The best ISO for Quality and low noise on a 7D?

Submariner wrote in post #14842164
I was told by Canon today that I would be better using ISO 160 than 100 (for stills, flash stills and even video) because it gave a cleaner picture. Reason the Senosor on a 7D was "tuned" to 160 ISO.

Well coming from the film era it sounds crazy, that ISO 160 would deliver a better image quality [all other things being equal] than ISO 100. But discussing my Sony A77 with one of their top techies. He said the A77 Sensor was tuned to ISO 200 - and that it would deliver noticably better Image Quality.

Interested what your 'real world' experiene of this is?

When I first got the 7D I did some noise/resolution tests, here http://www.zen20934.ze​n.co.uk …D%20ISO%20Perfo​rmance.htm

These are very simple tests, but i think reasonably representative. There is very little difference between the RMS noise of 100 to 400 iso, a small upward trend is just discernable above the measearment uncirtany, so I would say the Canon rep is talking Bull.

Having said that I only tested the real whole stop ISO values and not the pushed intermediates. Others have noted the intermediates trade noise for reduced dynamic range. The same effect can be achieved by using the whole stops and changing you exposure. This is known as exposing to the right (of the histogram).

Submariner wrote in post #14842164
I was told by Canon today that I would be better using ISO 160 than 100 (for stills, flash stills and even video) because it gave a cleaner picture. Reason the Senosor on a 7D was "tuned" to 160 ISO.

Well coming from the film era it sounds crazy, that ISO 160 would deliver a better image quality [all other things being equal] than ISO 100. But discussing my Sony A77 with one of their top techies. He said the A77 Sensor was tuned to ISO 200 - and that it would deliver noticably better Image Quality.

Interested what your 'real world' experiene of this is?

I set my camera to full stops of ISO only a while ago. I was shooting midday in winter light and getting horrible amounts of noise. I read, that using intermediate ISO's on the 7d permitted the camera to over/under expose from the full settings to get a correct exposure for the available light settings. As we all know, the 7d is an unforgiving camera when under exposed. The 1/3 stop ISO settings allowed the camera to under expose and give me noise levels that were unacceptable even to an amateur like me.
When I changed to full ISO stops only I began to get the shots I had bought the 7d for.

IMAGE LINK: http://www.flickr.com/​photos/ianhatch/545349​0545/

Ian,

Very nice.

Eight_Blade wrote in post #14842363
For someone "just starting out" don't worry about it.

Er how does one learn, and understand how to maximise the potential of a peice of equipment?

Following that philosophy, I would have been miles ahead just sticking with my Sony A77 for photos, in terms of picture IQ and keeper rates; and light years ahead in terms of video.

I think the 7D is a very difficult beast to master, but I think it can be rewarding if you make the effort to squeeze every drop of blood out of it. Then it might be 5% better than the A77. But straight out of the box with no effort the Sony blows it away.
Hence all my questions - as I don't like having downgraded.
And before you ask I did it for the camera's and the L series lens's physical build quality. And Canon's support [which is light years ahead of Sony].

JohnB57 wrote in post #14843435
Take a look at this. It's counter-intuitive for film shooters (including me), but it seems that on many Canon models, analogue noise is best controlled at 160/320/640 etc. On the video, you'll note the comment that ISO 100 is as noisy as ISO 640 and ISO 2500 cleaner than ISO 2000.

This is exactly what the guy told me - multiples of ISO 160 are best.
As Sony also confirmed they have a sweet spot tuned to their Sensor (which is naturally different as it's a different sensor in their case 200) I guess there must be something in it.

Maybe its like engines, more petrol is not necessarily the best way to deliver HP and Torque - it's the right or 'perfect' mix of petrol and air that makes it fly.

Submariner wrote in post #14848123
Er how does one learn, and understand how to maximise the potential of a peice of equipment?

Following that philosophy, I would have been miles ahead just sticking with my Sony A77 for photos, in terms of picture IQ and keeper rates; and light years ahead in terms of video.

I think the 7D is a very difficult beast to master, but I think it can be rewarding if you make the effort to squeeze every drop of blood out of it. Then it might be 5% better than the A77. But straight out of the box with no effort the Sony blows it away.
Hence all my questions - as I don't like having downgraded.
And before you ask I did it for the camera's and the L series lens's physical build quality. And Canon's support [which is light years ahead of Sony].

Because until you're pushing the limits of your gear, you dont need to major in the minors.

ISO 1600 on a 7D. I don't think there is any objectionable noise here. There certainly will not be any at 100 ISO.

The cat is well lit though, isn't noise most apparent at long shutter openings in low light?

stevenkelby wrote in post #14848260
The cat is well lit though, isn't noise most apparent at long shutter openings in low light?

Yes, and you can see that in some of the pics folks have been posting of the recent Perseid meteor shower. The shots that run 20 secs or more have noise. But it's caused by the sensor getting hot. There are different type of noise (chroma, luma, etc) and some are more easily controlled than others. Long shutter shots are the worst. Correcting that much noise usually results in a mushy image.

My point was that in most cases where you would use a higher ISO to get a reasonable hand-held shutter speed, noise will not be an issue on the 7D.

Roger that, thanks

http://vimeo.com/m/202​39453

Here's a video I watched not too long ago. It's 5Dii specific, but the results are the same. The comments have some pretty interesting reading as to why this happens.

I don't buy what the Canon tech is selling.

IMO, the best way to compare is by measuring dark frame noise levels at all the different ISO settings. In other words, shoot a series with the lens cap on, then import the images in to Adobe Camera Raw and increase exposure in all of them equally, then compare the images by displaying the noise data shown in the histogram against each other.

See Northlight Images website for examples and results with 7D. The problem with this is that it's done with all noise reduction turned off.... so noise is mazimized. In camera and post-production noise reduction processes might do wonders and may be more effective with some ISOs than with others.

Basically, though, there is so little difference from 160 to 200, I'll just use 200. In fact, I've set my 7Ds to full ISO stops only (makes for faster ISO changes) and have used them that way for better part of three years now.

Another consideration.... A while ago Canon posted a white paper about 7D being more sensitive to camera movement or shake than other models, probably because of it's ultra densely crowded sensor and really small photosites (presumably the same would apply to other 18MP models). The recommendation was to use higher ISO than usual, to keep shutter speeds a little faster. I've been doing that for a couple years and it seems to help. In fact, I can't recall the last time I used below ISO 200 with my 7Ds... and I often have them set to ISO 400 or higher.

With some previous Canon models (my old 30Ds, for example) I did find it helpful to use the 1/3 stop ISOs.... With those 160 was "cleaner" than 100, 125 and 200, for example. Enough so, that I didn't use the full stops on those cameras. But I don't find that necessary with 7Ds. There's just so little difference, I'd rather have the higher ISO to help keep shutter speeds up and prevent any movement blur.

Incidentally, Canon's native ISO is 100.... Sony and Nikon (who use Sony sensors) use ISO 200 as their native ISO.

As I understand it, beyond the native ISO all the full stops are hardware derived, and the 1/3 stops in between are software derived (similar to using push or pull processing with film... or underexposing or overexposing, then adjusting exposure after the fact).

7D are very sensitive to underexposure.

amfoto1 wrote in post #14848365
IMO, the best way to compare is by measuring dark frame noise levels at all the different ISO settings. In other words, shoot a series with the lens cap on, then import the images in to Adobe Camera Raw and increase exposure in all of them equally, then compare the images by displaying the noise data shown in the histogram against each other.

I don't think that the difference in dark frame noise is a useful measurement of real picture quality. The dark frame noise can be almost completely removed using "long exposure noise reduction" aka dark frame subtraction in camera.

The dominant noise source in a well exposed photo is usually in the photon counting statistics where the noise is proportional to the square root of the number of photons total. The dark frame noise test collects no photons. In a real photo, the intermediate ISO values still collect more light on the sensor from the longer exposure time than the next whole value above and as such have less photon noise.

I could quantify this when I did the test of using the identical scene through all of the ISOs shot raw with no noise reduction or sharpening. Converting to maximum quality jpg lead to a file size that increased with increasing ISO. More noise means less compressibility.

I am under the impression that noise is the result of poor metering / lighting in a shot regardless of the camera used. For example, I take a shot of a flower & meter correctly for the flower. If the background is shaded, noise appears in the shaded area where an insufficient amount of light falls, but there is no noise visible in the flower itself. Also, if some extra light was directed into the shaded area, the amount of noise visible would be reduced. How much noise will be determined by how well the sensor & meter will work together IMO, but I could be wrong.

Miki G wrote in post #14849269
I am under the impression that noise is the result of poor metering / lighting in a shot regardless of the camera used. For example, I take a shot of a flower & meter correctly for the flower. If the background is shaded, noise appears in the shaded area where an insufficient amount of light falls, but there is no noise visible in the flower itself. Also, if some extra light was directed into the shaded area, the amount of noise visible would be reduced. How much noise will be determined by how well the sensor & meter will work together IMO, but I could be wrong.

While you are right, noise appears first in darker areas, it should not! That just shows that Canon and others have work still to do. No matter what, photographs are made up of a range of tones, including deep shade!

And very deep shadow can show some VERY weird patterning indeed.

So we are still a long way from achieving the perfect sensor.