Pentax K-3 iii - Can it really be this noisy?
Posted 26/01/2024 - 19:03
Link
Like you, I'd expect better. And I got better yesterday when shooting the K3iii at ISO 3200, albeit different subject and background (sparrowhawk sitting on fence, OOF vegetation behind). Yours looks a bit underexposed to me, might that be contributing to the noise? I've always found that underexposed blue skies tend to be a bit noisy, going back to the K20D.
Steve
Steve
Posted 26/01/2024 - 20:03
Link
This was exposed with no compensation. Guess I should have gone with F5.6 instead of F8.
K3iii, K3ii, K-5, K-x, DA150-450mm, DA16-85WR, DA16-45, DA18-55WR, DA18-135WR, DA35 F2.4, M100mm F4 Macro, DA55-300mm, FA50mm 1.4, AF360 Flash, AF540 Flash
Posted 26/01/2024 - 20:39
Link
That looks cropped in tight..... can you show the image in its entirety?
The Legendary Terry Pratchett once said:
At the beginning there was nothing... which exploded
At the beginning there was nothing... which exploded
Posted 26/01/2024 - 21:36
Link
@Spad. Can I upload a full size image here? I am showing a crop to keep the size down. The rest of the image is all just sky. The one I have shown has been cropped to 1340px square.
K3iii, K3ii, K-5, K-x, DA150-450mm, DA16-85WR, DA16-45, DA18-55WR, DA18-135WR, DA35 F2.4, M100mm F4 Macro, DA55-300mm, FA50mm 1.4, AF360 Flash, AF540 Flash
Posted 26/01/2024 - 21:40
Link
tigershoot wrote:
@Spad. Can I upload a full size image here? I am showing a crop to keep the size down. The rest of the image is all just sky. The one I have shown has been cropped to 1340px square.
@Spad. Can I upload a full size image here? I am showing a crop to keep the size down. The rest of the image is all just sky. The one I have shown has been cropped to 1340px square.
Yes, you can but it needs to be in JPEG format.
The Legendary Terry Pratchett once said:
At the beginning there was nothing... which exploded
At the beginning there was nothing... which exploded
Posted 27/01/2024 - 00:03
Link
In case you haven't already seen it, here's a link to the 'other forum' where the K3 iii's noise is reviewed with examples link. As far as RAW files are concerned it concludes that "Up to ISO 12800, it is hard to notice any noise contribution". Perhaps their subject was more forgiving than a blue sky.
Just had a thought (well, it is late at night!). Since only 1 in 4 pixels on the sensor records blue, about 3 in 4 of the pixels won't be detecting much light when a blue sky is photographed. A precise rendering of each pixel in the RAW file will therefore necessarily appear rather blotchy, just how blotchy will depend on how closely the sky blue matches the blue of the sensor's filter. The RAW renderer must presumably do some averaging over adjacent pixels to address that sort of thing. Perhaps the sky picture is more a comment on what the RAW processing is doing rather than what the K3iii is doing.
I hope that sounds as sensible in the morning as it does now!
Steve
Just had a thought (well, it is late at night!). Since only 1 in 4 pixels on the sensor records blue, about 3 in 4 of the pixels won't be detecting much light when a blue sky is photographed. A precise rendering of each pixel in the RAW file will therefore necessarily appear rather blotchy, just how blotchy will depend on how closely the sky blue matches the blue of the sensor's filter. The RAW renderer must presumably do some averaging over adjacent pixels to address that sort of thing. Perhaps the sky picture is more a comment on what the RAW processing is doing rather than what the K3iii is doing.
I hope that sounds as sensible in the morning as it does now!
Steve
Posted 27/01/2024 - 12:41
Link
Tigershoot, can you upload the dng file to dropbox or the like. Then we can have a proper look at it.
Peter
My Flickr page
My Flickr page
Posted 27/01/2024 - 13:42
Link
I have just been outside to get a street shot. Taken on a tripod and I just kept upping the ISO. The difference between ISO100 and ISO1600 is huge. Check out the white plastic area of the garage. This is grabbed from my screen and the saved as a JPEG at the highest setting.
K3iii, K3ii, K-5, K-x, DA150-450mm, DA16-85WR, DA16-45, DA18-55WR, DA18-135WR, DA35 F2.4, M100mm F4 Macro, DA55-300mm, FA50mm 1.4, AF360 Flash, AF540 Flash
Posted 27/01/2024 - 15:58
Link
Thanks. I have the dng and will look at it as soon as i can
Peter
My Flickr page
My Flickr page
Posted 27/01/2024 - 16:16
Link
Any comments on the garage screenshot above? I cannot believe this level of noise, but I somehow cannot believe either that my K3iii is somehow noisier than others. It was made in Sept 2023 so is very new. Hopefully it's not a lemon.
In the garage shot, both shots were perfectly exposed. Just in case someone think the one shot was massively underexposed and then pushed in processing.
In the garage shot, both shots were perfectly exposed. Just in case someone think the one shot was massively underexposed and then pushed in processing.
K3iii, K3ii, K-5, K-x, DA150-450mm, DA16-85WR, DA16-45, DA18-55WR, DA18-135WR, DA35 F2.4, M100mm F4 Macro, DA55-300mm, FA50mm 1.4, AF360 Flash, AF540 Flash
Posted 27/01/2024 - 22:49
Link
This is a very interesting topic which I'm sure gets very technical very quickly.
Just a quick look at your full image it seems in DxO seems to show the noise is dominated by Luminance Noise not Chroma Noise. I guess that's not surprising as the blue sky is being chiefly captured by the 'blue' sensor pixels. Does Chroma noise become more visible when the colour being captured chiefly by two or three of the sensor's colour channels.
This whole topic is one for the experts to think and argue about And of course we have no idea what noise processing is performed by the camera even in producing a RAW data file let alone a compressed JPG file.
Makes my head hurt.
Mike
Just a quick look at your full image it seems in DxO seems to show the noise is dominated by Luminance Noise not Chroma Noise. I guess that's not surprising as the blue sky is being chiefly captured by the 'blue' sensor pixels. Does Chroma noise become more visible when the colour being captured chiefly by two or three of the sensor's colour channels.
This whole topic is one for the experts to think and argue about And of course we have no idea what noise processing is performed by the camera even in producing a RAW data file let alone a compressed JPG file.
Makes my head hurt.
Mike
Posted 28/01/2024 - 11:33
Link
Just curious as to what lenses you have used for the images?
The Legendary Terry Pratchett once said:
At the beginning there was nothing... which exploded
At the beginning there was nothing... which exploded
Posted 28/01/2024 - 14:50 - Helpful Comment
Link
Soo.... I took a couple of images of a white surface (white plastic window frame), one with K3iii the other with K20D (yeah, vintage!), both at ISO 3200, and compared DNGs with the DNG of the aircraft and blue sky (thanks, tigershoot ). The comparison was done with Affinity Photo v1 (I don't use LR) with noise reduction turned off. To sum up what I saw:
1. K3 iii noise was qualitatively much the same in my K3iii image and the aircraft shot.
2. K20D image noise was significantly worse than K3 iii - more and uglier (not unexpected).
I then tried to remove noise using the luminance slider. The result was that about 20% did a pretty good job of both K3iii images, less so for the K20D image. Playing around with colour sliders had very little effect.
Conclusion is that most of the noise is luminance and easy to remove.
There will, of course, always be some noise in what a sensor records because both light and electricity have random fluctuations ('shot noise' and thermal noise) which will remain however perfect the sensor. How much noise to remove in camera and how much to leave in the RAW file is a choice for the designers, particularly as noise removal results in some loss of detail in the image. Interestingly, Richoh/Pentax recognises this compromise in their K3 iii webpages where they say " the PENTAX K-3 Mark III greatly reduces noise, while still retaining the subtle outlines of a subject. This advanced technology optimizes the camera’s resolving power, and faithfully reproduces the subject's texture and detail. It also improves image description in the lower sensitivity range, while minimizing unfaithful reproduction of blacks in the higher sensitivity range where color reproduction becomes more difficult.". OK, there's some techno-babble from the marketing department in that, but it is still a statement of the priorities they took when making the trade-off between noise and detail.
Personally, I think that Ricoh/Pentax has done a pretty good job with the K3iii and the noise/detail compromise seems in line with what they say on their website. What noise there is seems easy to remove (for the types of photos I take) and I usually don't bother because to my eyes it's usually acceptably unobtrusive (again, for the types of photos I take). I'd actually be very suspicious (and annoyed ) if RAW files were all squeaky clean as that would imply fairly heavy noise reduction before I could get my hands on the files and that, to my mind, defeats the purpose of RAW. Sure, things might be even better with bigger pixels on a bigger sensor, but I wouldn't be surprised if the K3 iii is about as good as it gets for this number of pixels on an APS-C sensor with today's technology.
Steve
1. K3 iii noise was qualitatively much the same in my K3iii image and the aircraft shot.
2. K20D image noise was significantly worse than K3 iii - more and uglier (not unexpected).
I then tried to remove noise using the luminance slider. The result was that about 20% did a pretty good job of both K3iii images, less so for the K20D image. Playing around with colour sliders had very little effect.
Conclusion is that most of the noise is luminance and easy to remove.
There will, of course, always be some noise in what a sensor records because both light and electricity have random fluctuations ('shot noise' and thermal noise) which will remain however perfect the sensor. How much noise to remove in camera and how much to leave in the RAW file is a choice for the designers, particularly as noise removal results in some loss of detail in the image. Interestingly, Richoh/Pentax recognises this compromise in their K3 iii webpages where they say " the PENTAX K-3 Mark III greatly reduces noise, while still retaining the subtle outlines of a subject. This advanced technology optimizes the camera’s resolving power, and faithfully reproduces the subject's texture and detail. It also improves image description in the lower sensitivity range, while minimizing unfaithful reproduction of blacks in the higher sensitivity range where color reproduction becomes more difficult.". OK, there's some techno-babble from the marketing department in that, but it is still a statement of the priorities they took when making the trade-off between noise and detail.
Personally, I think that Ricoh/Pentax has done a pretty good job with the K3iii and the noise/detail compromise seems in line with what they say on their website. What noise there is seems easy to remove (for the types of photos I take) and I usually don't bother because to my eyes it's usually acceptably unobtrusive (again, for the types of photos I take). I'd actually be very suspicious (and annoyed ) if RAW files were all squeaky clean as that would imply fairly heavy noise reduction before I could get my hands on the files and that, to my mind, defeats the purpose of RAW. Sure, things might be even better with bigger pixels on a bigger sensor, but I wouldn't be surprised if the K3 iii is about as good as it gets for this number of pixels on an APS-C sensor with today's technology.
Steve
Add Comment
To leave a comment - Log in to Pentax User or create a new account.
383 posts
15 years
Ashford,
Kent
I have been very busy since getting my new K3iii, and what with the grim weather have not really used it much. I have taken some shots at higher ISOs and have to say I am dismayed at the noise. Please have a look at this photo. Shot as DNG at only 3200 ISO and imported into LR. No noise reduction, no level adjustments and sharpening set to just 40. I was expecting amazing low noise performance from this camera, and so far I am not impressed. Am I doing something wrong?