Blown away by 5D mki

KungPOW · 10-08-2009, 12:47 PM

Originally posted by georgweb

Hi guys,

I'd like to know from your experience if you could rate the following cameras only regarding low light performance. Let's be more specific, I'd appreciate 'real world' experience with those cams, dark indoor shots, stage shots, artificial light.

The easiest is to talk ISO I guess (in the sense of keeping the DR and the pic not totally falling apart). Obviously my yardstick is the first camera.

Pentax K100D (my current DSLR) - acceptable ISO: 800

Canon 5D 'classic': 1600 ? (the most tempting upgrade for me)

Nikon D700: 6400 ?

Sony A850: 800 ?

Pentax K-X /Sony A500 (presumably same sensor): 1600 ?

As for many other people the low light perf. is crucial to me and I just can't tell if it would be worth upgrading for two stops. I am using manual f/1.2 - f/2 lenses in those situations whenever possible so I can't do much better lenswise with my current camera.

Any thoughts appreciated,Georg (the other)

Georg,

From what I've used;

The K10D can get good high ISO results, but you need to expose to the right. Anytime you need to push the exposure, the noise gets bad. This I find is an issue with the K10D because the camera tends to underexpose 2/3 to 1 stop. So you really need to be aware of your histogram.

The 5D shots I saw from my co-workers camera at ISO 1000 looked better then my K10D shots at ISO 400. Granted, I had a number of shots underexposed, because I forgot to watch the histogram.... I really need to learn not to trust the matrix metering on the K10D.

The D200 has a much better matrix meter then the K10D, producing far more correctly exposed shots over the K10D. But, When the K10D is correctly exposed, it has far better high ISO images then the D200. The D200 has a faint "ripple" texture to its high ISO noise. It can be seen from about ISO 600 and up.

When I had a K100D, I took very few high ISO shots. But the few that I did take had a very film like grain pattern to them. Usually I would convert the shot to B/W.

---------------------

I was looking at the high ISO images from the dcreviews "compaire-o-meter" from the A850 and the K-7. I found that downsizing the A850 images to the same res as the K-7 resulted in a better image then the K-7. It has made me reconsider the A850 as a camera. Has anyone else noticed that downsizing can improve image noise? This is an idea that is new to me.

Gooshin · 10-08-2009, 12:54 PM

Originally posted by KungPOW

Has anyone else noticed that downsizing can improve image noise? This is an idea that is new to me.

ofcourse

3 black dots 1 blue dot = average 1 black dot

noise gone, tada!

pingflood · 10-08-2009, 06:30 PM

Originally posted by Gooshin

ofcourse

3 black dots 1 blue dot = average 1 black dot

noise gone, tada!

Like most things, you just have to take it to its extremes and realize that if you averaged out all the pixels you'd end up with a big, noise free, grey print.

So yeah, downsampling = less noise.

KungPOW · 10-08-2009, 07:30 PM

So what is the downside to downsizing?

If it works, and to me it looked good, why would you want a lower mp camera for better low light shots? Why not have a 20+ mp camera, and if the shot is noisy, downsize to 10 mp?

I just priced out a Canon kit. I need to sell or rent a kidney.

Gooshin · 10-08-2009, 07:52 PM

Originally posted by KungPOW

So what is the downside to downsizing?

If it works, and to me it looked good, why would you want a lower mp camera for better low light shots? Why not have a 20+ mp camera, and if the shot is noisy, downsize to 10 mp?

I just priced out a Canon kit. I need to sell or rent a kidney.

nothing at all, assuming the program doing it is quality software.

the only downside would be printing.

this is why its so easy to be fooled on the internet, 800pix images get rid of so much noise.

when i work with my photos, i have a second monitor displaying shit at 1200X1000 so i'm much more critical

all of a sudden when i export to jpeg and downsize half th enoise goes away.

KungPOW · 10-08-2009, 08:06 PM

funny how something so obvious compleatly got past me...

Gooshin · 10-08-2009, 08:26 PM

Originally posted by KungPOW

funny how something so obvious compleatly got past me...

regarding NOISE in general

over the course of the next.. whatever your shooting habits are

assemble a set of images you consider noisy and clean.

then take them and print them at 8X12 or whatever

that will be the true test, once you SEE how the "noise" looks on the final product, your tolerances for what is bad and good might change... significantly.

Class A · 10-08-2009, 11:59 PM

Originally posted by KungPOW

...why would you want a lower mp camera for better low light shots?

You don't.

A high MP camera not only allows downsizing of images to get the same noise levels but it also shows the finer noise structure before downsizing.

When looking at DxOMark results, make sure to select the "Print" tab. Here the results are normalised to the same image size. The "Screen" tab shows results "per Pixel" and thus favours cameras with lower resolution. However, results per "pixel" are rather irrelevant in practice. They only indicate what you would see if you chose to reproduce pixels with the same size, resulting in correspondingly larger images for the high MP cameras. That's normally not a fair way of comparing IQ.

pingflood · 10-09-2009, 05:01 AM

Originally posted by Class A

You don't.

A high MP camera not only allows downsizing of images to get the same noise levels but it also shows the finer noise structure before downsizing.

When looking at DxOMark results, make sure to select the "Print" tab. Here the results are normalised to the same image size. The "Screen" tab shows results "per Pixel" and thus favours cameras with lower resolution. However, results per "pixel" are rather irrelevant in practice. They only indicate what you would see if you chose to reproduce pixels with the same size, resulting in correspondingly larger images for the high MP cameras. That's normally not a fair way of comparing IQ.

HALLELUJAH!

I can't tell you how many times I see people "prove" that camera X is noisier than camera Y because they look on a per pixel basis rather than on an output size basis. The infamous "50D is noisier than 40D" debate is a good example; when you use dxomark and compare "screen" numbers, it looks worse. When using "print" numbers, better.

Alfisti · 10-09-2009, 08:08 AM

Originally posted by Gooshin

regarding NOISE in general

over the course of the next.. whatever your shooting habits are

assemble a set of images you consider noisy and clean.

then take them and print them at 8X12 or whatever

that will be the true test, once you SEE how the "noise" looks on the final product, your tolerances for what is bad and good might change... significantly.

and here is an example, this is a 100% crop of an image pushed like HELL then upsampled like hell to an 11*14 print.

You'd be surprised how good the print looks, first class? No, does ANYONE notice? Nope.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Abstract Blown Out	justDIY	Post Your Photos!	4	01-20-2010 08:41 PM
Nature Blown shot	photolady95	Photo Critique	7	01-10-2010 10:00 AM
Blown out Yellow	dadipentak	Digital Processing, Software, and Printing	8	11-23-2009 02:12 PM
Blown background on purpose	hinman	Post Your Photos!	2	09-01-2007 10:12 PM
Full-blown LBA!	rfortson	Pentax SLR Lens Discussion	14	02-21-2007 05:13 PM

10-08-2009, 12:47 PM	#31
KungPOW Veteran Member Join Date: Nov 2007 Location: Vancouver, Canada Photos: Gallery Posts: 1,699 Original Poster	Originally posted by georgweb Hi guys, I'd like to know from your experience if you could rate the following cameras only regarding low light performance. Let's be more specific, I'd appreciate 'real world' experience with those cams, dark indoor shots, stage shots, artificial light. The easiest is to talk ISO I guess (in the sense of keeping the DR and the pic not totally falling apart). Obviously my yardstick is the first camera. Pentax K100D (my current DSLR) - acceptable ISO: 800 Canon 5D 'classic': 1600 ? (the most tempting upgrade for me) Nikon D700: 6400 ? Sony A850: 800 ? Pentax K-X /Sony A500 (presumably same sensor): 1600 ? As for many other people the low light perf. is crucial to me and I just can't tell if it would be worth upgrading for two stops. I am using manual f/1.2 - f/2 lenses in those situations whenever possible so I can't do much better lenswise with my current camera. Any thoughts appreciated,Georg (the other) Georg, From what I've used; The K10D can get good high ISO results, but you need to expose to the right. Anytime you need to push the exposure, the noise gets bad. This I find is an issue with the K10D because the camera tends to underexpose 2/3 to 1 stop. So you really need to be aware of your histogram. The 5D shots I saw from my co-workers camera at ISO 1000 looked better then my K10D shots at ISO 400. Granted, I had a number of shots underexposed, because I forgot to watch the histogram.... I really need to learn not to trust the matrix metering on the K10D. The D200 has a much better matrix meter then the K10D, producing far more correctly exposed shots over the K10D. But, When the K10D is correctly exposed, it has far better high ISO images then the D200. The D200 has a faint "ripple" texture to its high ISO noise. It can be seen from about ISO 600 and up. When I had a K100D, I took very few high ISO shots. But the few that I did take had a very film like grain pattern to them. Usually I would convert the shot to B/W. --------------------- I was looking at the high ISO images from the dcreviews "compaire-o-meter" from the A850 and the K-7. I found that downsizing the A850 images to the same res as the K-7 resulted in a better image then the K-7. It has made me reconsider the A850 as a camera. Has anyone else noticed that downsizing can improve image noise? This is an idea that is new to me.

10-08-2009, 12:54 PM	#32
Gooshin Veteran Member Join Date: Sep 2007 Location: Toronto, the one in Canada. Posts: 5,610	Originally posted by KungPOW Has anyone else noticed that downsizing can improve image noise? This is an idea that is new to me. ofcourse 3 black dots 1 blue dot = average 1 black dot noise gone, tada!

10-08-2009, 06:30 PM	#33
pingflood Veteran Member Join Date: Jun 2008 Location: Borås, Sweden Photos: Gallery Posts: 3,169	Originally posted by Gooshin ofcourse 3 black dots 1 blue dot = average 1 black dot noise gone, tada! Like most things, you just have to take it to its extremes and realize that if you averaged out all the pixels you'd end up with a big, noise free, grey print. So yeah, downsampling = less noise.

10-08-2009, 07:30 PM	#34
KungPOW Veteran Member Join Date: Nov 2007 Location: Vancouver, Canada Photos: Gallery Posts: 1,699 Original Poster	So what is the downside to downsizing? If it works, and to me it looked good, why would you want a lower mp camera for better low light shots? Why not have a 20+ mp camera, and if the shot is noisy, downsize to 10 mp? I just priced out a Canon kit. I need to sell or rent a kidney.

10-08-2009, 07:52 PM	#35
Gooshin Veteran Member Join Date: Sep 2007 Location: Toronto, the one in Canada. Posts: 5,610	Originally posted by KungPOW So what is the downside to downsizing? If it works, and to me it looked good, why would you want a lower mp camera for better low light shots? Why not have a 20+ mp camera, and if the shot is noisy, downsize to 10 mp? I just priced out a Canon kit. I need to sell or rent a kidney. nothing at all, assuming the program doing it is quality software. the only downside would be printing. this is why its so easy to be fooled on the internet, 800pix images get rid of so much noise. when i work with my photos, i have a second monitor displaying shit at 1200X1000 so i'm much more critical all of a sudden when i export to jpeg and downsize half th enoise goes away.

10-08-2009, 08:06 PM	#36
KungPOW Veteran Member Join Date: Nov 2007 Location: Vancouver, Canada Photos: Gallery Posts: 1,699 Original Poster	funny how something so obvious compleatly got past me...

10-08-2009, 08:26 PM	#37
Gooshin Veteran Member Join Date: Sep 2007 Location: Toronto, the one in Canada. Posts: 5,610	Originally posted by KungPOW funny how something so obvious compleatly got past me... regarding NOISE in general over the course of the next.. whatever your shooting habits are assemble a set of images you consider noisy and clean. then take them and print them at 8X12 or whatever that will be the true test, once you SEE how the "noise" looks on the final product, your tolerances for what is bad and good might change... significantly.

10-08-2009, 11:59 PM	#38
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by KungPOW ...why would you want a lower mp camera for better low light shots? You don't. A high MP camera not only allows downsizing of images to get the same noise levels but it also shows the finer noise structure before downsizing. When looking at DxOMark results, make sure to select the "Print" tab. Here the results are normalised to the same image size. The "Screen" tab shows results "per Pixel" and thus favours cameras with lower resolution. However, results per "pixel" are rather irrelevant in practice. They only indicate what you would see if you chose to reproduce pixels with the same size, resulting in correspondingly larger images for the high MP cameras. That's normally not a fair way of comparing IQ.

10-09-2009, 05:01 AM	#39
pingflood Veteran Member Join Date: Jun 2008 Location: Borås, Sweden Photos: Gallery Posts: 3,169	Originally posted by Class A You don't. A high MP camera not only allows downsizing of images to get the same noise levels but it also shows the finer noise structure before downsizing. When looking at DxOMark results, make sure to select the "Print" tab. Here the results are normalised to the same image size. The "Screen" tab shows results "per Pixel" and thus favours cameras with lower resolution. However, results per "pixel" are rather irrelevant in practice. They only indicate what you would see if you chose to reproduce pixels with the same size, resulting in correspondingly larger images for the high MP cameras. That's normally not a fair way of comparing IQ. HALLELUJAH! I can't tell you how many times I see people "prove" that camera X is noisier than camera Y because they look on a per pixel basis rather than on an output size basis. The infamous "50D is noisier than 40D" debate is a good example; when you use dxomark and compare "screen" numbers, it looks worse. When using "print" numbers, better.

10-09-2009, 08:08 AM	#40
Alfisti Veteran Member Join Date: Mar 2007 Location: Toronto (for now) Posts: 1,748	Originally posted by Gooshin regarding NOISE in general over the course of the next.. whatever your shooting habits are assemble a set of images you consider noisy and clean. then take them and print them at 8X12 or whatever that will be the true test, once you SEE how the "noise" looks on the final product, your tolerances for what is bad and good might change... significantly. and here is an example, this is a 100% crop of an image pushed like HELL then upsampled like hell to an 11*14 print. You'd be surprised how good the print looks, first class? No, does ANYONE notice? Nope.