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.3 #15D2 low ISO poor v 1Ds3
brainiac wrote:
Sadly I have just sold my 1Ds3. I say sadly because I did not expect my new 5D2 to have significantly worse performance at low ISOs, but it does seem to.
This is an ISO 200 HTP file, which is the same as an underexposed ISO 100 image. Having ALO activated only exacerbates the problem. I'm also guessing you have peripheral illumination activated while your ultra wide angle lens is shot nearly wide open. Essentially, what you have here is the shadow area of a severely underexposed ISO 100 image where the deepest, darkest parts have been pushed up in exposure by about 3 to 4 stops (ALO + HTP + Peripheral Correction).
No wonder banding occured. I do not doubt the 1Dsmk3, 1Dmk3, and D700 will all buckle under those conditions as well. The D3x is probably the only camera that can take so much punishment
Funny enough, I started a thread about low ISO banding in the 5Dmk2 a while ago. I caught a lot of flak for it from people saying the noise doesn't manifest itself in real life situations. Well, your sample proves other wise. HTP + ALO + Peripheral Correction + Underexposure = The Perfect Storm.
Edited on Feb 03, 2009 at 06:15 PM · View previous versions
Sadly I have just sold my 1Ds3. I say sadly because I did not expect my new 5D2 to have significantly worse performance at low ISOs, but it does seem to.
This is an ISO 200 HTP file, which is the same as an underexposed ISO 100 image. Having ALO activated only exacerbates the problem. I'm also guessing you have peripheral illumination activated while your ultra wide angle lens is shot nearly wide open. Essentially, what you have here is the shadow area of a severely underexposed ISO 100 image where the deepest, darkest parts have been pushed up in exposure by about 3 to 4 stops (ALO + HTP + Peripheral Correction).
No wonder banding occured. I do not doubt the 1Dsmk3, 1Dmk3, and D700 will all buckle under those conditions as well. The D3x is probably the only camera that can take so much punishment
Funny enough, I started a thread about low ISO banding in the 5Dmk2 a while ago. I caught a lot of flak for it from people saying the noise doesn't manifest itself in real life situations. Well, your sample proves other wise. HTP + ALO + Peripheral Correction + Underexposure = The Perfect Storm.
Edited on Feb 03, 2009 at 06:15 PM · View previous versions
Feb 03, 2009 at 06:10 PM
p.3 #25D2 low ISO poor v 1Ds3
ward1066 wrote:
how do you underexpose shadows
When you under-expose a shot you under-expose the highlights as well as the shadow areas. Isn't that obvious?
how do you underexpose shadows
When you under-expose a shot you under-expose the highlights as well as the shadow areas. Isn't that obvious?
Feb 03, 2009 at 06:11 PM
p.3 #35D2 low ISO poor v 1Ds3
Pixel Perfect wrote:
My standard ISO setting now is 400/800.When using 5D, I try avoid even ISO 400, if possible. There are obvious noise differences between ISO 200 and 400.
Don't know about 5DII, but I assume that since its pixels are smaller than 5D, ...ISO 400 on the 5D is very clean and even on something like blue skies, if the exposure is good, they show very little noise, noticeably less than my 1D II and 40D.
I seem to see obvious noise, especially when dealing with things like flowers. Of course I am looking at files at100% , not printing.
Edited on Feb 03, 2009 at 06:13 PM · View previous versions
rhtml wrote
Earlier quote
:My standard ISO setting now is 400/800.
Don't know about 5DII, but I assume that since its pixels are smaller than 5D, ...
I seem to see obvious noise, especially when dealing with things like flowers. Of course I am looking at files at100% , not printing.
Edited on Feb 03, 2009 at 06:13 PM · View previous versions
Feb 03, 2009 at 06:11 PM
p.3 #45D2 low ISO poor v 1Ds3
I have a 5D2 and actually have run into the same noise problem in ONE picture out of thousands. It, too, was taken at a low ISO (200, possibly) and it, too, had a combination of a bright center and a gray border area. In the gray border area there was quite a bit of noise with similar vertical bands. It was a horrendous picture, so I threw it out, but it's interesting to see another example of this. It might be a bug, in which case you should report it to Canon so that they can fix it in a future firmware version.
Feb 03, 2009 at 06:12 PM
p.3 #55D2 low ISO poor v 1Ds3
rhtml wrote:
I seem to see obvious noise, especially when dealing with things like flowers. Of course I am looking at files at100% , not printing.
That's strange as I do a lot of macro work and ISO 400 is my most common setting (with flash) and I never need to do NR if the exposure is good. In contrast my 40D needs NR a lot more. Also if I'm using a polarizer and handholding I'm usually at ISO 400.
I seem to see obvious noise, especially when dealing with things like flowers. Of course I am looking at files at100% , not printing.
That's strange as I do a lot of macro work and ISO 400 is my most common setting (with flash) and I never need to do NR if the exposure is good. In contrast my 40D needs NR a lot more. Also if I'm using a polarizer and handholding I'm usually at ISO 400.
Feb 03, 2009 at 06:25 PM
p.3 #65D2 low ISO poor v 1Ds3
anthonygh wrote:
Does anyone on this forum ever engage in photography rather than magnifying bits of images then engaging in numerous debates about those bits?
Anyone who can't produce superb images with either of these cameras....just isn't any good. What is the universal saying about a poor workman blaming his tools?
sure they do, the problem is they are out taking pictures and not posting ^_^
Does anyone on this forum ever engage in photography rather than magnifying bits of images then engaging in numerous debates about those bits?
Anyone who can't produce superb images with either of these cameras....just isn't any good. What is the universal saying about a poor workman blaming his tools?
sure they do, the problem is they are out taking pictures and not posting ^_^
Feb 03, 2009 at 06:31 PM
p.3 #75D2 low ISO poor v 1Ds3
Pixel Perfect wrote:
What's that supposed to mean? If any part of the shot is underexposed then the whole shot is underexposed. Brainiacs shot is at least 1EV underexposed, so the shadows will be also and shadows suffer more from underexposure than bright areas. Noise rises very quickly in shadows with underexposure.
bingo
ward1066 wrote
how do you underexpose shadowsbingo
Feb 03, 2009 at 06:33 PM
p.3 #85D2 low ISO poor v 1Ds3
Daan B wrote:
Oh, before I forget... Brainiac: I assume you shot RAW? If so, what converter did you use? And what FW does the 5D2 have?
I will have to do some further testing, but it seems ACR/LR has some troubles with 5D2 RAW files after the FW 1.0.7 update: https://www.fredmiranda.com/forum/topic/735917 (I hope to receive some feedback on this).
sometimes i wonder if ACR doesn't do deep, deep, deep shadow NR even with slider at 0 to a larger degree for the older cams like the 20D,40D, Mkiii's than the 50D and 5dmkii.
i was processing some ISO200 shot in HTP no less from my 40D and i could pull the deep shadows under bushes up and up and up and magically never see much fine grain at all (then again, my 40D, did measure somewhat anomalously to have exceptionally low levels of read noise at the owest two ISOs compared to other measured 40D, i measured mine to have less read noise per pixel than 20D,50D and even 5dmkii) and the extent to which i can do this seems hard to believe.
i should test with some other converters....
anyway i did actually measure my 40D to have lower read noise at the very lowest ISOs, per pixel of course, than the 5dmkii although the 5dmkii rapidly started to beat it, I think even by ISO400 per pixel read noise was already better, nevermind per image (and of course the 5dmkii collects a little bit more light per photosite too).
Edited on Feb 03, 2009 at 06:43 PM · View previous versions
Oh, before I forget... Brainiac: I assume you shot RAW? If so, what converter did you use? And what FW does the 5D2 have?
I will have to do some further testing, but it seems ACR/LR has some troubles with 5D2 RAW files after the FW 1.0.7 update: https://www.fredmiranda.com/forum/topic/735917 (I hope to receive some feedback on this).
sometimes i wonder if ACR doesn't do deep, deep, deep shadow NR even with slider at 0 to a larger degree for the older cams like the 20D,40D, Mkiii's than the 50D and 5dmkii.
i was processing some ISO200 shot in HTP no less from my 40D and i could pull the deep shadows under bushes up and up and up and magically never see much fine grain at all (then again, my 40D, did measure somewhat anomalously to have exceptionally low levels of read noise at the owest two ISOs compared to other measured 40D, i measured mine to have less read noise per pixel than 20D,50D and even 5dmkii) and the extent to which i can do this seems hard to believe.
i should test with some other converters....
anyway i did actually measure my 40D to have lower read noise at the very lowest ISOs, per pixel of course, than the 5dmkii although the 5dmkii rapidly started to beat it, I think even by ISO400 per pixel read noise was already better, nevermind per image (and of course the 5dmkii collects a little bit more light per photosite too).
Edited on Feb 03, 2009 at 06:43 PM · View previous versions
Feb 03, 2009 at 06:38 PM
p.3 #95D2 low ISO poor v 1Ds3
well are not some shadows darker than others?, by underexposing are you not simulating maybe the darkest of shadows. i.e. viewing the left most of the histogram?
Feb 03, 2009 at 06:40 PM
p.3 #105D2 low ISO poor v 1Ds3
Developed with DPP, no exposure adjustment necessary for the look I wanted, and therefore not underexposed. At 200 ISO a £2000 camera ought to be able to produce tidy shadows in this file, with no exposure adjustment.
Feb 03, 2009 at 10:44 PM
p.3 #115D2 low ISO poor v 1Ds3
brainiac wrote:
Developed with DPP, no exposure adjustment necessary for the look I wanted, and therefore not underexposed. At 200 ISO a £2000 camera ought to be able to produce tidy shadows in this file, with no exposure adjustment.
When you have NOT opened up shadows by a severe amount of have adjusted exposure, the 5D2 shouldn't look like this at ISO200. PERIOD! Not on DPP or in ACR/LR for that matter.
The thing that comes to my mind is that maybe your 5D2 isn't working properly. I now have had three 5D2's in less than 3 weeks. All of them had different built quality. One had a line running throught the back LCD. The other had a creaking CF card door. One had more pronounced AF point bleeding than the others. And so on...
Maybe there are also 5D2's out there that produce more noise than others... maybe it is sensor related? I don't know, just a thought... But after all that I have gone through, it wouldn't surprise me.
I would swap it for another.
Developed with DPP, no exposure adjustment necessary for the look I wanted, and therefore not underexposed. At 200 ISO a £2000 camera ought to be able to produce tidy shadows in this file, with no exposure adjustment.
When you have NOT opened up shadows by a severe amount of have adjusted exposure, the 5D2 shouldn't look like this at ISO200. PERIOD! Not on DPP or in ACR/LR for that matter.
The thing that comes to my mind is that maybe your 5D2 isn't working properly. I now have had three 5D2's in less than 3 weeks. All of them had different built quality. One had a line running throught the back LCD. The other had a creaking CF card door. One had more pronounced AF point bleeding than the others. And so on...
Maybe there are also 5D2's out there that produce more noise than others... maybe it is sensor related? I don't know, just a thought... But after all that I have gone through, it wouldn't surprise me.
I would swap it for another.
Feb 04, 2009 at 01:51 AM
Andi Dietrich
Offline
[X]
Details
p.3 #125D2 low ISO poor v 1Ds3
It is quite difficult to expose an indoor/outdoor photo without snow, with snow it's just bound to turn out ugly without proper exposure.
Low contrast in shadows, low contrast in lights, high contrast all over! There is no evidence that any other camera would give a better image.
If you look at the image below you see the image is quite underexposed. As you know probably, the left 50% of the levels contain only 25% of the information of an image. The better part of the levels are wasted
http://www.andidietrich.com/elements/fm3/brain.jpg
good morning BTW
Edited on Feb 05, 2009 at 03:07 PM · View previous versions
Low contrast in shadows, low contrast in lights, high contrast all over! There is no evidence that any other camera would give a better image.
If you look at the image below you see the image is quite underexposed. As you know probably, the left 50% of the levels contain only 25% of the information of an image. The better part of the levels are wasted
http://www.andidietrich.com/elements/fm3/brain.jpg
good morning BTW
Edited on Feb 05, 2009 at 03:07 PM · View previous versions
Feb 04, 2009 at 02:49 AM
p.3 #135D2 low ISO poor v 1Ds3
Rubber Soul wrote:
This is an ISO 200 HTP file, which is the same as an underexposed ISO 100 image. Having ALO activated only exacerbates the problem. I'm also guessing you have peripheral illumination activated while your ultra wide angle lens is shot nearly wide open. Essentially, what you have here is the shadow area of a severely underexposed ISO 100 image where the deepest, darkest parts have been pushed up in exposure by about 3 to 4 stops (ALO + HTP + Peripheral Correction).
No wonder banding occured. I do not doubt the 1Dsmk3, 1Dmk3, and D700 will all buckle under those...Show more →
This is exactly correct. Turn off HTP+ALO+Peripheral Correction and do not underexpose and you should not see these results.
I shot some tonight at a hockey game and just finished downloading them. Shot some of the stands and such with the 5D2 at ISO 1600 and my results at 100% are less noisy (Chroma and banding) than the ISO 200 shot posted here. The luminance noise is a tad higher but not much and that is at ISO 1600. I also shot the 1D3 at ISO2500 all night long. Couldn't do any low ISO 100-400 tests unfortunately. Once the client has the images, I'll see about posting some. Or perhaps, I'll just do a controlled test here at home tomorrow.
Have a good night!
Andrew
brainiac wrote
Sadly I have just sold my 1Ds3. I say sadly because I did not expect my new 5D2 to have significantly worse performance at low ISOs, but it does seem to.No wonder banding occured. I do not doubt the 1Dsmk3, 1Dmk3, and D700 will all buckle under those...Show more →
This is exactly correct. Turn off HTP+ALO+Peripheral Correction and do not underexpose and you should not see these results.
I shot some tonight at a hockey game and just finished downloading them. Shot some of the stands and such with the 5D2 at ISO 1600 and my results at 100% are less noisy (Chroma and banding) than the ISO 200 shot posted here. The luminance noise is a tad higher but not much and that is at ISO 1600. I also shot the 1D3 at ISO2500 all night long. Couldn't do any low ISO 100-400 tests unfortunately. Once the client has the images, I'll see about posting some. Or perhaps, I'll just do a controlled test here at home tomorrow.
Have a good night!
Andrew
Feb 04, 2009 at 03:04 AM
p.3 #145D2 low ISO poor v 1Ds3
apdieb wrote:
This is exactly correct. Turn off HTP+ALO+Peripheral Correction and do not underexpose and you should not see these results.
I don't get it... none of those features (HTP+ALO+Peripheral Correction) affects RAW data, right
On another note... Now that I am using FW 1.0.6 (instead of FW 1.0.7) the images are much more comparable to the 1Ds3 RAW files. FW 1.0.7 and LR 2.2 are no perfect match...
This is exactly correct. Turn off HTP+ALO+Peripheral Correction and do not underexpose and you should not see these results.
I don't get it... none of those features (HTP+ALO+Peripheral Correction) affects RAW data, right
On another note... Now that I am using FW 1.0.6 (instead of FW 1.0.7) the images are much more comparable to the 1Ds3 RAW files. FW 1.0.7 and LR 2.2 are no perfect match...
Feb 04, 2009 at 03:31 AM
p.3 #155D2 low ISO poor v 1Ds3
I think all of the fancy features were turned off in DPP except peripheral illumination. I will see if disabling that helps - it ought to as it has the effect of pushing the corners.
All this nonsense about underexposure misses the point entirely. I exposed the shot how I wanted it to look. I don't want white in the sky. The kindly provided histogram shows that the shadows are not clipped. The exposure is perfect. Could I reduce noise by overexposing and then pulling in DPP? Sure, but I did not buy the camera to do that to every shot. Most shots have higher contrast than this scene, and it would not be possible to expose further right without sacrificing highlights. I should not have to beware of putting data in the left quarter of the histogram at 200 ISO. Regardless of this red herring, I have been using a 1Ds3 for a year, and I don't think handling of shadows is the same.
Tonal problems also emerge in large graduated areas. 5D2 ouput has the appearance you would expect if the camera had a bandwidth bottleneck somewhere in its processing, which leads to posterisation in the shadow noise, and in graduated tones. People are calling that banding, but it seems to me that it is posterisation which is exacerbating shadow noise and spoiling continuous grades. These are not the 14 bit files that I grew used to in the 1Ds3. A simple solution will be to add very fine noise, so that the flat tonal blocks will to some extent become less visible, but it is a shame that we have to use noise to mask a camera shortcoming.
Good morning Andi!
All this nonsense about underexposure misses the point entirely. I exposed the shot how I wanted it to look. I don't want white in the sky. The kindly provided histogram shows that the shadows are not clipped. The exposure is perfect. Could I reduce noise by overexposing and then pulling in DPP? Sure, but I did not buy the camera to do that to every shot. Most shots have higher contrast than this scene, and it would not be possible to expose further right without sacrificing highlights. I should not have to beware of putting data in the left quarter of the histogram at 200 ISO. Regardless of this red herring, I have been using a 1Ds3 for a year, and I don't think handling of shadows is the same.
Tonal problems also emerge in large graduated areas. 5D2 ouput has the appearance you would expect if the camera had a bandwidth bottleneck somewhere in its processing, which leads to posterisation in the shadow noise, and in graduated tones. People are calling that banding, but it seems to me that it is posterisation which is exacerbating shadow noise and spoiling continuous grades. These are not the 14 bit files that I grew used to in the 1Ds3. A simple solution will be to add very fine noise, so that the flat tonal blocks will to some extent become less visible, but it is a shame that we have to use noise to mask a camera shortcoming.
Good morning Andi!
Feb 04, 2009 at 05:58 AM
p.3 #165D2 low ISO poor v 1Ds3
You do realize that you're looking at a very high resolution image, so naturally you're going to see some difference when juxtaposed to a 10 megapixel image.
No offense to anyone in this thread, but I swear that some people here never use their cameras, but rather nitpick every damn pixel on the sensor.
No offense to anyone in this thread, but I swear that some people here never use their cameras, but rather nitpick every damn pixel on the sensor.
Feb 04, 2009 at 06:04 AM
p.3 #175D2 low ISO poor v 1Ds3
Well I said this of my experiences with the 5DMKII's in real world use .... and wasn't believed
I'm glad there are others (and there are) coming out and confirming my view of the camera ....
Low ISO isn't as good as the 5D MKI or the 1DMKIII or the 1DSMKIII
Reason I sold mine and have gone to 1DMKIII's - the quality of the files are excellent.
No doubt I'll be termed a poor practiser of photography - but my clients seem to like what I do.
I'm glad there are others (and there are) coming out and confirming my view of the camera ....
Low ISO isn't as good as the 5D MKI or the 1DMKIII or the 1DSMKIII
Reason I sold mine and have gone to 1DMKIII's - the quality of the files are excellent.
No doubt I'll be termed a poor practiser of photography - but my clients seem to like what I do.
Feb 04, 2009 at 06:11 AM
p.3 #185D2 low ISO poor v 1Ds3
CanAm wrote:
You do realize that you're looking at a very high resolution image, so naturally you're going to see some difference when juxtaposed to a 10 megapixel image.
No offense to anyone in this thread, but I swear that some people here never use their cameras, but rather nitpick every damn pixel on the sensor.
On Saturday I shot over 4000 pictures with it and was paid handsomely for my efforts. If you don't like talking to others about their experience of camera performance, what are you doing here?
You do realize that you're looking at a very high resolution image, so naturally you're going to see some difference when juxtaposed to a 10 megapixel image.
No offense to anyone in this thread, but I swear that some people here never use their cameras, but rather nitpick every damn pixel on the sensor.
On Saturday I shot over 4000 pictures with it and was paid handsomely for my efforts. If you don't like talking to others about their experience of camera performance, what are you doing here?
Feb 04, 2009 at 06:14 AM
p.3 #195D2 low ISO poor v 1Ds3
spxxxx wrote:
Well I said this of my experiences with the 5DMKII's in real world use .... and wasn't believed
I'm glad there are others (and there are) coming out and confirming my view of the camera ....
Low ISO isn't as good as the 5D MKI or the 1DMKIII or the 1DSMKIII
Reason I sold mine and have gone to 1DMKIII's - the quality of the files are excellent.
No doubt I'll be termed a poor practiser of photography - but my clients seem to like what I do.
I am not sure it's _worse_ than the 5D in regards of shadow pushing bandwidth, but it's not enormously better. I certainly agree with you that the both 1D cameras seem to be better in this regard.
Maybe eosfun can enlighten us on how many bits the camera thinks with at each stage of image processing. I don't believe I am seeing a full 14 bit path.
Well I said this of my experiences with the 5DMKII's in real world use .... and wasn't believed
I'm glad there are others (and there are) coming out and confirming my view of the camera ....
Low ISO isn't as good as the 5D MKI or the 1DMKIII or the 1DSMKIII
Reason I sold mine and have gone to 1DMKIII's - the quality of the files are excellent.
No doubt I'll be termed a poor practiser of photography - but my clients seem to like what I do.
I am not sure it's _worse_ than the 5D in regards of shadow pushing bandwidth, but it's not enormously better. I certainly agree with you that the both 1D cameras seem to be better in this regard.
Maybe eosfun can enlighten us on how many bits the camera thinks with at each stage of image processing. I don't believe I am seeing a full 14 bit path.
Feb 04, 2009 at 06:17 AM
p.3 #205D2 low ISO poor v 1Ds3
cineski wrote:
Personal experience with the 1Ds3. However, my fanboy comments go toward people who discount people's findings as user problems. I've seen quite a few responses about Canon's poor low ISO performance, and tested it out myself and got the same results. No user error involved, and very real.
Exactly, I remember when I posted about the original 5d's noise at ISO 100 while shooting in the studio, I received a few pages of really ignorant comments. The fact of the matter is that if you have dark areas in a photograph then you're going to have noise. You can overexpose a photo and still have areas that are dark and going to have noise. And if you think you should have to "expose to the right" and adjust in post process then you've been trained to work around their engineering faults. Otherwise they would have the camera do this for you.
Edited on Feb 04, 2009 at 06:24 AM · View previous versions
Personal experience with the 1Ds3. However, my fanboy comments go toward people who discount people's findings as user problems. I've seen quite a few responses about Canon's poor low ISO performance, and tested it out myself and got the same results. No user error involved, and very real.
Exactly, I remember when I posted about the original 5d's noise at ISO 100 while shooting in the studio, I received a few pages of really ignorant comments. The fact of the matter is that if you have dark areas in a photograph then you're going to have noise. You can overexpose a photo and still have areas that are dark and going to have noise. And if you think you should have to "expose to the right" and adjust in post process then you've been trained to work around their engineering faults. Otherwise they would have the camera do this for you.
Edited on Feb 04, 2009 at 06:24 AM · View previous versions
Feb 04, 2009 at 06:20 AM