[Fauxton]

Another point. There is a scientific paper that argues that, in an ideal imaging system, 1000 is the minimum number of photons per sensel required to get an image without shot noise becoming visible:

Psychophysical thresholds and digital camera sensitivity: the thousand photon limit

Quoting: In summary, an ideal single capture camera must be designed to capture 10^3 photons in the dark part of an image to avoid visible photon noise. The pixel must be able to capture 10^6 photons to encode the dynamic range of natural images. These are the basic constraints for an ideal camera that can render the vast majority of natural images with no visible noise and no saturated pixels. For real cameras, the requirement can be even higher due to the addition of electronic noise and color process functions (such as color correction).

So if the sensel in the K-3 has a 30Ke- FWC (I haven't seen the actual value reported by anyone yet, but I suspect it's around 30Ke-), then 1000e-, which would needed for a 33:1 (30dB) SNR, would be 2 stops below ISO12800 i.e. at ISO3200.

BTW, my K-5 calculations indicate that its sensels have a 34:1 (30.5dB) SNR at ISO3200.

Of course, real-life sensors also have read noise, made up of sensor read noise, PGA noise and ADC noise, and images will start looking noisy at lower ISOs.

Dan.

HI DOSDAN
I calculated that the sat point of the K3 at base ISO - 94 (100) the read noise was 3.33, Saturation 24500 and DR 12.84
See K3 Read Noise, Sat and DR: Photographic Science and Technology Forum: Digital Photography Review for a bit more. QE is fairly high at 48%
The K5 is quite a bit better in its saturation calculated at 47159 from Sensorgen. I used the same method as Sensorgen.
The tradeoff in pixel noise levels on the K3 vs K5 is that we have many more pixels on the K3. So when sized and viewed similarly the apparent noise is similar.
I have the K3 and K5. In most situations its very hard to tell which photo the K3 versus the K5 took. However the resolution difference is quite noticeable.
I find that under 3200 the K3 has better IQ. Above that its about the same due to noise and what I need to do to mitigate it - I lose most of the improvement in resolution to mitigate the noise. Below 1600 and in particular below 800 the I find the K3 is quite a bit better in many ways.
Although I shoot raw mostly I have been quite happy with the default tuning of the JPEGs. Now that I've spent some time tuning the parameters I like them better than my Olympus. And that is saying something.
Jim

[Stagnant]

The Exif on Flickr says all pics are taken @ ISO12800 which is more like I would expect when I look at them. I wonder what Stagnant is trying to pull...

That's due to a method I used for creating the crops in the photoshop. Each next image was layered on top of the 12800 file and then all the layers were cropped simultaneously. Then layers were exported to files. Download the original fullzise images from a link in the first post and you can compare the crops against them and examine exifs.
Link : https://drive.google.com/file/d/0B8INe0_VKyQ8Z3ppNXhsaWpGVjQ/edit?usp=sharing

That's due to a method I used for creating the crops in the photoshop. Each next image was layered on top of the 12800 file and then all the layers were cropped simultaneously. Then layers were exported to files. Download the original fullzise images from a link in the first post and you can compare the crops against them and examine exifs.
Link : https://drive.google.com/file/d/0B8INe0_VKyQ8Z3ppNXhsaWpGVjQ/edit?usp=sharing

Ok, looked kind of suspicious. What I don't understand is where all the colour noise is coming from. I have no colour noise in my night pics. I have noise reduction turned off, shooting in RAW+ L***. Luminance noise, yes but not colour noise.

[Rondec]

Rondec, would you mind, uploading the original image size, or at least a 100% crop ?

I probably can upload a one hundred percent crop later, but of course, it will show a lot more noise...

[Stagnant]

What I don't understand is where all the colour noise is coming from. I have no colour noise in my night pics. I have noise reduction turned off, shooting in RAW+ L***. Luminance noise, yes but not colour noise.

This is perhaps due to the several facts :
* normally you would not use F9 in such condtions.
* normally you would not turn EVERYTHING off, when doing post processing. For example in DxO the preset "default" already shapes you image in a certain way. The same goes for SilkyPix and to a lesser extent it applies to Lightroom. PDCU is a thing of its own but it still acts in a similar fashion.

[gosman]

It is probably the photographer and the conditions at the time of shooting. Lighting conditions vary not the sensors! Technology is consistent not the photographers and environment!

[RonHendriks1966]

No idea about those numbers, except that I know that with hi iso the K-5 is better then the K-3. Up to iso3200 both deliver great images at times. With the K-3 I didn't had a slammer at iso3200 yet, until last week.

Put it here:

Soccer Dance

[JinDesu]

Ok, looked kind of suspicious. What I don't understand is where all the colour noise is coming from. I have no colour noise in my night pics. I have noise reduction turned off, shooting in RAW+ L***. Luminance noise, yes but not colour noise.

One thing I noticed when processing k-3 files on my computer (sample RAWs) was that my Lightroom 5 showed more chroma noise than my Lightroom 3, and both showed more chroma noise than my Photoshop CS5.

It may be how the programs read the files.

[Stagnant]

- Processed using default settings in DxO.
- Processed using default settings in SilkyPix.

Default settings in these programmes imply noise reduction, sharpening, contrast, etc. Click the preview images to get the original size from flickr. You may notice, that colour noise is gone and that the images look much smoother now.

[j2photos]

- Processed using default settings in DxO.
- Processed using default settings in SilkyPix.

Default settings in these programmes imply noise reduction, sharpening, contrast, etc. Click the preview images to get the original size from flickr. You may notice, that colour noise is gone and that the images look much smoother now.

It looks like silkypix really messed with the WB. In my experience the K-3 has been spot on with the in body AWB. The one from Dxo doesnt look bad at all but I still like Lightroom.

[Stagnant]

In my experience certain software has advantage when it comes to certain scenes.

[dosdan]

Stagnant, are you using SP Developer Studio Pro 5 with the Sharpness setting set to "No sharpness"?

Did you have the K-3's WB set to Multi AWB?

If you want to see the raw file converted without any undocumented "optimisations" being applied, use the open-source RawTherapee.

Dan.

[Stagnant]

@dosdan, in all the samples, except for the last one (default settings), I manually set all the sharpness and noise adjustments to 0 in silkypix. White balance was set to AWB, because initially I intended to do a comparison with K-01, which does not have M AWB. Unfortunately the shooting conditions changed considerably between the two sessions, so this ended up being just K-3 at high iso.
I used to have RAWTherapee on my PC, but I haven't used it, so it got removed.

[rawr]

I know these shots are just a test, but no-one would normally consider using ISO 12800 for such a night scene 'landscape' shot, especially if they were interested in detail preservation, and had access to a tripod.

IS0 200 or 400 + f5.6 + a sturdy tripod would probably be how one would normally shoot such a scene.

ISO 12800 for indoors low-light handheld, gigs in dark bars, low-light sports with telephoto, maybe OK. ISO 12800 for static night landscapes (or astrophotography), probably never OK.

[JinDesu]

In consideration, that is pretty true - when you are subjecting the sensor to small points of lights/details in a very dark scene, it hurts more at high ISO than other subjects.