High ISO - Not such a good deal
Oct 17, 2015 14:28:59 #
Still another page...
(In progess)
High ISO or even super high ISO is touted by many manufacturers but you are not given the full information on this.
First off, this is not about trying to tell you not to use ISO. This is about giving you a fair warning of the two major problems created.
The first one is known to all "Noise".
The second is about the dynamic range and that is largely unmentioned simply because few are aware of it.
When you the specification of a camera you see the dynamic range of a camera as a fixed numer when in reality this is a variable number that tends to go down quickly as soon as you crank up the ISO. This is one of the causes for the higher noise by the way*.
While some will dismiss it as they use JPG so, who cares, I am limited anyway, well, you should care. Your preciously limited JPG is influenced even more than you first realize. To create a JPG a camera uses the sensor capture, in other words, the raw data the manufacturer does not want you to have. The result is that a JPG camera produced is more detoriated than a raw capture.
Once again, this is not about putting you off high ISO. You should still use it, just with the full knowledge of what takes place at the sensor level.
-----
* Noise is always present, even when using the optimal status, it is just less visible. Noise is also present in the highlights. When you use high ISO weak functioning pixels on the sensor start failing, enhancing the noise. This additional noise is not random by the way but always at the same place so can be corrected. For the few who are starting to wonder f I am looking at 'red noise', I am not. Red noise is created by long exposure time, overheated failing pixels, not High ISO, that one is corrected with 'black frames' in post processing.
Opinions - comments?
(In progess)
High ISO or even super high ISO is touted by many manufacturers but you are not given the full information on this.
First off, this is not about trying to tell you not to use ISO. This is about giving you a fair warning of the two major problems created.
The first one is known to all "Noise".
The second is about the dynamic range and that is largely unmentioned simply because few are aware of it.
When you the specification of a camera you see the dynamic range of a camera as a fixed numer when in reality this is a variable number that tends to go down quickly as soon as you crank up the ISO. This is one of the causes for the higher noise by the way*.
While some will dismiss it as they use JPG so, who cares, I am limited anyway, well, you should care. Your preciously limited JPG is influenced even more than you first realize. To create a JPG a camera uses the sensor capture, in other words, the raw data the manufacturer does not want you to have. The result is that a JPG camera produced is more detoriated than a raw capture.
Once again, this is not about putting you off high ISO. You should still use it, just with the full knowledge of what takes place at the sensor level.
-----
* Noise is always present, even when using the optimal status, it is just less visible. Noise is also present in the highlights. When you use high ISO weak functioning pixels on the sensor start failing, enhancing the noise. This additional noise is not random by the way but always at the same place so can be corrected. For the few who are starting to wonder f I am looking at 'red noise', I am not. Red noise is created by long exposure time, overheated failing pixels, not High ISO, that one is corrected with 'black frames' in post processing.
Opinions - comments?
Oct 17, 2015 14:46:28 #
coj
Loc: NJ, USA
But it can be easily corrected in Windows using their picture editing. Granted it dulls some detail, but then sharpening can be added (or contrast) and noise reduced again until you get the best image possible.
Oct 17, 2015 14:52:38 #
coj wrote:
But it can be easily corrected in Windows using their picture editing. Granted it dulls some detail, but then sharpening can be added (or contrast) and noise reduced again until you get the best image possible.
You can correct the dynamic range missing data? That is news to me.
Oct 17, 2015 15:04:58 #
coj
Loc: NJ, USA
I was referring to noise reduction.
Rongnongno wrote:
You can correct the dynamic range missing data? That is news to me.
Oct 17, 2015 15:09:49 #
coj wrote:
I was referring to noise reduction.
We have no dispute there albeit likely on the method used.
Noise created by high ISO even more so because it seems to be predictable. (I need to verify that by reliable sources hence: seems)
Oct 17, 2015 15:14:55 #
coj wrote:
But it can be easily corrected in Windows using their picture editing. Granted it dulls some detail, but then sharpening can be added (or contrast) and noise reduced again until you get the best image possible.
All you are doing is throwing image detail and adding edge distortion, you are not correcting anything.
Oct 17, 2015 15:41:28 #
amfoto1
Loc: San Jose, Calif. USA
You are correct...
Set up a test. Camera on a tripod, exact same subject and lighting, set the camera to shoot RAW and aperture priority to allow the camera to adjust for the variations in ISO... For these purposes, full ISO steps (100, 200, 400, 800, etc.) are adequate. Being careful to keep everything the same, except for the ISO you set and the shutter speed the camera uses, take a series of test shots using different ISOs from low to high and download the image files to your computer.
Now look at the details of those image files. The higher the ISO, the smaller the file size. That's the resolution gradually decreasing with each step higher ISO. On close inspection you also can see the difference in the images.
But, hey, what you gonna do? When you need high ISO, you need high ISO.
Best to learn your limits in advance, whatever camera you use. And, yes, I mean your limits... not the camera's. How high ISO are you willing to use, depending upon your purposes for the image? (I'd rather the camera manufacturer enable a camera to take images at ridiculously high ISOs if I wish, than have them limit me.)
Also, learn noise reduction techniques, in order to get the best possible images you can at higher ISOs. NR has come a long way in recent years. Back in 2004, I wouldn't use a camera above ISO 800.... in 2008, my limit moved to ISO 1600, and again to 3200 in 2010. Now I sometimes use ISO 6400. There have been advancements both in the cameras and in the softwares we use for noise reduction.
Still, it's all better than film. The fastest slide film I used was ISO 200... and much of it was ISO 100 or even ISO 50. When I needed faster, I'd use ISO 400 print film (and ISO 160). Occasionally I'd push 400 B&W film to ISO 800 or 1600. But that was about the limit. There were ISO 1600 and 3200 films... but they basically sucked. Digital today is far better at high ISOs, than film ever was.
It's generally best to shoot RAW and avoid under-exposure as much as possible. If you shoot JPEG, you are pretty much stuck with whatever NR was done to the image in-camera, might be able to do a little more in post-processing... but not much. On the other hand, if you shot RAW, you can work in 16 bit mode and apply more extensive and even image-specific NR, or selective NR to parts of the image, just to the shadows or just to the highlights, or even to individual color channels.... with potentially much better results. Also, archive those RAW files. NR keeps improving and there's a good chance there will be a better process in years to come, that you can go back and re-process older images, if you wish.
How much high ISO noise is tolerable also will depend upon the final use of the image. If it's just being displayed small online at Internet resolutions, or being used to make a 5x7 print, much higher ISO can be tolerated, than - say - if you are planning to make a 20x30 print or it will be used for a billboard.
Also, ultra-high ISO images might be usable if converted to black & white. A lot of B&W films were pretty grainy, so we are accustomed to that... and color noise looks like grain, after it's converted.
Set up a test. Camera on a tripod, exact same subject and lighting, set the camera to shoot RAW and aperture priority to allow the camera to adjust for the variations in ISO... For these purposes, full ISO steps (100, 200, 400, 800, etc.) are adequate. Being careful to keep everything the same, except for the ISO you set and the shutter speed the camera uses, take a series of test shots using different ISOs from low to high and download the image files to your computer.
Now look at the details of those image files. The higher the ISO, the smaller the file size. That's the resolution gradually decreasing with each step higher ISO. On close inspection you also can see the difference in the images.
But, hey, what you gonna do? When you need high ISO, you need high ISO.
Best to learn your limits in advance, whatever camera you use. And, yes, I mean your limits... not the camera's. How high ISO are you willing to use, depending upon your purposes for the image? (I'd rather the camera manufacturer enable a camera to take images at ridiculously high ISOs if I wish, than have them limit me.)
Also, learn noise reduction techniques, in order to get the best possible images you can at higher ISOs. NR has come a long way in recent years. Back in 2004, I wouldn't use a camera above ISO 800.... in 2008, my limit moved to ISO 1600, and again to 3200 in 2010. Now I sometimes use ISO 6400. There have been advancements both in the cameras and in the softwares we use for noise reduction.
Still, it's all better than film. The fastest slide film I used was ISO 200... and much of it was ISO 100 or even ISO 50. When I needed faster, I'd use ISO 400 print film (and ISO 160). Occasionally I'd push 400 B&W film to ISO 800 or 1600. But that was about the limit. There were ISO 1600 and 3200 films... but they basically sucked. Digital today is far better at high ISOs, than film ever was.
It's generally best to shoot RAW and avoid under-exposure as much as possible. If you shoot JPEG, you are pretty much stuck with whatever NR was done to the image in-camera, might be able to do a little more in post-processing... but not much. On the other hand, if you shot RAW, you can work in 16 bit mode and apply more extensive and even image-specific NR, or selective NR to parts of the image, just to the shadows or just to the highlights, or even to individual color channels.... with potentially much better results. Also, archive those RAW files. NR keeps improving and there's a good chance there will be a better process in years to come, that you can go back and re-process older images, if you wish.
How much high ISO noise is tolerable also will depend upon the final use of the image. If it's just being displayed small online at Internet resolutions, or being used to make a 5x7 print, much higher ISO can be tolerated, than - say - if you are planning to make a 20x30 print or it will be used for a billboard.
Also, ultra-high ISO images might be usable if converted to black & white. A lot of B&W films were pretty grainy, so we are accustomed to that... and color noise looks like grain, after it's converted.
Oct 17, 2015 15:47:59 #
High ISO is just another tool. You must decide what is best for the conditions you have to work with. Do you use a wide open aperture and lose the depth of field you need or do you use high ISO and get noise? Or do you use a slow shutter speed and get blurred action shots so you can have low ISO and no noise? Best to know your camera so you can choose the right compromise to get the as close to the picture you want. I do know that with the higher ISO of my D7100 I can outdo my D50 hands down. - Dave
Oct 17, 2015 16:00:13 #
Rongnongno wrote:
Still another page... br br (In progess) br br H... (show quote)
I've gotten good results at 3200 with my Df.
Oct 17, 2015 16:01:57 #
wilsondl2 wrote:
High ISO is just another tool. You must decide wh... (show quote)
You are absolutely correct. I thought I emphasized that first in the opening statement and the conclusion.
I wrote:
First off, this page is not about trying to tell you not to use high ISO. This is about giving you a fair warning of the two major problems created.
and
Quote:
Once again, this is not about putting you off high ISO. You should still use it, just with the full knowledge of what takes place at the sensor level and the influences it has on your initial capture, whatever the format.
Please note that this is a work in progress. Spelling grammar and information needs to be corrected and triple verified (twice verified so far - not using UHH)
Oct 17, 2015 20:40:07 #
R -- so like grain in film -- sometimes good sometimes not. I recall pushing Tri X and developing in Dektol for max grain. Or using Panatomic or Plus X for almost none to mention. I don't do a lot of PP so shoot JPG Fine / Large. Pretty satisfactory. If I want PP, Elements offers me the option of opening JPG in Raw, even if I only shot JPG. Seems to let me have access to all the tools what's that all about? With my "new" D 750, I can shoot both at the same time. I'm not up on all of this, but what do I gain if I do? Thanks in advance for your patience.
Amphoto1, from the time warp that I am in, Tri X at 2400, Accufine, carefully controlled, not bad!
Amphoto1, from the time warp that I am in, Tri X at 2400, Accufine, carefully controlled, not bad!
Oct 17, 2015 21:04:49 #
quixdraw wrote:
R -- so like grain in film -- sometimes good somet... (show quote)
Not quite. Comparing noise to grain is an error.
Grain was random (as is noise) but it was of various sizes and blocking the light. Noise is created randomly over the sensor pixel pattern, always of the same size and randomly colored (blue/green shades) so transparent red is weak or missing).
So? Let's compare noise vs grain
Quality -{}- Grain -{}- Noise
Random-{}- Yes, free -{}- Yes, over a matrix
Size -{}- Random-{}- Always the same (one pixel)
Opaque -{}- Yes -{}- No, colored GrBG (red is missing or weakened)
The only way to have grain in this new digital world is to create it using PP, even so that will be 'pseudo' random.
Oct 17, 2015 21:08:45 #
Rongnongno wrote:
Still another page...
When you the specification of a camera you see the dynamic range of a camera as a fixed numer when in reality this is a variable number that tends to go down quickly as soon as you crank up the ISO. This is one of the causes for the higher noise by the way*.
Opinions - comments?
When you the specification of a camera you see the dynamic range of a camera as a fixed numer when in reality this is a variable number that tends to go down quickly as soon as you crank up the ISO. This is one of the causes for the higher noise by the way*.
Opinions - comments?
While DxO and many others who test cameras report dynamic range of the cameras I don't see a manufacturer specify the camera dynamic range in the specification.
Oct 17, 2015 21:08:51 #
Oct 17, 2015 21:12:51 #
If you want to reply, then register here. Registration is free and your account is created instantly, so you can post right away.