Raw, noise and D300

HarlanBear

Mod edit: I merged these two threads--RSinMadrid.

HarlanBear

Raw, noise and the D300
I originally posted this is Finishing School, but realize it may belong here.

I recently upgraded to the Nikon D300 (from D70 and love it) and have been doing research on the noise reduction settings in the D300 but have now learned that some settings may not work for me. I shoot exclusively in RAW and do conversions in ACR and post in PSCS3. The noise settings in the D300 include "High ISO Noise Reduction" and "Long Exposure Noise Reduction". I believe that both these adjustments have to do with jpeg images and will have no real effect on my RAW images in-camera. Is that correct?

And if so, then however the camera handles noise for shooting RAW is how it handles it and if there is excessive noise I will need to deal with it in post. The D300 is certainly better at noise than the D70, but I still do find some noise in long exposures, which is about all I've done so far.

Currently I'm using Noise Ninja for all noise handling and do like it, but haven't tried the Capture NX I got with the camera.

Any collaboration of this information or more enlightenment would be appreciated. Thanks!

Bigmitch

I use Canon not Nikon, but the noise parameters your mention have nothing to do with RAW or JPEG, High ISO is just that, where you dont have the option of a flash (such as in a church) you would set the ISO to 800-1600 in order to get an acceptable exposure. This cant be changed in RAW processing or we all would be doing it.

Long Exposure Noise Reduction would be used if you were to set your camera shutter to open for a minute (for example night shots), being open this long generates noise on a digital SLR so this setting causes the camera to perform noise reduction (on my camera again it takes equal time to the shutter duration) before another pic can be taken.

Again this has nothing to do with JPEG or RAW.

Hope this helps.

arodney

IF you're shooting Raw, I don't think these settings from the camera do a thing. This is for JPEG generation.

HarlanBear

HarlanBear wrote:

Mod edit: I merged these two threads--RSinMadrid.

Thanks for your help, Richard. Greatly appreciated

Bigmitch wrote:

Long Exposure Noise Reduction would be used if you were to set your camera shutter to open for a minute (for example night shots), being open this long generates noise on a digital SLR so this setting causes the camera to perform noise reduction (on my camera again it takes equal time to the shutter duration) before another pic can be taken.

Again this has nothing to do with JPEG or RAW.

Actually, I think it does, as Andrew points out. Unless we're missing somthing here. That's why I posted the question. My research points to the jpeg creation and not Raw, but wasn't sure.

colourbox

Bigmitch wrote:

I use Canon not Nikon, but the noise parameters your mention have nothing to do with RAW or JPEG, High ISO is just that...

Let's hear from someone who has used that Nikon. The feature was not "High ISO," but "High ISO Noise Reduction." It sounds like a form of in-camera noise processing, not simply turning up the ISO.

And I thought Long Exposure Noise Reduction (the subtraction method, right?) could be saved into a RAW by a camera, but I could be wrong there as I don't use the D300. [Update: A Nikonians thread found on Google implies that D300 LENR is saved in RAW.]

Also, correct me if I'm wrong but I didn't think Long Exposure Noise Reduction is something that most raw processors can do on top of their standard noise reduction, you'd have to take two separate exposures and subtract in Photoshop...right? If so, and if you need this, then it would be good to do it in camera.

HarlanBear

arodney wrote:

IF you're shooting Raw, I don't think these settings from the camera do a thing. This is for JPEG generation.

Thanks, Andrew. I was hoping to hear from you or one of the other real gurus here at dgin. This is what I kinda figured, and perhaps I need to experiment with Raw and jpeg to get the definative. But this sounds right.

And if anyone else has actual experience with this or has done testing, I'd love to hear.

HarlanBear

colourbox wrote:

And I thought Long Exposure Noise Reduction (the subtraction method, right?) could be saved into a RAW by a camera, but I could be wrong there as I don't use the D300. [Update: A Nikonians thread found on Google implies that D300 LENR is saved in RAW.]

Also, correct me if I'm wrong but I didn't think Long Exposure Noise Reduction is something that most raw processors can do on top of their standard noise reduction, you'd have to take two separate exposures and subtract in Photoshop...right? If so, and if you need this, then it would be good to do it in camera.

After reading colourbox's update above I'm now more confused than before. Seems to me that the idea of High ISO is a setting that Raw could not change, but having the camera computer process a long exposure at lower ISO and producing the same type of results as a post processing method, like the subtraction method using a separate "frame" within the camera, could be possible. So I will try to test this tonight and see what's up.

Internaut

The dark frame subtraction noise reduction is applied to longer exposures only (anything over 8 seconds on my Olympus, others apply different criteria). This typically takes as long for the camera to perform as it did to take the exposure in the first place.

Any type of high ISO noise reduction is done on the out of camera JPEG only (a notable exception being Sony who also mess with the raw capture from their A700). As such, you would need to do your own noise reduction when doing your own post processing on a raw capture.

HarlanBear

I did some testing last night and it does appear that while the High ISO NR and the Long Exposure NR do effect jpegs. there is no real effect to RAW images with either settings. This was not the most scientific test, but all things being equal, this was my findings.

If anyone else does any tests, I'd love to hear the results.

ziggy53

HarlanBear wrote:

I did some testing last night and it does appear that while the High ISO NR and the Long Exposure NR do effect jpegs. there is no real effect to RAW images with either settings. This was not the most scientific test, but all things being equal, this was my findings.

If anyone else does any tests, I'd love to hear the results.

High ISO noise reduction is probably applied during the processing to JPG, so I wouldn't expect any difference in RAW, which means you can leave the high ISO reduction on because it doesn't affect RAW.

Long exposure noise reduction probably does affect RAW. You can test the degree yourself:

Make sure you start with a fully charged battery for each test. These tests can be a problem if you don't have enough power for the duration of the test. You are wholly responsible for the consequences should you perform the test. Long duration exposures can be hard on an imager.

Leave the long exposure NR off and cap the lens. Set your exposure manually and expose for 15 minutes in RAW at ISO 800. The total exposure time should be 15 minutes. (I put the camera in a bag and set that in a dark corner of the house.)

Now set the long exposure NR on and duplicate the exposure of 15 minutes and RAW/ISO 800. The exposure should take 15 minutes followed by a 15 minute "black (or dark) frame", for a total of 30 minutes.

Bring both files into ACR and allow neutral/default processing, but save as a 16 bit TIFF.

Open the file without long exposure NR in PS and allow "Auto Levels". Open a second instance of the same file and visually adjust the Levels until you achieve similar results to the Auto Levels, but record your procedure. (A Levels adjustment of 5 worked fairly well for me in a 30 second test, but you should derive your own amount.)

Open the second file, with the long exposure NR on, and process according to the same procedure as you previously recorded.

You should now have a visual comparison of the degree of NR the long exposure NR applies to an image at 15 minutes exposure and ISO 800.

BTW, I tested at 30 seconds which was not long enough to activate the long exposure NR on my Canon XT, so the file with the NR "on" was actually noisier than the first file without NR. It just proved that long exposure noise does increase with subsequent long exposures due to "hot" pixels. At some point, the imager should achieve equilibrium and not get much noisier.

I will run a test of 15 minutes sometime, when I construct a method of holding the shutter open in "bulb". Don't hold your breath waiting for me but proceed with your own tests if you desire. Understand that these tests can take their toll on an imager and it is possible to get "stuck pixels" or "dead pixels" as a result.

HarlanBear

Thanks, Ziggy. I appreciate your input on this. Very interesting information and test procedures. But I’m wondering from a practicality stand-point what this might do for me. Or anyone dealing with long exposures and the resultant noise. Beyond the obvious idea of having more knowledge about this situation.

Perhaps this is peripheral to the OP, but a practical question is how to deal with this in the real world. I know about the two image noise reduction procedure wherein you take a second image with the cap on for the same exposure duration as your primary image and “merge’ the two in Photoshop. But I wonder if the results are any better than using noise reduction software. I currently use Noise Ninja on a duplicate of the image on another layer, and of course that will soften the image, depending on how much reduction is needed. I then deal with it by either sharpening a bit or masking off portions of the duplicate, noise reduced layer.

Do you have any experience with the two different methods?

One aside is that I ran my tests using only a 10 second exposure since that is a little longer than the time needed for Long Exposure NR to kick in, so that would certainly have less effect on an image than a much longer exposure.

ziggy53

HarlanBear wrote:

Thanks, Ziggy. I appreciate your input on this. Very interesting information and test procedures. But I’m wondering from a practicality stand-point what this might do for me. Or anyone dealing with long exposures and the resultant noise. Beyond the obvious idea of having more knowledge about this situation.

Perhaps this is peripheral to the OP, but a practical question is how to deal with this in the real world. I know about the two image noise reduction procedure wherein you take a second image with the cap on for the same exposure duration as your primary image and “merge’ the two in Photoshop. But I wonder if the results are any better than using noise reduction software. I currently use Noise Ninja on a duplicate of the image on another layer, and of course that will soften the image, depending on how much reduction is needed. I then deal with it by either sharpening a bit or masking off portions of the duplicate, noise reduced layer.

Do you have any experience with the two different methods?

One aside is that I ran my tests using only a 10 second exposure since that is a little longer than the time needed for Long Exposure NR to kick in, so that would certainly have less effect on an image than a much longer exposure.

Long exposure noise reduction is based on the predictable nature of individual chips in particular situations. The manual process of doing a "dark frame" and then creating a subtraction mask is doing exactly the same thing. In both cases you are working against known and specific noise created by "profiling" your own chip in those circumstances.

In the case of noise reduction software, it is designed to handle "random" noise, and most modern software does a pretty good job at that task. Since long exposures generate non-random noise, it is at best a guess whether random noise reduction software would be as effective, but probably not. If it is effective enough is a very personal decision that only you can make.

My experience with the Canon XT is that even with a 30 second exposure of darkness, the resulting "dark noise" at ISO 800 is fairly negligible. A 15 minute exposure would probably be quite different.

Another consideration is subject matter. If you are shooting an image with a fairly busy subject that consumes the entire frame you are less likely to notice the problem than if your image is mostly dark or black, where the long exposure noise is more obvious.

Internaut

I can only speak from the point of view of someone who's idea of a long exposure is hardly ever more than 30 seconds. Whether it be a point and shoot (a Panasonic LX2 in my case) or an SLR, the dark frame subtraction stuff just works...

I've seen this discussed elsewhere and the reason for switching off the in camera dark frame subtraction (and having to do it yourself later) is to avoid the risk of "missing that shot" while the camera spends a minute or two (or 15 in the case of astro-photographers) on the processing. Seems reasonable enough as I discovered how important timing is just a few hours ago... I missed a rainbow from Heaven itself in this afternoon's extreme weather because it went away while I was taking my LX2 (which goes everywhere with me) out of the bag! Ho hum :cry

HarlanBear

Thanks again, Ziggy. Great information, and I will certainly give the "dark frame" technique a try. Makes sense that working with information from the cameras sensor would have a better result in certain cases.

As always, I know I can count on the folks here at Dgrin for helpful information and timely responses.

dadwtwins

Just my 2 cents here.

I am a Nikon shooter and own the D300. The High ISO settings blur the sharpness of picture way too much and does not allow you to control how much blur effect is applied to the picture.

I would rather shot with no "in-camera" noise reduction and do all my noise reduction in post processing. I feel that I have much more control over how I want the picture to look and to comprimise the sharpness with the noise. In post processing, you can also apply noise reduction in localized areas compared to the "in-camera" function.

Since the D300 does a good job shooting at ISO 800 compared to the D200, I accept some of the noise.