D500 ISO testing

jeff5150

Hi folks,
So, Im still getting to know my D500, but, I've seen some results grainer than expected in the 800-3200 ISO range (I didn't expect much above that). With that, a couple questions...
1) Anyone else had issues?
2) Can someone share with me a best practice for a true test of ISO performance? I have a D750 as well, so, was thinking I would do a comparison as the D500 should perform better (assumption).

Thanks, looking forward to any advice/tips you may have.

Ironheart

Here's a shot at ISO 204800

Go ahead and click on it and check out the full sized on my Flickr account. It really depends on what you are shooting, how much light there is and how you post process. Do you have some photos to post?

WestEndFoto

Jeff5150, I cannot imagine a D500 camera, which is DX, performing better on ISO, than a D750, which is FX and current generation FX to boot.

jeff5150

WestEndFoto said:

Jeff5150, I cannot imagine a D500 camera, which is DX, performing better on ISO, than a D750, which is FX and current generation FX to boot.

I'm not saying you're wrong, but, the D500 is a newer gen processor, and better ISO sensitivity (two extra stops on the D500). So, yes, Im assuming it should perform relatively the same if not better.

jeff5150

Ironheart said:

Here's a shot at ISO 204800



Go ahead and click on it and check out the full sized on my Flickr account. It really depends on what you are shooting, how much light there is and how you post process. Do you have some photos to post?

Hi Ironheart...I don't see the link. Can you send again or add me on flickr? I am jeffmiller_photography . I haven't posted the images in question yet, I've just been shaking my fist at them.

WestEndFoto

Yes, it is newer generation, but the improvements are incremental and only a fraction of the difference between the inherent characteristics of the sensor sizes. I suspect that the issue is your expectations.

The D500 is a fabulous camera though.

jeff5150

WestEndFoto said:

Yes, it is newer generation, but the improvements are incremental and only a fraction of the difference between the inherent characteristics of the sensor sizes. I suspect that the issue is your expectations.



The D500 is a fabulous camera though.

While you may be correct, that isn't helpful to just assume I'm the issue. Isn't it possible there is something else wrong? Again, my original question is asking for advice on best practices to test if there is a performance issue with my particular camera. Perhaps I got a bad one compared to others off the manufacturing line.

heartyfisher

The results from a bunch of testers on the Internet seem to show that the D750 does better at high ISO. Although the D500 is very close. and yes I am planning to get a D500 over a D750. Would be interesting to see your test results once you have worked out how you want to proceed.

Sports

My experience with the D500 is the same, I suppose. I did not expect FX performance, though, and it doesn't deliver FX performance.
After seeing some impressive test shots, and Ironheart's "magic" shot at ISO 204800, it came as a bit of a surprise that my own results from the 1600-3200 ISO range were mixed, and that noise reduction had to be applied more often than expected in post to make them acceptable. Too high expectations?
On the other hand I took a picture of a hedgehog at ISO 8000 (in darkness), and was almost blown away by an in focus, almost "daylight good" picture (without any postprocessing).
I don't know if these can be called "issues" but I still haven't figured out how to consistently get good results at high ISO.
Q: If you believe your D500 is faulty, then how can you set up a reference shoot to get one photo that can be compared to "the same" photo taken by @Ironheart or some other guy that knows his stuff?

Paperman

We could easily tell if the problem was with the D500 or your expectation from it, had you included an image ...

Ironheart

Repost, for some reason the comment above won't display the photo:

Ironheart

@jeff5150 We do need to see a photo in order to get a basis. I would like to know what ISO, shutter speed, aperture, and post it to flickr with the EXIF intact. Even better would be a raw file. I can shoot some books in my office with a single bulb as lighting, should be easily reproduced. The above shot is at 3EV (f/4, 1/1000th, ISO 204800) so almost no light

WestEndFoto

jeff5150 said:

WestEndFoto said:

Yes, it is newer generation, but the improvements are incremental and only a fraction of the difference between the inherent characteristics of the sensor sizes. I suspect that the issue is your expectations.



The D500 is a fabulous camera though.

While you may be correct, that isn't helpful to just assume I'm the issue. Isn't it possible there is something else wrong? Again, my original question is asking for advice on best practices to test if there is a performance issue with my particular camera. Perhaps I got a bad one compared to others off the manufacturing line.

Hi Jeff. I do think that you are part of the issue if you have an expectation that your D500 will outperform, in respect of noise at high ISO settings, an FX camera from a recent generation. I feel confident making this assertion for any FX camera from the D4 forward.

Sensors have been making dramatic improvements in the last fifteen years. However, the latest generations have only produced small improvements. The improvement from the D4s to the D5 was very small. These improvements are too small to compensate for the one stop advantages the FX sensors have over DX sensors. I am even sceptical that a DX sensor ten years from now will best an FX sensor from today - though I am hopeful that we will see at least that level of improvement.

That said, there COULD be an issue with your particular camera. I think that the easiest way to test that would be to take it to a camera store, take a picture in a dark corner with your camera, put your lens on another D500, and take another picture with exactly the same settings. Then compare the images at home. Do that a few times and you will get an idea if you have a problem or not.

Ironheart

The D750 is about a stop better than the D500. Shouldn't really be noticeable unless you are zooming in to 100%. Remember this would be the equivalent of 4' by 8' print and looking at it from a foot away. Where the D500 wins is at the top end, if you need to go there.

http://photonstophotos.net/Charts/PDR.htm#Nikon D500,Nikon D750

spraynpray

Most important is the amount of light you are shooting in. For instance if I am shooting a motor race on a dull day and want to boost ISO, I am going to see lower levels of noise than if I am shooting night/evening shots using the same ISO.

The thing is with processor updates is that they can never make a better image than the number of photons captured, they only 'smooth' the image better. I first started to notice that when I went from D7000 to D7100. The image from the D7100 appeared much better at first, but looking harder at it the image is 'smoothed' too much at the highest ISO's. When images from the two taken at ISO's they were both good at, there was little difference but the processing was much better at the higher ISO's on the newer camera. When I got the D750 as well, I saw that there was less smoothing in the image so it was a more 'honest' and accurate image than the DX.

Don't get me wrong, If had more money to throw at gear, I'd be all over the D500 for it's AF and better ISO performance for my macro shots, but at the moment, I'm happy to stick D750/7100.

WestEndFoto

Spraynpray, you touch on another point, though you may not agree with my choice of words. Shooting at higher resolution is fundamentally a trade off between resolution and noise. The resolution is reduced by the smoothing that you discuss. The choice about which trade off I select depends on the subject.

You also lose colour rendition and perhaps a few other parameters as well.

Ironheart

A picture is worth (at least) a thousand words

jeff5150

Ironheart said:

A picture is worth (at least) a thousand words

Thanks Ironheart...I'm going to do some test shots over the weekend and will post them up.

Sports

@jeff5150 - I just realized that maybe some of us have been fooled a bit by the impressive JPEGs sample photos and the JPEGs directly from the camera, because a fair amount of noise reduction has already been applied. Then, when we look at our own RAW files, we suddenly find lots of noise and graining. For my part, I just import RAWs into Lightroom, and I NEVER look at the JPEGs from the camera. It's pretty easy to never see those JPEGs. I must admit that my own view on high ISO image quality has probably suffered because of this, and I could have "wowed" myself more, had I looked at the JPEGs from my D500 and not just the RAW files.
In theory, we can get the same (and even better) results from RAW, using the right post processing software, but I assume it's high ISO image quality directly from the camera that is the issue that we're discussing.

Ironheart

@Sports couldn't agree more. Wait, I think you are agreeing with me

"It really depends on what you are shooting, how much light there is and how you post process" - Ironheart (2nd comment)

I was going to spill the beans, after we saw some images from @jeff5150 (are you a Van Halen fan, or just a psycho?) but I must confess used PRIME noise reduction software from DxO on the above image. However, the noise reduction that the free Capture NX-D does is on par with what the camera can do on its own, meaning not too shabby. I haven't been able to get as good results with LR, but maybe it's just me.

Pistnbroke

Quote ..the free Capture NX-D does is on par with what the camera can do on its own,
I did read some time ago that the in camera adjustments ie picture control and other menu settings were in fact a version of NX-D within the camera so your observations are not surprising.

Paperman

About 1.5 years ago or so, the D500 was about to come out and many here were expecting up to 2 stops better high ISO performance. I do remember posting ISO figures from DxO showing only incremental ( 1/10-1/5 stop ) improvements from previous models over the years. Some were under the impression DxO figures did not reflect reality and there was 1.5-2 stops performance difference between actual images from previous & next models easily visible on actual images.

Always trusted science/math/test figures more than my eye ....

WestEndFoto

It is easy to get rid of noise. But how much resolution are you willing to give up.

Ironheart

Pistnbroke said:

Quote ..the free Capture NX-D does is on par with what the camera can do on its own,
I did read some time ago that the in camera adjustments ie picture control and other menu settings were in fact a version of NX-D within the camera so your observations are not surprising.

The way I understand it, Nikon takes the same algorithms that the Expeed processor uses and implements them in NX-D. This means that if you use the same picture controls, etc... you get the same exact resulting JPG (or TIFF) from the RAW file whether it is processed by the camera or your computer. You can understand why this is important. My testing bears this out.

This is super important because Nikon engineers (software, firmware, and hardware) are the only ones that don't have to "reverse engineer" these algorithms. They already know what they are! This means that all of the hard work that goes into say, the de-mosaic process, noise reduction, color rendition, etc... inside the camera, can be easily replicated in software. EVERYONE else (Adobe, Apple, DxO, etc...) has to "reverse engineer" the RAW file, and apply their own algorithms in order to produce a JPG (or TIFF). Some folks like this, they think that say, LR produces a better look for their final images. No doubt, it provides additional controls over and above what NX-D can do.

Now comes the cool part, you can apply additional algorithms/settings to your RAWs on top of the normal picture controls, etc... that the camera does. Things like "Astro noise reduction" "axial color aberration" "PF flare reduction" etc... And all of these things are non-destructive.

You are not however limited to one tool. For images that I want really high-quality, but I want to use some of the additional power of LR or other post-processing, I will use NX-D to first produce either a) the best quality TIFF I can or b) the FLATTEST best quality TIFF I can (depends on what I'm trying to do) and then pull that into LR, or PS, or whatever to do additional post. For instance panoramic stitching, HDR, or other multi-image manipulations.
Anyway, long post, hope it helps.
Now let's get back to dissecting @jeff5150 's high-ISO images

Msmoto

After about 1800 clicks on my D500, my conclusion is the noise is about what one might expect from the extremely high pixel density. The D500 is like a 54 MP D8XX. So, on this image, full of shadows and highlights, ISO 6400, the noise is evident. But, as one who came from a film background there are two points I think justify the "cost" of a slightly noisy image. One is resolution, the other is dynamic range. I like my D500, but in low light, D4 is my choice, heavier, but better suited for very low light, IMO.

ISO 6400

https://www.flickr.com/photos/fantinesfotos/30772897511/sizes/o/

ISO 12,800

https://www.flickr.com/photos/fantinesfotos/30229243624/sizes/o/

Paperman

Well, those images are not exactly like the high ISO test shots posted by Nikon or some websites/magazines when a new camera is released, are they ?