D300 noise between 200 and 400iso?

[JohnC]

John,

Cool! That was pretty much my setup except I use two florsecent 5500k bulbs in reflectors.

I'll take a look at them as well when I get home.

I just clicked on the link to look at the files and notice that they are not named according to which picture control was used. I don't think Gabor's program reads any picture controls. So I'm guessing the first one is Standard then followed by Neutral, Vivid, D2XMODE1, 2, 3?

MikeT

I shot:

Standard - 0518
Neutral - 0519
VividI - 0520
D2X1 - 0521
D2X2 - 0522
D2X3 - 0523

Hope this works! :smile:

 

[MikeT]

I shot:

Standard - 0518
Neutral - 0519
VividI - 0520
D2X1 - 0521
D2X2 - 0522
D2X3 - 0523

Hope this works! :smile:

Me too!

Thanks John,

Miket

 

[JohnC]

Me too!

Thanks John,

Miket

How do they look to you? Good for test samples? I'm a rookie at this test shot stuff. :smile:

 

[MikeT]

How do they look to you? Good for test samples? I'm a rookie at this test shot stuff. :smile:

They look good John.
The only thing that may be a concern is distingushing the noise of the images from the grain of the paper. We'll see what Gabor says about them. I have his program but can't run it because it crashes on my Vista machine.


I'm going to post a set as well as backup to your's for comparison between bodies.

Thanks for doing the test and posting your images.


MikeT

 

[Panopeeper]

I downloaded and cheked out John's images. I don't see any relevant difference between the raw files.

There is a slight "bump" in the exposure from "vivid": "standard" and "neutral" are a tiny bit lower exposed than all the following shots. However, the difference is only about 0.06 EV (one sixteenth of a stop) - this is not enough to induce any noticable effect.

The ISO values are recorded in the Nikon-specific metadata as well (beside Exif). The smallest difference, which can appear there is IIRC 1/12 EV. These recordings too show identical ISO in the test schots.

I am convinced, that the exposure (i.e. aperture and shutter) will not be implicitely changed due to any setting. The camera can change only the applied ISO.

All in all: these settings are like WB, contrast, saturation, sharpness: affecting the in-camera JPEG creation (even those embedded in a raw file), but do not affect the raw data directly; it is CNX, which carrier out the requested functions.

 

[MikeT]

Anoter Set In Additon to John's

Gabor,
I brought home a couple of steel blue painted signs from work to use as a noise test subject.

They look like this:

Subscribe to see EXIF info for this image (if available)

Here's a screen shot from NX2 with all seven images loaded:
View attachment 252761

The NEF files can be downloaded from here.

All shots were taken with two 5500k flourescent lights, camera on manual, tripod, cable release, ISO set to sun and mup on. All were taken in a span of 60 seconds from start to finish.

The picture control modes are labeled on the NEF files:
Standard
Neutral
Vivid
DX2MODE1
DX2MODE2
DX2MODE3
Standard


MikeT

 

[Panopeeper]

Mike,

the above shots are so well-exposed, that one can not talk about noise at all.

Anyway, they too shows, that these settings do not affect the raw data, in other words the noise does not depend on these settings.

I loaded the "standard" and "vivid" in ACR; they are equal. The difference visible in your NCX samples must come from the raw conversion: NCX recognizes the setting and acts accordingly. I guess "vivid" means "more contrasty" and/or "more saturated".

 

[JohnC]

Mike,

the above shots are so well-exposed, that one can not talk about noise at all.

Anyway, they too shows, that these settings do not affect the raw data, in other words the noise does not depend on these settings.

I loaded the "standard" and "vivid" in ACR; they are equal. The difference visible in your NCX samples must come from the raw conversion: NCX recognizes the setting and acts accordingly. I guess "vivid" means "more contrasty" and/or "more saturated".

VividI is the most saturated mode on the D300. :smile:

 

[MikeT]

Mike,

... The difference visible in your NCX samples must come from the raw conversion: NCX recognizes the setting and acts accordingly. I guess "vivid" means "more contrasty" and/or "more saturated".

Yes and the fact that I had +5 in camera sharpening on all shots.

Here is a screen shot of the Vivid file loaded in both ACR and NX2 with sharpening turned off:

Subscribe to see EXIF info for this image (if available)

In my tower shot first posted here showing some noise was underexposed somewhat. So a better exposure would lessen the noise to a certain extent.


Thanks for your help and report Gabor.

MikeT

 

[Panopeeper]

There is one more issue to be cleared up re the D300: which are the real ISO settings.

ISO 1600 is real. ISO 3200 is a numerical derivative. However, 3200 does not look like a derivative of 1600. I guess 2000 is the top real ISO.

If someone posts *any* image with ISO 2000 and 2500, 14 bit, losslessly compressed, then I can determine if it pays to use ISO 2000 or even 2500.

Furthermore, I don't know if the other fractional settings, like 250 and 320 are real or numerical derivatives. For example my Canon 40D fakes all fractional settings: 250 is 200 multiplied by 1.27, 300 is made out of 400 multiplied by (1/1.27), etc. This means, that there is no reason to use ever the fractional ISO settings when working with raw files.

I would need one ISO 250 and one 320 image, anything, 14bit, losslessly compressed.

Uncompressed is good too, but why would not one compress these large files?

 

[JeffKohn]

AFAIK the incremental ISO settings on Nikon DSLR's are "real" unlike the consumer/prosumer Canons. At least, this has been confirmed on some past Nikon DSLR's though I don't know if it's beeen specifically confirmed on the D300. I'll try to get you some test shots this afternoon/evening if I remember and get a chance to do so before someone else does.

 

[Panopeeper]

Jeff,

if you can make these shots, I can take a look at them (this is a simple thing, one needs to verify only the fine histograms). The same question came up on DPReview, someone promised to make the shots for several days ago but did not.

 

[JeffKohn]

Sorry, been a little busy with work. I did fire off a couple of quick shots, but I don't have any easy way to make them available so I just took a screen-cap of both NEF's loaded in Rawnalyze. I just realized you wanted 250 and 320, not 200 and 320, if these don't help I can try again this weekend.

http://www.pbase.com/image/103654466/original

 

[Panopeeper]

Jeff,

I am sorry for the trouble; however, these shots are not useful. They are 12bit, which is on its own a derivative of the 14bit version.

Only the 14bit data shows the numerical factoring.

 

[MMarz]

I've spent the past three weekends out on the beach shooting BiF's and power boats..all at last light of the day, so ISO 400 - 640 is pretty much standard. I haven't noticed any excessive noise to speak about.

Shooting RAW, in camera NR off, LR2 for conversion.

 

[MikeT]

I'll backup Jeff's post with a set also.


The photo looks like:

Subscribe to see EXIF info for this image (if available)

The items are: 18% gray card, blue sign and a 6500k calibrated tile.
Camera histogram was centered nicely but NX2 looked a little underexposed.

The NEF's can be downloaded here.


MikeT

 

[Panopeeper]

Mike, thanks for the test shots; they are very helpful.

1. The 1/3 ISO steps are real. There is no reason not to use them in raw as well.

2. The top real ISO is somewhere between 1600 and 2000. I don't know if 1/6 steps are selectable, perhaps the camera can select them on its own.

Look at following fine histograms, 1600, 200, 2560 and 3000. If one loads these in Photoshop, magnifies and counts the red or blue colored columns in some reasonable range, one comes to something like

ISO 1600: five out of six levels are used = 0.83% (this is the same at all lower ISOs).
ISO 2000: 106 outof 145 used = 0.73%.
ISO 2560: 82 out of 142 used = 0.58%
ISO 3200: 55 out of 118 used = 0.47%

ISO 1600 / ISO 2000: 1.14, close to 1/6 stop
ISO 2000 / ISO 2560: 1.27, 1/3 stop
ISO 2560 / ISO 3200: 1.24, close to 1/3 stop

Thus ISO 2000 is halfway real, halfway fake (numerical derivative).

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

 

[Longshanks]

Very good info Martin!
I have seen noise in some of my ISO 200 shots while others look fine. Changing to different picture controls in NX2 on one of my images did make a difference in noise levels. It looks like Neutral and D2XMODE1 shows less while the other picture controls shows more.

Here's an image of mine taken at 120 sec, f/8, Nikon Cir-PL filter, Vivid picture control with in camera sharpening set to 5. No changes were made in post other than saving for web.

Full photo:

Subscribe to see EXIF info for this image (if available)

100% Crop
View attachment 252765


The NEF file can be downloaded here if anyone wants to take a look at it.

Question: Does any under exposure show more noise? I always try to expose ETTR but sometimes I slip up.


MikeT

I have downloaded your nef and viewed it with NX2.
You did some PP here and it seems that the high pass filter was set to 6.92.
When I turned it off the grain was gone.