Noise

[Bobobird]

Please take a look at this and then read the photographer's comment.

https://www.facebook.com/TinManPhoto...type=1&theater

600mm, f/4, 1/1000s, ISO 256,00, handheld

In one of the comments he say "no noise on the owl, but noisy on the background so I did denoising only on the background."

how can that be ???

[dje]

Please take a look at this and then read the photographer's comment.

https://www.facebook.com/TinManPhoto...type=1&theater

600mm, f/4, 1/1000s, ISO 256,00, handheld

In one of the comments he say "no noise on the owl, but noisy on the background so I did denoising only on the background."

how can that be ???

Hi Bobo

You'd really need to see the full size image to assess the noise. However I think the noise would be a lot more noticeable on the background compared to on the bird which has a lot of texture with light and dark tones all over the place.

Also, if he shot jpeg, there would probably be noise reduction done on the image in camera ?

Dave

[Bobobird]

Thanks Dave. Tin Man Lee, the photographer, is pretty well known and has won many awards with his wildlife photography. I think he shoots RAW.

I might go and just experiment for myself and see what this is all about.

[Manfred M]

Bobo - I suspect Dave has nailed it with his answer - noise is most noticeable in the darkest areas of the image and especially in the blue channel. Improper exposure will exacerbate the noise issue as well. Texture (like the feather details) will tend to hide it, whereas larger areas that are relatively evenly lit will tend to accentuate it. This suggests background noise reduction in more important that anything on the textured surface of the bird itself.

As Dave has stated, without a larger photo to evaluate the shot, we can only speculate, but

[Bobobird]

I am not able to find any larger.

Yes I agree about where noise will be more apparent. Was just wondering how he managed 256000 on that shot. I would have made everything so bad that the shot would not see the light of day. hehe

[pnodrog]

I am not able to find any larger.

Yes I agree about where noise will be more apparent. Was just wondering how he managed 256000 on that shot. I would have made everything so bad that the shot would not see the light of day. hehe

You maybe using a different priced camera.......

[xpatUSA]

I am not able to find any larger.

Yes I agree about where noise will be more apparent. Was just wondering how he managed 256000 on that shot. I would have made everything so bad that the shot would not see the light of day. hehe

The image has no EXIF data.

The file's JPEG meta-data is tagged as YCbCr4:2:0 (2 2). not unusual, but which means the color information was squeezed by half and then interpolated back up with reduced chroma noise when rendered on your screen.

The embedded ICC profile is named 'c2' and appears to be an Apple profile and some of the darker yellows are actually out of sRGB gamut - not that that is of any great significance.

At 720x388px, the image has likely been downsized a fair bit, which also reduces noise.

And then, the amount of NR used at these outrageous ISOs is often a wonder to behold - perfect for shooting pimply models

[JHzlwd]

First of all, the image doesn't look very sharp. That will soften noise all-round. The bird itself has a lot of texture further obscuring noise in that region. The photographer should, perhaps, have said "Noise not as visible in the bird" rather than "no noise on the owl". The latter phrase leaves an incorrect impression.

[Dave Humphries]

I bet the number of digits is correct, but the comma is in the wrong place.

He says "256,00", but must have meant 25,600, which is only two stops above 6,400 and capable with most recent DSLRs.

As to noise processing only the background - yes, that's a technique I have used before.

The reasons for noise being more noticeable in background (and less on subject with fine texture/detail) have already been stated, so it makes sense to not risk degrading the detail.

'Tin Man' being a professional, he's not going to give everyone access to a 'free' large image, hence the small size - which makes meaningful analysis impossible.

Cheers, Dave

[Bobobird]

Thanks guys - I am leaning with 25,600 rather than 256,000. For the latter ISO it would have been too dark to even see the owl let alone get an exposure on a moving subject.

@L.Paul - He uses the same camera that I use (though the 7D2 seems to be taking a beating at the moment).

[pnodrog]

Thanks guys - I am leaning with 25,600 rather than 256,000. For the latter ISO it would have been too dark to even see the owl let alone get an exposure on a moving subject.

@L.Paul - He uses the same camera that I use (though the 7D2 seems to be taking a beating at the moment).

Being a Nikon man I don't have to worry about noise......

[rpcrowe]

In one of the comments he say "no noise on the owl, but noisy on the background so I did denoising only on the background."

how can that be ???

I am interpreting the above statement as a question regarding how can you selectively reduce noise in certain areas of an image.

If you use NIK Dfine (the noise reduction portion of the NIK Software Suite) you can reduce noise globally. However, you can also select which areas in which to reduce the noise or not to reduce noise in other image areas by using the NIK Control Point system.

Additionally, you can select what type of noise luminance or color noise you will be reducing.

https://www.youtube.com/watch?v=SzIsMAk4Fqk

[Bobobird]

Thanks Richard. Good tip.

[Polar01]

Richard do not forget that with Nik Define 2 you can also apply with a brush instead of globally, however you can not do that if the image is a smart object.

Cheers: Allan

[chauncey]

Aah...why not just select the background and apply de=noise to it?