Canon's high-ISO secret? - FM Forums

Register · Software · Search · Image Upload · Buy & Sell · Reviews · Hosting

Moderated by: guardian

Visit the FM Store · Image Upload · Buy & Sell
FM Forum Rules
Canon SLRs, primes, and zooms lenses reviews

FM Forums | Canon-mount SLRs | Join Image Upload

1

2 end

alfarmer
Offline
Buy and Sell: On

p.1 #1 · Canon's high-ISO secret?

Most of what I've read suggests that Canon's high-ISO prowess is due to CMOS sensor technology, but I was curious if any of it was due to software algorithms used to process that data?

In an extreme difference example, Canon's NR technology doesn't result in super-smudgy images whereas other manufacturers suffer from their aggressive or different application of NR. That or with NR turned down there's tons of sensor noise.

Just curious...

ALF

Jul 22, 2008 at 09:59 PM

gfiksel
Offline
Image Upload: On

p.1 #2 · Canon's high-ISO secret?

If we knew the secret it wouldn't be a secret anymore, would it?

Edited on Jul 22, 2008 at 10:03 PM

Jul 22, 2008 at 10:02 PM

trumpet_guy
Offline
Image Upload: On

p.1 #3 · Canon's high-ISO secret?

I think it's predominantly the sensor and the associated read electronics.

If it weren't, then most any DSLR of a given sensor size would be able to achieve
comparable noise performance when shooting RAW.

Jul 22, 2008 at 10:11 PM

LMCasey
Offline
Image Upload: Off

p.1 #4 · Canon's high-ISO secret?

I don't think Canon has much of a lead in sensor noise any more. They used to, but with the advent of the Nikon D3 and D700, Canon is a bit behind Nikon in the noise department. In APS-C cameras, it looks like Canon may still have a bit of the lead judging by raw captures. Nikon still uses too much NR to achieve their clean looking jpegs.

Jul 22, 2008 at 10:20 PM

Stunnaz
Offline
Image Upload: Off

p.1 #5 · Canon's high-ISO secret?

gfiksel wrote:
If we knew the secret it wouldn't be a secret anymore, would it?

That made me chuckle.

Jul 22, 2008 at 11:31 PM

Pixel Perfect
Offline
Image Upload: On

p.1 #6 · Canon's high-ISO secret?

Sony's new backlit sensor looks like it could be a very big step forward. And then there's Kodak's new tech for low light with a alternate Bayer pattern. I wonder how that's progressing?

Interesting in scientific applications CCD is still king. I know of a camera that can shoot 100 million fps for watching the evolution of microbubbles and the sensor is a CCD with a native ISO of 100000 (at room temperature) and the noise is very good.

Jul 22, 2008 at 11:32 PM

Ariel Bravy
Offline
Buy and Sell: On

p.1 #7 · Canon's high-ISO secret?

Supposedly CCDs have better noise control than CMOS so I'm not sure exactly what changes and optimizations Canon did with their sensors to lower the noise so much. It took Nikon a while to catch up with their CCD sensors to Canon's CMOS.

Jul 22, 2008 at 11:37 PM

DavidP
Offline
Image Upload: On

p.1 #8 · Canon's high-ISO secret?

If I told you, I'd have to kill you.

Jul 22, 2008 at 11:47 PM

Alan321
Offline
Image Upload: On

p.1 #9 · Canon's high-ISO secret?

The modern CMOS technology uses circuitry to reduce noise by draining electric charge from each pixel before capturing an image but the modern cameras also use in-camera post-capture processing. Much of the improvement in recent cameras is due to improved in-camera processing rather than improved CMOS technology. That is why the results are not overwhelmingly better than for the previous generation cameras once you apply an external noise reduction program to the images, and why there was no great leap in usefl (clean) dynamic range. The hardware that does the best seems to be the hardware with the latest CMOS technology PLUS large pixels. There's no smudging caused by that combination but there can be some caused by the in-camera data processing, especially at extreme ISOs.

Nikon's CCD sensors never did catch up with Canon's CMOS sensors. They only jumped ahead when they introduced the large-pixel CMOS sensor in their D3. Having an AF system that worked was an added bonus.

- Alan

Jul 24, 2008 at 09:07 AM

Beni
Offline
Buy and Sell: On

p.1 #10 · Canon's high-ISO secret?

Ariel Bravy wrote:
Supposedly CCDs have better noise control than CMOS so I'm not sure exactly what changes and optimizations Canon did with their sensors to lower the noise so much. It took Nikon a while to catch up with their CCD sensors to Canon's CMOS.

They didn't, there are no CCD chips even close to the CMOS chips now in the new Nikon cameras.

Jul 24, 2008 at 10:18 AM

milanissimo
Offline
Image Upload: Off

p.1 #11 · Canon's high-ISO secret?

alfarmer wrote:
Most of what I've read suggests that Canon's high-ISO prowess is due to CMOS sensor technology, but I was curious if any of it was due to software algorithms used to process that data?

In an extreme difference example, Canon's NR technology doesn't result in super-smudgy images whereas other manufacturers suffer from their aggressive or different application of NR. That or with NR turned down there's tons of sensor noise.

Just curious...

ALF

My vastly amateur, uneducated and non-technical guess is you can do hardware NR on CMOS, while you can't do that on CCD (hence the software NR and the impressionist pictures we've seen from Nikons). Anyway this is what I've read in a magazine some years ago, I don't trust the magazine in the first place so it might be wrong

. It also helps that Canon makes their own sensors, so apparently they have a pretty good idea how every single part of it works, while most other manufacturers were shopping for sensors at Sony (how many cameras used their 6mpix sensor? A lot).

Jul 24, 2008 at 02:49 PM

alfarmer
Offline
Buy and Sell: On

p.1 #12 · Canon's high-ISO secret?

Interesting info. Thanks, everyone.

What prompted the question was my recent purchase of an Olympus E420 and what is arguably their best lens (the 50mm f/2 macro). In just over a month with the camera, I've found two things on my 5D that have forever spoiled me -- high ISO performance and AF intelligence/accuracy.

For whatever reason, ISO 800 on the E420 is noisy and ISO 1600 can border on completely unacceptable in darker situations. Thank goodness for Noise Ninja! I was just wondering if Canon has some software algorithm patented, which allows them to have cleaner images that aren't "smudged" by NR.

The more-difficult thing to get used to coming from the 5D to the E420 is AF. I don't understand "EV" numbers but apparently the E420 can only focus down to EV-0 while the Canons can focus at EV-2. Oddly enough, that's a HUGE thing. A much bigger deal than I'd have suspected, so luckily my E420 is for vacation where there's typically plenty of light.

While on the subject of AF -- does Canon do it differently than Olympus or is their sensor node just bigger? I just notice the sensore "square" in the viewfinder is MUCH bigger on my 5D when it blinks/focuses than the one on the E420.

Cheers,
ALF

Edited on Jul 24, 2008 at 02:59 PM

Jul 24, 2008 at 02:58 PM

brainiac
Offline
Image Upload: On

p.1 #13 · Canon's high-ISO secret?

LMCasey wrote:
I don't think Canon has much of a lead in sensor noise any more. They used to, but with the advent of the Nikon D3 and D700, Canon is a bit behind Nikon in the noise department.

Name these cameras, at iso 3200:

Edited by brainiac on Jul 27, 2008 at 04:16 PM GMT

Edited on Jul 27, 2008 at 04:16 PM

Jul 24, 2008 at 03:11 PM

Alan321
Offline
Image Upload: On

p.1 #14 · Canon's high-ISO secret?

ALF, Canon info about the size of the AF sensors is a bit difficult to find. Info about how many main-sensor image pixels corespond to each AF-sensor pixel is impossible to find because Canon won't release it.

Small cameras use small sensors with small pixels which have a relatively poor signal to noise ratio regardless of the level of sensor and processing technology being used. e.g. as evidence of this the Canon G9 is quite noisy at ISO 400 even though it is a recent model and supposedly has the same Digic III computer as in the up-market DSLRs.

Brainiac, I won't even try to guess which image came out of which camera, but an even harder test would be to correctly apply Neat Image Pro to each one and then try to tell them apart.

- Alan

Jul 24, 2008 at 04:01 PM

DoubleNegative
Online
Image Upload: Off

p.1 #15 · Canon's high-ISO secret?

CMOS sensor, Digic II/III, algorithms, etc.

Jul 24, 2008 at 04:07 PM

Daniel Heineck
Offline
Buy and Sell: On

p.1 #16 · Canon's high-ISO secret?

5D, D3, D700?

The last two look nearly identical, so that's my bet.

Jul 24, 2008 at 04:08 PM

GeneO
Offline
Image Upload: Off

p.1 #17 · Canon's high-ISO secret?

Ariel Bravy wrote:
Supposedly CCDs have better noise control than CMOS so I'm not sure exactly what changes and optimizations Canon did with their sensors to lower the noise so much. It took Nikon a while to catch up with their CCD sensors to Canon's CMOS.

CMOS has readout electronics at each photosite while CCDs do not. Placing electronics at each photosite introduces pattern noise and reduces real estate for the photosite. Canon has technology that eliminates the pattern noise. Having the readout electronics closer to the site reduces noise while on the other hand so does a larger photosite.

Gene

Edited on Jul 24, 2008 at 04:43 PM

Jul 24, 2008 at 04:43 PM

gabimaster
Offline
Image Upload: Off

p.1 #18 · Canon's high-ISO secret?

HEI,you,people,did you saw what"JPEG"s are making NIKON Dslr-s

DAMN,are so ,so soft;mostly I'm taking weddings pictures and I only shoot JPEGs(I don't need to shoot RAW because after so many years I think I'm very good on controling exposure)and for me THE CANON DSLRs are the best because are delivering the best JPEGs in the world,I SAW Jpegs made by NIKON D300, D3 and I think the D700 IT will be the same:AWFUL Jpegs, extremely soft,and very unnatural(because of very strong NR aplied in the camera)I THINK ,CANON is still the best on quality image department,IT'S MORE IMPORTANT TO DELIVER A SENSOR THAT PRODUSE A VERY LITTLE NOISE THAT TO ELIMINATE NOISE IN CAMERA(like NIKON DOES)

Jul 24, 2008 at 10:18 PM

M Burlingame
Offline
Image Upload: On

p.1 #19 · Canon's high-ISO secret?

"Most of what I've read suggests that Canon's high-ISO prowess is due to CMOS sensor technology, but I was curious if any of it was due to software algorithms used to process that data?"

It's a Christmas Miracle....

Mary

Jul 24, 2008 at 10:27 PM

brainiac
Offline
Image Upload: On

p.1 #20 · Canon's high-ISO secret?

> 5D, D3, D700? The last two look nearly identical, so that's my bet.

The cameras are... 1Ds3, 1Ds3, D3. Sic.

My point is that Canon's best kept high iso secret is that the 1Ds3 is better than the D3 at high iso's.

Now I expect a deluge of people saying 'rubbish! look at this online comparison, and that, and that'. Please don't bother unless the comparisons show the D3 ouput uprezzed to 21 megapixel. Comparing 100% crops at different magnifications, as done on almost all test sites, is thoroughly misleading. It is like showing a square inch cut out from an A4 print, and another square inch cut out of an A3 print. The A4 cut-out looks less noisy, even if the camera isn't as good.

When comparing 100% crops from a 12 Mpixel and 21 Mpixel camera, the 21 Mpixel file is equivalent to a much bigger print. A fair comparison on a perfect monitor would be to examine the noise in the 12 megapixel file at 175%. It's bigger and worse than it seemed at 100%.

The crops above are from no less trusted a source than Phil Askey's DPReview. He plainly states that the D3 is better at iso's above 1600, because he failed to uprez the D3 file to 21 Mpixels, and so misjudged his own crops.

Edited on Jul 27, 2008 at 04:28 PM

Jul 27, 2008 at 04:17 PM

jvarszegi
Offline
Image Upload: Off

p.1 #21 · Canon's high-ISO secret?

Perhaps we have discovered a new principle: an important factor in high ISO performance is marketing.

Jul 27, 2008 at 04:44 PM

Makten
Offline
Image Upload: Off

p.1 #22 · Canon's high-ISO secret?

brainiac wrote:
He plainly states that the D3 is better at iso's above 1600, because he failed to uprez the D3 file to 21 Mpixels, and so misjudged his own crops.

Did he? It looks very much like the D3 pic is "uprezed" since it is much softer. Why would it be softer if both were 100% crops?
Where can one read about the comparsion?

Edited on Jul 27, 2008 at 05:11 PM

Jul 27, 2008 at 05:10 PM

brainiac
Offline
Image Upload: On

p.1 #23 · Canon's high-ISO secret?

Makten wrote:
brainiac wrote:
He plainly states that the D3 is better at iso's above 1600, because he failed to uprez the D3 file to 21 Mpixels, and so misjudged his own crops.

Did he? It looks very much like the D3 pic is "uprezed" since it is much softer. Why would it be softer if both were 100% crops?
Where can one read about the comparsion?

I did the uprezzing in the crop above to illustrate that the original comparison on DPReview was a scientifically reliable way of handicapping the higher Mpixel camera. The original, with crops shown at different magnifications in order to handicap the 1Ds3, is here: http://www.dpreview.com/reviews/NikonD3/page32.asp

Uprezzing a file has little or no detrimental effect on real image data, as can be seen in a comparison of an uprezzed file and the original at 200%:

While it is true that the crops look different, I would hesitate to say that the interpolated one lacks something that the original has.

Edited on Jul 27, 2008 at 05:19 PM

Jul 27, 2008 at 05:17 PM

Makten
Offline
Image Upload: Off

p.1 #24 · Canon's high-ISO secret?

brainiac wrote:
I did the uprezzing in the crop above to illustrate that the original comparison on DPReview was a scientifically reliable way of handicapping the higher Mpixel camera. The original, with crops shown at different magnifications in order to handicap the 1Ds3, is here: http://www.dpreview.com/reviews/NikonD3/page32.asp

I see, but I don't think he had any intention to "handicap" the 1Ds. People reading dpreviews probably know enough to understand that a higher resolution camera with the same noise per pixel as a lower resolution camera, will have the lower total noise. Still, I can't see that the 1Ds has lower noise in your improved comparsion. I don't say it isn't better, but I can't see it because I don't know if it is noise or just the label that looks like that. Both cameras probably gives very low noise at ISO 3200, so a better scene for the comparsion is needed in my opinion.

Edit: And, JPG or RAW?

brainiac wrote:
Uprezzing a file has little or no detrimental effect on real image data, as can be seen in a comparison of an uprezzed file and the original at 200%:

While it is true that the crops look different, I would hesitate to say that the interpolated one lacks something that the original has.

Ehhh, what's the difference of "uprezzing" and doing whatever-it-takes to show it at 200%?? At 200%, the image IS "uprezzed". I don't get what you mean. You have just used two different types of interpolation.

Edited on Jul 27, 2008 at 05:32 PM

Jul 27, 2008 at 05:26 PM

brainiac
Offline
Image Upload: On

p.1 #25 · Canon's high-ISO secret?

Makten wrote:
People reading dpreviews probably know enough to understand that a higher resolution camera with the same noise per pixel as a lower resolution camera, will have the lower total noise.

So why does the reviewer say "There's no doubt that the D3 is the winner once you get above ISO 1600".

> Still, I can't see that the 1Ds has lower noise in your improved comparsion.

OK - I will try to explain what I see. The D3 has applied some chrominance noise reduction. That is why you see very limited local colour variation (colour detail) over the range of a few pixels. It's the familiar paint by numbers look. An example of this is that the chrominance NR thinks it has found a big green object and a big purple object in the tree, on roughly the 50 pixel wavelength. That's why there is a big green splodge in the top left of the tree, and a big pink splodge in the lower foliage. The 1Ds3 shows much more of the subtle colour variation there. The big splodges of colour are false, and due to an attempt to average chrominance noise in a way that identifies areas of uniform colour. You can duplicate this effect on Canon files in photoshop by applying the dust & scratches filter at increasing radii to the chrominance channels and not the luminance channel. The 1Ds3 crop is more accurate at retaining the variations of colour in the tree and elsewhere. In other words it needs LESS noise reduction to get an acceptable result, and therefore retains much more colour detail. Colour detail is a good thing, especially if you want a colour photo.

brainiac wrote:
Uprezzing a file has little or no detrimental effect on real image data, as can be seen in a comparison of an uprezzed file and the original at 200%:

While it is true that the crops look different, I would hesitate to say that the interpolated one lacks something that the original has.

Ehhh, what's the difference of "uprezzing" and doing whatever-it-takes to show it at 200%?? At 200%, the image IS "uprezzed". I don't get what you mean. You have just used two different types of interpolation.

No. The 200% non-interpolated file is the original pixels provided by the camera. The fact that in this case each one is represented by 4 monitor pixels should make no difference to the content of the picture, unless there is something very wrong with your monitor. From further away, the 200% crop I show should look identical to a 100% from closeup. This illustrates that interpolating upwards is not very destructive of image data, if at all, because the two crops look pretty much the same in terms of image detail.

Edited on Jul 27, 2008 at 06:01 PM

Jul 27, 2008 at 05:59 PM


FM Forums \| Canon-mount SLRs \| Join Image Upload 1 2 end