This topic is in the 2009 archive. You can read it and like its posts here. It is read only.
Archive 20095D2 low ISO poor v 1Ds3
p.13 #15D2 low ISO poor v 1Ds3
Am I missing something here
Maybe I should change my nick to Simpleton 
The OP did post non-pushed pictures which shows plenty noise in shadows.
We all know by pushing a stop or two it will bring out more noise, but I though the topic was visible noise on non-pushed pics.
I'll try to capture a similar picture tomorrow, with daylight, using different ISO's and see what I get.
Edited on Feb 06, 2009 at 03:31 PM · View previous versions
Maybe I should change my nick to Simpleton The OP did post non-pushed pictures which shows plenty noise in shadows.
We all know by pushing a stop or two it will bring out more noise, but I though the topic was visible noise on non-pushed pics.
I'll try to capture a similar picture tomorrow, with daylight, using different ISO's and see what I get.
Edited on Feb 06, 2009 at 03:31 PM · View previous versions
Feb 06, 2009 at 03:16 PM
p.13 #25D2 low ISO poor v 1Ds3
Dang. With this hide me button there is hardly anybody left.
Feb 06, 2009 at 03:28 PM
p.13 #35D2 low ISO poor v 1Ds3
WebDog wrote:
Am I missing something here
Yes, you are. But you'd probably have to read a lot of long boring posts - including my long boring posts - to figure it out. ;-)
Dan
Am I missing something here
Yes, you are. But you'd probably have to read a lot of long boring posts - including my long boring posts - to figure it out. ;-)
Dan
Feb 06, 2009 at 05:00 PM
p.13 #45D2 low ISO poor v 1Ds3
WebDog wrote:
Am I missing something here Maybe I should change my nick to Simpleton
The OP did post non-pushed pictures which shows plenty noise in shadows.
We all know by pushing a stop or two it will bring out more noise, but I though the topic was visible noise on non-pushed pics.
I'll try to capture a similar picture tomorrow, with daylight, using different ISO's and see what I get.
All I am trying to say is that compared to 5D, the Mk2 has more noise at ISO 100. The best way to see it is to push the shadows 1.5 or 2 stops. The noise is there without pushing as well, the ISO 100 pictures from 5D were noiseless, the 5D2 , even at 50% are not. But if you expose correctly and don't have to push in post processing, it is not worth of any discussion in my opinion.
Am I missing something here
Maybe I should change my nick to Simpleton The OP did post non-pushed pictures which shows plenty noise in shadows.
We all know by pushing a stop or two it will bring out more noise, but I though the topic was visible noise on non-pushed pics.
I'll try to capture a similar picture tomorrow, with daylight, using different ISO's and see what I get.
All I am trying to say is that compared to 5D, the Mk2 has more noise at ISO 100. The best way to see it is to push the shadows 1.5 or 2 stops. The noise is there without pushing as well, the ISO 100 pictures from 5D were noiseless, the 5D2 , even at 50% are not. But if you expose correctly and don't have to push in post processing, it is not worth of any discussion in my opinion.
Feb 06, 2009 at 05:29 PM
p.13 #55D2 low ISO poor v 1Ds3
Browsing the ten pages I realize people do not really discuss the same issue!
I would be perfectly happy to try to find out if Brainiac's 5D2 do have worse noise in shadows than others (i.e. mine) or if this is what to expect from this camera.
Now there is some good advice in later posts on how to improve on noise in shadows, but that (to me) is another topic, really!
BTW I seems unable to recreate the amount of noise with my 5D2 at ISO 200.
/Mr Simpleton
danmitchell wrote:
Yes, you are. But you'd probably have to read a lot of long boring posts - including my long boring posts - to figure it out. ;-)
Dan
Edited on Feb 06, 2009 at 07:33 PM · View previous versions
I would be perfectly happy to try to find out if Brainiac's 5D2 do have worse noise in shadows than others (i.e. mine) or if this is what to expect from this camera.
Now there is some good advice in later posts on how to improve on noise in shadows, but that (to me) is another topic, really!
BTW I seems unable to recreate the amount of noise with my 5D2 at ISO 200.
/Mr Simpleton
danmitchell wrote:
WebDog wrote
Am I missing something hereDan
Edited on Feb 06, 2009 at 07:33 PM · View previous versions
Feb 06, 2009 at 06:11 PM
p.13 #65D2 low ISO poor v 1Ds3
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor. If I could afford the 1D3s, I would have never considered the 5D2. Unless of course I really needed the video feature, a smaller body, etc. Also, pixel peeping IMHO is no way to compare cameras, especially ones that have comparable sensors/MPs. The 1D series cameras are in a class by themselves. I'm not aware of anyone who buys a 1Ds3 vs. a 5D2 based on (perceived or actual) image quality. I would have happily traded you my beloved 5D2 for your 1Ds3!! 
Feb 06, 2009 at 07:20 PM
p.13 #75D2 low ISO poor v 1Ds3
+ some more thousands eh ? 
ohyeah wrote:
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor. If I could afford the 1D3s, I would have never considered the 5D2. Unless of course I really needed the video feature, a smaller body, etc. Also, pixel peeping IMHO is no way to compare cameras, especially ones that have comparable sensors/MPs. The 1D series cameras are in a class by themselves. I'm not aware of anyone who buys a 1Ds3 vs. a 5D2 based on (perceived or actual) image quality. I would have happily traded you my beloved 5D2 for your...Show more →
ohyeah wrote:
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor. If I could afford the 1D3s, I would have never considered the 5D2. Unless of course I really needed the video feature, a smaller body, etc. Also, pixel peeping IMHO is no way to compare cameras, especially ones that have comparable sensors/MPs. The 1D series cameras are in a class by themselves. I'm not aware of anyone who buys a 1Ds3 vs. a 5D2 based on (perceived or actual) image quality. I would have happily traded you my beloved 5D2 for your...Show more →
Feb 06, 2009 at 07:21 PM
p.13 #85D2 low ISO poor v 1Ds3
BTW I seems unable to recreate the amount of noise with my 5D2 at ISO 200.BTW I seems unable to recreate the amount of noise with my 5D2 at ISO 200.
I can't create that amount of noise in genuine dark gray tones, but when I very heavily push the shadows--bring black areas up to nearly middle gray--I do see that hash-pattern noise in 10x15-inch enlargements.
I still think the point a couple of us have been trying to express is being misunderstood.
Look, I do low-key portraits with dark backgrounds and dark shadows all the time. It's my style, my specialty. I don't get any kind of noise like that in my shadows or my dark tones with a properly exposed image.
By "properly exposed," I mean the highlights that must retain detail are placed at the top of the range of the sensor; when that image is printed, those highlights are so far above the base noise that it can't be seen even in the black shadows.
As I've mentioned before, the contrast curve should be adjusted so that shadows without detail print as the eye would see them--black. That's the way we did it with film, and digital isn't any different because we're talking about the pictorial representation of the scene in the final print. Shadows without detail should not be gray, they should be black.
But a proper exposure will place any actual tone--any area that actually recorded pixels, even a very dark tone--far above the base noise.
Braniac's your results may turn out typical for the way you expose, but with scenes of similar dynamic range, they're not typical for the way I expose.
I can't create that amount of noise in genuine dark gray tones, but when I very heavily push the shadows--bring black areas up to nearly middle gray--I do see that hash-pattern noise in 10x15-inch enlargements.
I still think the point a couple of us have been trying to express is being misunderstood.
Look, I do low-key portraits with dark backgrounds and dark shadows all the time. It's my style, my specialty. I don't get any kind of noise like that in my shadows or my dark tones with a properly exposed image.
By "properly exposed," I mean the highlights that must retain detail are placed at the top of the range of the sensor; when that image is printed, those highlights are so far above the base noise that it can't be seen even in the black shadows.
As I've mentioned before, the contrast curve should be adjusted so that shadows without detail print as the eye would see them--black. That's the way we did it with film, and digital isn't any different because we're talking about the pictorial representation of the scene in the final print. Shadows without detail should not be gray, they should be black.
But a proper exposure will place any actual tone--any area that actually recorded pixels, even a very dark tone--far above the base noise.
Braniac's your results may turn out typical for the way you expose, but with scenes of similar dynamic range, they're not typical for the way I expose.
Feb 06, 2009 at 08:10 PM
p.13 #95D2 low ISO poor v 1Ds3
floris wrote:
Brainiac - I don't know if you missed my question in the midst of all this bickering about what a proper exposure is.. but have you tried other raw converters? It is quite possible that the raw converter is responsible for the banding/blobbing you see. I haven't seen any problems with my 5D2 at high or low iso...
Sorry - I didn't miss it, I just haven't had time to discuss it. I use DPP, latest version, which I think is the fairest way to assess Canon cameras since Canon produces the software. Also I am mean, and having spent much money on a camera that can produce raw files, I like to avoid spending yet more money on raw developers, when I think the camera maker should supply such software, and my experience is that the camera maker generally supplies the best available raw engine.
Brainiac - I don't know if you missed my question in the midst of all this bickering about what a proper exposure is.. but have you tried other raw converters? It is quite possible that the raw converter is responsible for the banding/blobbing you see. I haven't seen any problems with my 5D2 at high or low iso...
Sorry - I didn't miss it, I just haven't had time to discuss it. I use DPP, latest version, which I think is the fairest way to assess Canon cameras since Canon produces the software. Also I am mean, and having spent much money on a camera that can produce raw files, I like to avoid spending yet more money on raw developers, when I think the camera maker should supply such software, and my experience is that the camera maker generally supplies the best available raw engine.
Feb 06, 2009 at 08:41 PM
p.13 #105D2 low ISO poor v 1Ds3
ohyeah wrote:
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor.
If you're talking to me, then I agree with you. I sold my 1Ds3 because I never could afford two of them, or bear to carry two around my neck at once. The lighter weight 5D style camera suits me much better because I can put two around my neck, which is how I like to shoot, and for the price of one 21 megapixel camera (1Ds3) I can have and use two 21 Mpixel cameras. The 5D2 also has video, a smaller charger, a 10.5 Mpixel raw mode, a better monitor, and a face recognition focussing system. On balance, even with the low iso textured noise problem that I am seeing, it is still a better camera for my needs than the 1Ds3.
However, the low iso quality is a surprising disappointment to me, albeit not enough of a shortcoming to make me consider switching to another camera (immediately, at least).
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor.
If you're talking to me, then I agree with you. I sold my 1Ds3 because I never could afford two of them, or bear to carry two around my neck at once. The lighter weight 5D style camera suits me much better because I can put two around my neck, which is how I like to shoot, and for the price of one 21 megapixel camera (1Ds3) I can have and use two 21 Mpixel cameras. The 5D2 also has video, a smaller charger, a 10.5 Mpixel raw mode, a better monitor, and a face recognition focussing system. On balance, even with the low iso textured noise problem that I am seeing, it is still a better camera for my needs than the 1Ds3.
However, the low iso quality is a surprising disappointment to me, albeit not enough of a shortcoming to make me consider switching to another camera (immediately, at least).
Feb 06, 2009 at 08:56 PM
p.13 #115D2 low ISO poor v 1Ds3
brainiac you need to shoot to the right...didn't many people tell you that already ?
http://luminous-landscape.com/tutorials/expose-right.shtml
brainiac wrote:
If you're talking to me, then I agree with you. I sold my 1Ds3 because I never could afford two of them, or bear to carry two around my neck at once. The lighter weight 5D style camera suits me much better because I can put two around my neck, which is how I like to shoot, and for the price of one 21 megapixel camera (1Ds3) I can have and use two 21 Mpixel cameras. The 5D2 also has video, a smaller charger, a 10.5 Mpixel raw mode, a better monitor, and a face recognition focussing system. On balance,...Show more →
http://luminous-landscape.com/tutorials/expose-right.shtml
brainiac wrote:
ohyeah wrote
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor.Feb 06, 2009 at 09:08 PM
p.13 #125D2 low ISO poor v 1Ds3
Thanks! That solves it! Thanks for your brilliant insight! With more advice from people like you I'm really going to get the hang of this photography hobby!
maverick666 wrote:
brainiac you need to shoot to the right...didn't many people tell you that already ?
http://luminous-landscape.com/tutorials/expose-right.shtml
If you're talking to me, then I agree with you. I sold my 1Ds3 because I never could afford two of them, or bear to carry two around my neck at once. The lighter weight 5D style camera suits me much better because I can put two around my neck, which is how I like to shoot, and for the price of one 21 megapixel camera (1Ds3) I can have and use two 21 Mpixel cameras. The 5D2 also has video, a smaller charger, a 10.5 Mpixel raw mode, a better monitor, and a face recognition focussing system. On balance, even with the low iso textured noise problem that I am seeing, it is still a better camera for my needs than the 1Ds3.
However, the low iso quality is a surprising disappointment to me, albeit not enough of a shortcoming to make me consider switching to another camera (immediately, at least).
maverick666 wrote:
brainiac you need to shoot to the right...didn't many people tell you that already ?
http://luminous-landscape.com/tutorials/expose-right.shtml
brainiac wrote
ohyeah wrote
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor.However, the low iso quality is a surprising disappointment to me, albeit not enough of a shortcoming to make me consider switching to another camera (immediately, at least).
Feb 06, 2009 at 09:19 PM
p.13 #135D2 low ISO poor v 1Ds3
Hahahahaha....I guess you're getting sick of shooting to the right huh ? . Sorry didn't mean to upset you.
brainiac wrote:
Thanks! That solves it! Thanks for your brilliant insight! With more advice from people like you I'm really going to get the hang of this photography hobby!
http://luminous-landscape.com/tutorials/expose-right.shtml
If you're talking to me, then I agree with you. I sold my 1Ds3 because I never could afford two of them, or bear to carry two around my neck at once. The lighter weight 5D style camera suits me much better because I can put two around my neck, which is how I like to shoot, and for the price of one 21 megapixel camera (1Ds3) I can have and use two 21 Mpixel cameras. The 5D2 also has video, a smaller charger, a 10.5 Mpixel raw mode, a better monitor, and a face recognition focussing system. On balance, even with the low iso textured noise problem that I am seeing, it is still a better camera for my needs than the 1Ds3.
However, the low iso quality is a surprising disappointment to me, albeit not enough of a shortcoming to make me consider switching to another camera (immediately, at least).
. Sorry didn't mean to upset you.brainiac wrote:
Thanks! That solves it! Thanks for your brilliant insight! With more advice from people like you I'm really going to get the hang of this photography hobby!
maverick666 wrote
brainiac you need to shoot to the right...didn't many people tell you that already ?http://luminous-landscape.com/tutorials/expose-right.shtml
brainiac wrote
ohyeah wrote
I don't know why you'd sell your 1Ds3 just because possible image quality improvements in the 5D2 sensor.However, the low iso quality is a surprising disappointment to me, albeit not enough of a shortcoming to make me consider switching to another camera (immediately, at least).
Feb 06, 2009 at 09:23 PM
p.13 #145D2 low ISO poor v 1Ds3
OK, here is my attemt to replicate the noise.
http://www.webdog.se/tmp/5D2_noise/ISO200.jpg
Picture shot at ISO200 f:5,6 and 1/200. Developed through DPP with zero at both croma and luma noise reduction:
http://www.webdog.se/tmp/5D2_noise/ISO200_dpp_v2.jpg
THEN I increased the Brighness adjustment to +1 see:
http://www.webdog.se/tmp/5D2_noise/ISO200_RAW.jpg
and
http://www.webdog.se/tmp/5D2_noise/ISO200_RGB.jpg
and this is what it looked like:
http://www.webdog.se/tmp/5D2_noise/ISO200_dpp_v3.jpg
I had a go at ISO 100 (cleaner) and ISO 400 (bit more noise) but no real surprises there.
FWIW
/Mr Simpleton
http://www.webdog.se/tmp/5D2_noise/ISO200.jpg
Picture shot at ISO200 f:5,6 and 1/200. Developed through DPP with zero at both croma and luma noise reduction:
http://www.webdog.se/tmp/5D2_noise/ISO200_dpp_v2.jpg
THEN I increased the Brighness adjustment to +1 see:
http://www.webdog.se/tmp/5D2_noise/ISO200_RAW.jpg
and
http://www.webdog.se/tmp/5D2_noise/ISO200_RGB.jpg
and this is what it looked like:
http://www.webdog.se/tmp/5D2_noise/ISO200_dpp_v3.jpg
I had a go at ISO 100 (cleaner) and ISO 400 (bit more noise) but no real surprises there.
FWIW
/Mr Simpleton
Feb 07, 2009 at 07:47 AM
p.13 #155D2 low ISO poor v 1Ds3
Your contrast is set to 5. That's why you have to push in order to see the pattern. If you were attempting to capture a high dynamic range scene, with maximum info in the highlights and shadows, then you would set your contrast to the lowest setting. Notice that the criss-cross pattern is becoming visible with a 1 stop push at iso 200. That, in my opinion, is a disappointing performance compared to the 1Ds3. This issue is going to matter with a picture where you want to maintain highlights and rescue as much information out of the shadows as possible, like this one:
http://cyberphotographer.com/5d2/turnersreachb.jpg
The expose to the right fans should note that there are many situations in which you can't expose further to the right. Off the top of my head here are four:
- dynamic range exceeds the camera's range and overexposing will destroy highlights
- light is very low, aperture is wide open, and shutter speeds are too low to freeze handshake or subject movement
- circumstances are changing very quickly and photographer does not have time to chimp and adjust exposure for each shot (e.g. bride walking down aisle and exiting church into sunlight)
- you accidentally underexposed an image (seriously - this is a real working constraint for any pro)
Exposing to the right is a luxury, not a planned requirement. One of the great features of the 1Ds3 was that one could afford to push by a considerable amount in post if it turned out to be necessary. This seems to be less true of the 5D2 at lower isos.
http://cyberphotographer.com/5d2/turnersreachb.jpg
The expose to the right fans should note that there are many situations in which you can't expose further to the right. Off the top of my head here are four:
- dynamic range exceeds the camera's range and overexposing will destroy highlights
- light is very low, aperture is wide open, and shutter speeds are too low to freeze handshake or subject movement
- circumstances are changing very quickly and photographer does not have time to chimp and adjust exposure for each shot (e.g. bride walking down aisle and exiting church into sunlight)
- you accidentally underexposed an image (seriously - this is a real working constraint for any pro)
Exposing to the right is a luxury, not a planned requirement. One of the great features of the 1Ds3 was that one could afford to push by a considerable amount in post if it turned out to be necessary. This seems to be less true of the 5D2 at lower isos.
Feb 07, 2009 at 08:00 AM
p.13 #165D2 low ISO poor v 1Ds3
I was not really trying to acheive a high dynamic range picture, rather replicate what you had in your first post. With the snow outdoors it is a bit hard to get the brittish dull lightning
Nevermind what's to the right, shouldn't the noise be similar when looking in the shadows??
Had a go setting contrast to -4 (previous set to 0) which is the lowest it will go... but stll does not show the same amount of noise that you show in your pictures.
This is with the above adjustment and brightness set to zero
http://www.webdog.se/tmp/5D2_noise/ISO200_dpp_v4.jpg
Nevermind what's to the right, shouldn't the noise be similar when looking in the shadows??
Had a go setting contrast to -4 (previous set to 0) which is the lowest it will go... but stll does not show the same amount of noise that you show in your pictures.
This is with the above adjustment and brightness set to zero
http://www.webdog.se/tmp/5D2_noise/ISO200_dpp_v4.jpg
Feb 07, 2009 at 08:16 AM
p.13 #175D2 low ISO poor v 1Ds3
My results match WebDog's, but I am using Lightroom.
The one option that does seem to generate some type of pattern that might be interpreted as banding is the "fill light" option. But even then I have to do what looks like close to a 3 stop push. The original values in the area show all 3 channels under 3% (single digits on the 0-255 scale).
The one option that does seem to generate some type of pattern that might be interpreted as banding is the "fill light" option. But even then I have to do what looks like close to a 3 stop push. The original values in the area show all 3 channels under 3% (single digits on the 0-255 scale).
Feb 07, 2009 at 09:01 AM
p.13 #185D2 low ISO poor v 1Ds3
Now do the same with the clasic 5D at ISO200 f:5,6 and 1/200 and look at the shadows, just wondering
Feb 07, 2009 at 09:01 AM
p.13 #205D2 low ISO poor v 1Ds3
brainiac wrote:
Exposing to the right is a luxury, not a planned requirement.
No.
Exposing to the right is the default starting point for optimum exposure with digital sensors. It is virtually always the best choice in the more common situations in which dynamic range is no larger than that of the sensor. Certainly exposing to the right is not a "requirement," but is is a good "plan" unless other factors require a different approach.
(Exposing to the left in low contrast scenes - which is what we saw in the original example - is, as I think most agree, a poor choice, especially if your goal is to minimize noise in the final image
I won't repeat what I wrote before about situations in which we would both agree that exposing to the right would not necessarily be the best choice, or what I wrote about what must happen when the dynamic range of the sensor is exceeded.
I still think that if you would just let go of what seems like an obsession about digital capture noise at low luminosity levels and move on most everything would be fine. The 5D2 is what the 5D2 is, and its low light noise performance is what its low light noise performance is.
An image for consideration - a 100% crop of a section of a 5D photograph that includes luminosities at both ends of the scale - in fact, as the histogram shows, they are somewhat clipped in this very small excerpt from a much larger photograph. I'm not seeing any significant noise even at 100%:
http://www.gdanmitchell.com/images/RhyoliteSchoolWindowsSkySample.jpg
The dynamic range of the original scene exceeded the dynamic range of the sensor on my camera. I "exposed to the right," sacrificing some exposure in the shadowed areas of the window frame shown here so that I could retain some detail in the fully illuminated clouds seen through the window in the full image. In post I compensated by bringing up the values of the shadows a bit - just enough to get some detail there to suggest the wood texture in the darkest shadow areas.
Is there noise in the dark area? Of course. Does it matter in the print? Of course not - it isn't even visible, and if it were it wouldn't make a difference anyway? Did the dynamic range of the scene exceed that of the sensor? Yes. Did I "expose to the right?" Yes. Did I boost shadows in post? Yes.
Take care,
Dan
Exposing to the right is a luxury, not a planned requirement.
No.
Exposing to the right is the default starting point for optimum exposure with digital sensors. It is virtually always the best choice in the more common situations in which dynamic range is no larger than that of the sensor. Certainly exposing to the right is not a "requirement," but is is a good "plan" unless other factors require a different approach.
(Exposing to the left in low contrast scenes - which is what we saw in the original example - is, as I think most agree, a poor choice, especially if your goal is to minimize noise in the final image
I won't repeat what I wrote before about situations in which we would both agree that exposing to the right would not necessarily be the best choice, or what I wrote about what must happen when the dynamic range of the sensor is exceeded.
I still think that if you would just let go of what seems like an obsession about digital capture noise at low luminosity levels and move on most everything would be fine. The 5D2 is what the 5D2 is, and its low light noise performance is what its low light noise performance is.
An image for consideration - a 100% crop of a section of a 5D photograph that includes luminosities at both ends of the scale - in fact, as the histogram shows, they are somewhat clipped in this very small excerpt from a much larger photograph. I'm not seeing any significant noise even at 100%:
http://www.gdanmitchell.com/images/RhyoliteSchoolWindowsSkySample.jpg
The dynamic range of the original scene exceeded the dynamic range of the sensor on my camera. I "exposed to the right," sacrificing some exposure in the shadowed areas of the window frame shown here so that I could retain some detail in the fully illuminated clouds seen through the window in the full image. In post I compensated by bringing up the values of the shadows a bit - just enough to get some detail there to suggest the wood texture in the darkest shadow areas.
Is there noise in the dark area? Of course. Does it matter in the print? Of course not - it isn't even visible, and if it were it wouldn't make a difference anyway? Did the dynamic range of the scene exceed that of the sensor? Yes. Did I "expose to the right?" Yes. Did I boost shadows in post? Yes.
Take care,
Dan
Feb 07, 2009 at 11:42 AM