Nikon.
I rejected David's example because he's using a known inferior raw converter which is bad at extracting DR. His example actually kind of supports my point, when he turns down the highlights and turns on Linear, suddenly the white sky becomes a beautiful blue with nice clouds, and the road turns black. I'm pretty sure LR could give you that and the foreground all properly exposed in one image.
I think we're looking for SOC DR. If Canon wanted a home run, they would have addressed that in this camera-maybe they have indirectly, idk.
Moderator Journeys/Sports/Big Picture :: Need some help with dgrin?
Here's one. I'm not shooting into the sun, in fact, it is behind me and the building. I was trying to get an exposure that would capture both the sky (there is a blue sky with clouds) and the foreground subjects. A little too dark in the foreground - though that is easily fixable. The sky, however, is toast. Clicking the Linear view, there is some sky there, but the clouds are still blown out. I do realize that the clouds are front-lit, but that's what clouds do in the sky. A couple more stops of DR would probably have got it all in (yet with a little more work in post).
And here is the same file with the EV reduced to -0.10, highlights are set to -5, and Linear is checked. There's the sky! But the clouds are still at 255. It just can't handle an exposure of medium low light foreground and sky at the same time, even if I shoot opposite the sun. I find that lacking.
I could shoot all scenes with these not-so-tough conditions as HDR's. But I shouldn't have to now or in the near future. The sensor tech is mostly available, but Canon isn't using it. This was from a 5D3, made two years ago and shot last year. I can't change the sensor in that cam, but the replacement had better the heck be able to do shots like this in one click.
The latest version of Magic Lantern firmware overlay has a feature called, "Dual ISO", which can capture around 14 EVs of sensor data in a single exposure.
Here are a couple of examples from the same image file (links only because these are not my images):
Yes, Magic Lantern and Dual-ISO are some extra work, but it's true tone blending of a broader/larger dynamic range for an imager and (more importantly) an amplifier and image processor not otherwise designed to produce this latitude and dynamic range.
Anyone with a suitable Canon camera (many of the midrange and higher end dSLR bodies work well) may wish to try. Start here:
I mean no disrespect, I guess I would like to see a real-life comparison between Sony/Nikon and Canon sensors in this department. And not images of the inside of someone's office. Like, here's a sunset, here's the best I could make it look with Canon, here's the best I could make it look with Nikon.
I rejected David's example because he's using a known inferior raw converter which is bad at extracting DR. His example actually kind of supports my point, when he turns down the highlights and turns on Linear, suddenly the white sky becomes a beautiful blue with nice clouds, and the road turns black. I'm pretty sure LR could give you that and the foreground all properly exposed in one image.
I don't have access to it, but Marc Muench did extensive testing before moving from Canon to Nikon, and his real-world testing showed that there's 1.5 stops additional dynamic range in the shadows of the Nikons (at the time the D800, the D810 only performs better) than there was in the Canon (5DMKIII). You try to boost the shadows on the Canons and they've got terrible noise in the shadows that's not there in the Nikon.
He's got a new blogpost due to be published soon where he's testing low light capabilities of various cameras against the Nikon and didn't bother including the current Canon cameras because he's already dismissed them as poorer performers in low light.
The ONLY way to test this is to shoot the same scene with the same settings with different bodies, and check the details, apples to apples.
All of these cameras will take good pictures and the Canon cameras are still amazing cameras. But you can't deny the value of having increased dynamic range or improved low-light performance. The new 5DS/R bodies look to be poor performers in both respects. The rumored 5DMKIV is supposed to have improved dynamic range over the 5DMKIII with a limited increase in resolution.
Canon's choice to cram more pixels in this camera leads to necessary compromises, and I find it interesting that they announced it ahead of the 5DMKIV. My guess is that they figure lots of folks will jump on the wow factor of having a 50MPX body, so they're holding off on the announcement of the MKIV to give the S/R lineup a better jump start. But I tell you, if the the THREE cameras were launched at the same time, the MKIV would be the much more appealing option. The S/R bodies are way pricey.
David, with respect, can you explain to me the importance of low noise in the lower tones, when one is shooting 3-5 frame exposure brackets at low ISO on a tripod?
I can understand the desire for larger dynamic range if one is only using a single frame for large images, perhaps, but almost all of Marc's images are shot with multi-frame brackets, aren't they?
I am one of the Canon shooters who has not switched to Nikon, and probably will not because of my lens closet full of Canon glass which I am pretty happy with. And I have not decided yet to go the Sony body route yet either.
David, with respect, can you explain to me the importance of low noise in the lower tones, when one is shooting 3-5 frame exposure brackets at low ISO on a tripod?
I can understand the desire for larger dynamic range if one is only using a single frame for large images, perhaps, but almost all of Marc's images are shot with multi-frame brackets, aren't they?
I am one of the Canon shooters who has not switched to Nikon, and probably will not because of my lens closet full of Canon glass which I am pretty happy with. And I have not decided yet to go the Sony body route yet either.
I am interested in the 5DMkIV though.
Well, I can't speak for Marc, but just because you can avoid some of these issues by bracketing doesn't mean you wouldn't appreciate a few more stops of dynamic range. By the numbers the Nikon beats Canon by 3 stops, and in real world testing Marc found over 1.5 stops in the shadows alone.
My point isn't that anyone should ditch Canon, but that Canon doesn't yet have a camera that can do what the Nikon does, which is to find the sweet spot for high resolution and low-light capabilities at a good price. The 5DSR is $1200 more than the D810 and all it really adds as far as IQ goes (from what I can see) is more pixels. The question I have is whether those pixels will be any good with the sensors so small. The 5DMKIV is supposed to have better low light capabilities and a modest resolution increase.
It's too soon to pass judgment on any of these Canon cameras, I'm just saying it's interesting, the choices they're making, and by extension the choices they're forcing their customers to make.
They also made this same choice with the 11-24 vs. the 14-24, forsaking a stop of aperture for a few more mm of focal length.
I don't have access to it, but Marc Muench did extensive testing before moving from Canon to Nikon, and his real-world testing showed that there's 1.5 stops additional dynamic range in the shadows of the Nikons (at the time the D800, the D810 only performs better) than there was in the Canon (5DMKIII). You try to boost the shadows on the Canons and they've got terrible noise in the shadows that's not there in the Nikon.
He's got a new blogpost due to be published soon where he's testing low light capabilities of various cameras against the Nikon and didn't bother including the current Canon cameras because he's already dismissed them as poorer performers in low light.
That's funny, that I forgot that we pushed that blogpost.
That thread is probably what I'm talking about. I wonder what happened to the links? Maybe you guys turned off HTML in posts or Smugmug changed something?
And yes, it's the third video where he shows the ugly Canon shadows next to the perty Nikon shadows.
Specifically, this frame, where both cameras are shot at ISO 200 and boosted 3 stops in LR:
EDIT: And that improvement in quality is available in the D610, as well, which has similar but slightly higher resolution, at about 1/2 the cost. There's a good quick overview here.
Even the OM-D E-m5 sensor has better "grain" (i.e, random, not corduroy) in the dark zones than my 5D mark II. I understand the mark III is better though.
If you don't like ISO performance just use a noise reduction program, hard to believe but software is getting better
Mama, where's the noise ?
ISO 5000 on an old Canon 60D
0
Matthew SavilleRegistered Users, Retired ModPosts: 3,352Major grins
edited February 15, 2015
I think many people, including both megapixel fanboys AND dynamic range fanboys, are missing the point really.
Some complaints I've heard have been along the lines of, "well how come you all praised Nikon for raising the megapixel bar to 36 MP, yet now Canon gets slammed for raising the megapixel bar again? That's a double standard..."
Other complaints are more along the lines of, "why is dynamic range now the only yard stick that matters to all of photography? How come all of a sudden certain DSLRs are completely un-usable for landscape photography, when they still offer far more dynamic range than the films that were still very popular just a decade ago?"
Both of these are valid arguments, and anyone with a clear mind and a good sense of fairness will agree that, as far as an average photographer is concerned, all of this is being blown way out of proportion.
However in my opinion, that still isn't the point. The point is that Canon could have made a ~38 MP sensor with ~15 stops of DR instead, and it would have been a hundred times more well-received by, well, everybody!
In other words, what landscape or similar photographer in their right mind would trade 2+ stops of dynamic range, AND 1-2 stops of high ISO performance, ...in exchange for a measly 10-12 extra megapixels? Or, what astro-landscape photographer wouldn't rather have a D810 and a 14-24 f/2.8, instead of a 5DsR and a 11-24?
So yes, all modern DSLRs are jaw-droppingly capable, by yesterday's standards. But Canon shouldn't use that as an excuse to leave their dynamic range flat-lined where it has been for the past decade or so. They need to step it up, but they are clearly more interested in creating "trophy" bodies and lenses instead.
BTW if you'd like to see a whole slew of situations in which I'd have serious trouble with Canon's dynamic range, check out either my D750 review or my D810 review, both of which include numerous scenes that I would have simply not been able to capture in a single exposure on a Canon sensor.
Or there's this shot, a single exposure... (Actually a focus stack, but all frames were the same exposure and that still makes my life a whole lot easier!)
There's a trade-off between number of pixels and DR. For a fixed sensor size, the more pixels you have the less the DR (all other things being equal). The physics behind this is somewhat complicated, but there's a simple way to get an intuitive feel for why this is true. Each pixel on the sensor is a photon bucket. The more pixels, the smaller the bucket size. Since dynamic range is essentially the difference between the smallest number of photons the camera can detect in the bucket and the largest number of photons that can fit in the bucket, a smaller bucket necessarily has less DR.
Added to that is the signal-to-noise ratio of the detector, which goes down with pixel size.
Comments
I cannot disagree with this however, I think we can agree with a wider DR, it would be closer to what Joel actually saw.
Link to my Smugmug site
Link to my Smugmug site
I rejected David's example because he's using a known inferior raw converter which is bad at extracting DR. His example actually kind of supports my point, when he turns down the highlights and turns on Linear, suddenly the white sky becomes a beautiful blue with nice clouds, and the road turns black. I'm pretty sure LR could give you that and the foreground all properly exposed in one image.
An "accurate" reproduction of a scene and a good photograph are often two different things.
I still want to see how this new metering thing is gonna work. That could be interesting. Wonder who will be first with a real review?
I think we're looking for SOC DR. If Canon wanted a home run, they would have addressed that in this camera-maybe they have indirectly, idk.
Straight out of the camera JPEGs? Who cares about those?
An "accurate" reproduction of a scene and a good photograph are often two different things.
The latest version of Magic Lantern firmware overlay has a feature called, "Dual ISO", which can capture around 14 EVs of sensor data in a single exposure.
Here are a couple of examples from the same image file (links only because these are not my images):
First, the native data with base exposure for the background:
http://s8.postimg.org/fqeio2xs5/011.jpg
... and the Dual-ISO image:
http://s8.postimg.org/r5azspa4l/012.jpg
(Above images attributed to Magic Lantern forum user "Danne".)
Another image using Dual-ISO:
http://dcmakecreative.altervista.org/ML/dual_iso_test.jpg
(Above image attributed to Magic Lantern forum user "JackDaniel412".)
Yes, Magic Lantern and Dual-ISO are some extra work, but it's true tone blending of a broader/larger dynamic range for an imager and (more importantly) an amplifier and image processor not otherwise designed to produce this latitude and dynamic range.
Anyone with a suitable Canon camera (many of the midrange and higher end dSLR bodies work well) may wish to try. Start here:
http://www.magiclantern.fm/forum/index.php?topic=7139.0
Many more sample images here:
http://www.magiclantern.fm/forum/index.php?topic=7402
http://www.flickr.com/search/?q=dualiso
Moderator of the Cameras and Accessories forums
I don't have access to it, but Marc Muench did extensive testing before moving from Canon to Nikon, and his real-world testing showed that there's 1.5 stops additional dynamic range in the shadows of the Nikons (at the time the D800, the D810 only performs better) than there was in the Canon (5DMKIII). You try to boost the shadows on the Canons and they've got terrible noise in the shadows that's not there in the Nikon.
He's got a new blogpost due to be published soon where he's testing low light capabilities of various cameras against the Nikon and didn't bother including the current Canon cameras because he's already dismissed them as poorer performers in low light.
The ONLY way to test this is to shoot the same scene with the same settings with different bodies, and check the details, apples to apples.
All of these cameras will take good pictures and the Canon cameras are still amazing cameras. But you can't deny the value of having increased dynamic range or improved low-light performance. The new 5DS/R bodies look to be poor performers in both respects. The rumored 5DMKIV is supposed to have improved dynamic range over the 5DMKIII with a limited increase in resolution.
Canon's choice to cram more pixels in this camera leads to necessary compromises, and I find it interesting that they announced it ahead of the 5DMKIV. My guess is that they figure lots of folks will jump on the wow factor of having a 50MPX body, so they're holding off on the announcement of the MKIV to give the S/R lineup a better jump start. But I tell you, if the the THREE cameras were launched at the same time, the MKIV would be the much more appealing option. The S/R bodies are way pricey.
Dgrin FAQ | Me | Workshops
I can understand the desire for larger dynamic range if one is only using a single frame for large images, perhaps, but almost all of Marc's images are shot with multi-frame brackets, aren't they?
I am one of the Canon shooters who has not switched to Nikon, and probably will not because of my lens closet full of Canon glass which I am pretty happy with. And I have not decided yet to go the Sony body route yet either.
I am interested in the 5DMkIV though.
Moderator of the Technique Forum and Finishing School on Dgrin
Who said anything about anything but getting the exposure correct in the camera?
Nikon just announced a new ....wait for it.........D810A "specialty" camera, designed for astrophotography.
The price of this new baby is........wait for it...............$3799.95. Right in there with the Canon pricing.
Hummmm ...a fine mess you've gotten us into Stanly.
Sam
"You miss 100% of the shots you don't take" - Wayne Gretzky
Well, I can't speak for Marc, but just because you can avoid some of these issues by bracketing doesn't mean you wouldn't appreciate a few more stops of dynamic range. By the numbers the Nikon beats Canon by 3 stops, and in real world testing Marc found over 1.5 stops in the shadows alone.
My point isn't that anyone should ditch Canon, but that Canon doesn't yet have a camera that can do what the Nikon does, which is to find the sweet spot for high resolution and low-light capabilities at a good price. The 5DSR is $1200 more than the D810 and all it really adds as far as IQ goes (from what I can see) is more pixels. The question I have is whether those pixels will be any good with the sensors so small. The 5DMKIV is supposed to have better low light capabilities and a modest resolution increase.
It's too soon to pass judgment on any of these Canon cameras, I'm just saying it's interesting, the choices they're making, and by extension the choices they're forcing their customers to make.
They also made this same choice with the 11-24 vs. the 14-24, forsaking a stop of aperture for a few more mm of focal length.
Dgrin FAQ | Me | Workshops
Thanks!
Jim, I'll also note that more and more Marc is looking for ways to get it all in one capture, using the merge to 32bit plugin in Lightroom.
Dgrin FAQ | Me | Workshops
Marc's blog about the differences is up.
Also, is this the thread you were talking about in the first sentence above?
"You miss 100% of the shots you don't take" - Wayne Gretzky
That's funny, that I forgot that we pushed that blogpost.
That thread is probably what I'm talking about. I wonder what happened to the links? Maybe you guys turned off HTML in posts or Smugmug changed something?
Dgrin FAQ | Me | Workshops
Specifically, this frame, where both cameras are shot at ISO 200 and boosted 3 stops in LR:
EDIT: And that improvement in quality is available in the D610, as well, which has similar but slightly higher resolution, at about 1/2 the cost. There's a good quick overview here.
Dgrin FAQ | Me | Workshops
If you don't like ISO performance just use a noise reduction program, hard to believe but software is getting better
Mama, where's the noise ?
ISO 5000 on an old Canon 60D
Some complaints I've heard have been along the lines of, "well how come you all praised Nikon for raising the megapixel bar to 36 MP, yet now Canon gets slammed for raising the megapixel bar again? That's a double standard..."
Other complaints are more along the lines of, "why is dynamic range now the only yard stick that matters to all of photography? How come all of a sudden certain DSLRs are completely un-usable for landscape photography, when they still offer far more dynamic range than the films that were still very popular just a decade ago?"
Both of these are valid arguments, and anyone with a clear mind and a good sense of fairness will agree that, as far as an average photographer is concerned, all of this is being blown way out of proportion.
However in my opinion, that still isn't the point. The point is that Canon could have made a ~38 MP sensor with ~15 stops of DR instead, and it would have been a hundred times more well-received by, well, everybody!
In other words, what landscape or similar photographer in their right mind would trade 2+ stops of dynamic range, AND 1-2 stops of high ISO performance, ...in exchange for a measly 10-12 extra megapixels? Or, what astro-landscape photographer wouldn't rather have a D810 and a 14-24 f/2.8, instead of a 5DsR and a 11-24?
So yes, all modern DSLRs are jaw-droppingly capable, by yesterday's standards. But Canon shouldn't use that as an excuse to leave their dynamic range flat-lined where it has been for the past decade or so. They need to step it up, but they are clearly more interested in creating "trophy" bodies and lenses instead.
BTW if you'd like to see a whole slew of situations in which I'd have serious trouble with Canon's dynamic range, check out either my D750 review or my D810 review, both of which include numerous scenes that I would have simply not been able to capture in a single exposure on a Canon sensor.
http://www.slrlounge.com/nikon-d810-review/
http://www.slrlounge.com/nikon-d750-best-wedding-dslr-ever/
Or there's this shot, a single exposure... (Actually a focus stack, but all frames were the same exposure and that still makes my life a whole lot easier!)
My SmugMug Portfolio • My Astro-Landscape Photo Blog • Dgrin Weddings Forum
right, it's all about "IQ" - image quality
only 18 megapickles !!
Added to that is the signal-to-noise ratio of the detector, which goes down with pixel size.