RoamingScott wrote:
I can see the banding, though the color noise portion has been dealt with 🤷🏼♂️
Can you please perhaps show a crop of where you see banding - im probably not as observant as you are as i don't notice any banding in that image. I definitely see noise but no banding.
bernardl wrote:
Still interested in you pointing out where exactly you see banding in the violin player image.
Cheers,
Bernard
It's literally everywhere in the background, but you weren't able to see the LoCA on your 35/1.2 so...not sure this is a discussion worth going down for either of us.
I'll start and finish here: take a screenshot of the full Flickr image, throw it in photoshop, and raise the brightness if the horizontal PDAF artifacts aren't visible to you at the regular brightness. With some extra brightness the lines should jump out at you. It's extremely obvious to me at the regular brightness, but I've seen it before in my own work, tested lots of different noise abatement techniques, and know what to look for.
This is the reason I don't care about the charts that Geoff posted. You run into levels of funk that charts don't account for as you get into this range of ISOs.
RoamingScott wrote:
It's literally everywhere in the background, but you weren't able to see the LoCA on your 35/1.2 so...not sure this is a discussion worth going down for either of us.
I'll start and finish here: take a screenshot of the full Flickr image, throw it in photoshop, and raise the brightness if the horizontal PDAF artifacts aren't visible to you at the regular brightness. With some extra brightness the lines should jump out at you. It's extremely obvious to me at the regular brightness, but I've seen it before in my own work, tested lots of different noise abatement techniques, and know what to look for.
This is the reason I don't care about the charts that Geoff posted. You run into levels of funk that charts don't account for as you get into this range of ISOs....Show more →
Even after increasing the brightness a lot in photoshop, I don't see any horizontal banding anywhere in the image. Not sure what you are seeing but whatever it is, it is definitely not obvious.
curious80 wrote:
Even after increasing the brightness a lot in photoshop, I don't see any horizontal banding anywhere in the image. Not sure what you are seeing but whatever it is, it is definitely not obvious.
Everyone has their own standards of quality and IQ nuance. God has blessed you mightily if you can't see it
I've said my piece on the matter and my preferences. Y'all go shoot till your trigger finger falls off.
RoamingScott wrote:
Everyone has their own standards of quality and IQ nuance. God has blessed you mightily if you can't see it
I am glad to be blessed! Looks like you are able to see things that no one else is able to see - that can be a superpower or a curse depending on your viewpoint
RoamingScott wrote:
It's literally everywhere in the background, but you weren't able to see the LoCA on your 35/1.2 so...not sure this is a discussion worth going down for either of us.
I'll start and finish here: take a screenshot of the full Flickr image, throw it in photoshop, and raise the brightness if the horizontal PDAF artifacts aren't visible to you at the regular brightness. With some extra brightness the lines should jump out at you. It's extremely obvious to me at the regular brightness, but I've seen it before in my own work, tested lots of different noise abatement techniques, and know what to look for.
This is the reason I don't care about the charts that Geoff posted. You run into levels of funk that charts don't account for as you get into this range of ISOs....Show more →
We are talking about a screen shot of a flicker 1024 pix resized file, are we?
There is a faint horizontal patterns in some parts of the image on the left if you brighten significantly yes, very hard to see at normal brightness levels IMHO. I am pretty sure it would be totally invisible in a DxO conversion of this image. I am more than willing to so the test if the original raw file is provided to me.
I just checked in C1 Pro the ISO 25,600 images I shot last weekend at default denoise settings with the 400mm f2.8 S TC at 560mm (meaning basically no denounce) and I cannot see any banding. These were with the Z8 but the Z9 is the same. Here are C1 Pro and DXO conversions (.nef -> DNG -> DxO -> C1 Pro) for your review (NR applied). Not the best photograph obviously, but this is just for demonstration purposes. The full size jpg are available for download on Flickr.
So I am sorry, my opinion remains that correctly exposed ISO 25,600 images from Z8/Z9 are totally usable if handled through the correct workflow.
As far as the 35mm f1.2 S I commented about fringing not about LoCA and there was none in the file I shot, that's all In said. I could see some fringing over high contrast branches in some samples.
bernardl wrote:
There is a faint horizontal patterns in some parts of the image on the left if you brighten significantly yes, very hard to see at normal brightness levels IMHO. I am pretty sure it would be totally invisible in a DxO conversion of this image. I am more than willing to so the test if the original raw file is provided to me.
If you want to work with the RAW file, you can download it here:
I'm interested to see what DxO can do with the noise. I'll try downloading the trial version of DxO when I am in the office tomorrow (my laptop at home isn't sufficiently spec'ed). I settled on Lightroom since I already use the full Adobe CC suite, but I'm willing to try other software if it provides a real benefit...
I was out today and arrived at my location shortly before sunrise. With this particular thread in mind, I decided to begin my shoot at ISO 25000 +/- and drop the ISO when the conditions permitted.
Attached are three images... All were brought into LR Classic, Processed with DXO RAW, and then masked in LR for the fox and the background. Small adjustments on color and white balance were made and I exported them as jpgs w/ "standard sharpening for web."
I'd have no issues printing these at 16 x24", but probably would not go with a 30" on a glossy surface... but let's face it, who would buy a 30" print of a fox
Images like this are things that I'd happily use for promotions, blogs, or photo essays in a magazine.
regards,
bruce
NIKON Z 9NIKKOR Z 400mm f/2.8 TC VR S lens560mmf/4.01/1000s25600 ISO+1.3 EV
NIKON Z 9NIKKOR Z 400mm f/2.8 TC VR S lens400mmf/2.81/1000s12800 ISO+1.3 EV
NIKON Z 9NIKKOR Z 400mm f/2.8 TC VR S lens400mmf/2.81/1000s6400 ISO0.0 EV
bernardl wrote:
Lovely Bruce! Wonderful stuff and a beautiful demonstration about the high ISO abilities of the Z9 with DxO. No banding here either.
Since those were quickly processes, if I may I think I would prefer the first one with slightly darker shadows. But probably a matter of taste. :-)
Cheers,
Bernard
Thanks for the comment Bernard. It's funny, but I would like the top one to have a darker background as well, but you are seeing the impact of using an ultra high ISO. I biased my exposure to the highlights because I knew that the dynamic range would be reduced at ISO 25000. As such, the fox is perfectly exposed, but the background has doesn't look as good as it does in other shots. I tried to actually bring down the background's exposure, but the blacks started to look bad. In the end, I left the background as it was.
In contrast, I have little difficulty controlling, pushing, or pulling highlights and shadows when I shoot at ISO 6400 or slower.
I'm interested to see what DxO can do with the noise. I'll try downloading the trial version of DxO when I am in the office tomorrow (my laptop at home isn't sufficiently spec'ed). I settled on Lightroom since I already use the full Adobe CC suite, but I'm willing to try other software if it provides a real benefit...
PL8 Denoise PRIME XD/XD2s
Reduced luminance NR to 30 from default of 40
+2.4ev
Looks pretty sweet to me. But hey I shoot Sony so my standards are already so low
OwlsEyes wrote:
I was out today and arrived at my location shortly before sunrise. With this particular thread in mind, I decided to begin my shoot at ISO 25000 +/- and drop the ISO when the conditions permitted.
Attached are three images... All were brought into LR Classic, Processed with DXO RAW, and then masked in LR for the fox and the background. Small adjustments on color and white balance were made and I exported them as jpgs w/ "standard sharpening for web."
I'd have no issues printing these at 16 x24", but probably would not go with a 30" on a glossy surface... but let's face it, who would buy a 30" print of a fox
Images like this are things that I'd happily use for promotions, blogs, or photo essays in a magazine.
I really like the middle picture, it is beautifully showing the fox strotting about, nothing wrong with the image although the shallower dynamic range and thinner colour is typical at this ISO. A D6 or D5 might give slightly more dense colour in the images but obviously it's not going to be like mid or low ISO images on those cameras either.
I have no doubt that there would be someone who would want that image in a large print. There are a lot of people who love mammals and think foxes are among the cutest things, and this image looks natural and beautiful.
First, a concert situation indoors is different from natural light outdoors. Dim natural light outdoors typically has a lot of blue light and this is to the advantage of the photographer as the photosensor is natively more sensitive to red than shorter wavelengths and this means the blue channel is the noisiest. Since the early morning or late evening light has relatively more blue light in it, this helps improve the SNR of the pixels that are filtered to respond to blue light, and so even at high ISOs the images can be quite usable.
In indoor artificial light the colours are a lot warmer and this can result in the blue channel breaking up earlier than the others when the light is overall very dim and high ISO is used. This was worse in the D800 and D810 than the D850 and Z9/8 as Nikon made gradual improvements to how the high ISO images are recorded and processed. The D800 could break down even as early as ISO 6400 in candle light while cameras like D3, D3s, D5, D6 could do very nicely in similar conditions. The 45 MP sensors do fairly well at ISO 6400-12800 though not as well as the D6/D5 in terms of tonality, dynamic range, and colours.
As for the banding discussion, noise-reduction algorithms can sometimes "find" patterns in the data when they are trying to preserve image details while suppressing noise, even when there is no pattern that could be seen by naked eye in the original image. Thus one may need to try out different algorithms and settings to find the one that produces the best result in a given situation. I have not run into banding with my Z8 but I generally don't go above ISO 12800 and I find that the Zf and D6 work better at ultra high ISO and so I preferentially use those cameras in such conditions.
ilkka_nissila wrote:
I really like the middle picture, it is beautifully showing the fox strotting about, nothing wrong with the image although the shallower dynamic range and thinner colour is typical at this ISO. A D6 or D5 might give slightly more dense colour in the images but obviously it's not going to be like mid or low ISO images on those cameras either.
I have no doubt that there would be someone who would want that image in a large print. There are a lot of people who love mammals and think foxes are among the cutest things, and this image looks natural and beautiful. ...Show more →
Thank you for the kind words...
Your assessment of "ultra high" ISO images is spot on. Rather than seeing banding or blotchy noise after post-processing, I often feel as if the color depth suffers. As a result, I would need to punch the saturation to restore some of this. Of course, there is a cost to doing so, as sharpness is often reduced. Another cost is a "digital" effect where the image almost seems plastic...
To be honest, that 2nd shot is pretty close to how the scene appeared when it was taken. The fox was running across an open field as the sun was rising into an overcast sky. There was a hint of backlight due to light reflecting off the grasses and ground.... I was pretty pleased given that the ISO was 12500.
ilkka_nissila wrote:
A few comments on the discussion overall.
First, a concert situation indoors is different from natural light outdoors. Dim natural light outdoors typically has a lot of blue light and this is to the advantage of the photographer as the photosensor is natively more sensitive to red than shorter wavelengths and this means the blue channel is the noisiest. Since the early morning or late evening light has relatively more blue light in it, this helps improve the SNR of the pixels that are filtered to respond to blue light, and so even at high ISOs the images can be quite usable.
In indoor artificial light the colours are a lot warmer and this can result in the blue channel breaking up earlier than the others when the light is overall very dim and high ISO is used. This was worse in the D800 and D810 than the D850 and Z9/8 as Nikon made gradual improvements to how the high ISO images are recorded and processed. The D800 could break down even as early as ISO 6400 in candle light while cameras like D3, D3s, D5, D6 could do very nicely in similar conditions. The 45 MP sensors do fairly well at ISO 6400-12800 though not as well as the D6/D5 in terms of tonality, dynamic range, and colours.
As for the banding discussion, noise-reduction algorithms can sometimes "find" patterns in the data when they are trying to preserve image details while suppressing noise, even when there is no pattern that could be seen by naked eye in the original image. Thus one may need to try out different algorithms and settings to find the one that produces the best result in a given situation. I have not run into banding with my Z8 but I generally don't go above ISO 12800 and I find that the Zf and D6 work better at ultra high ISO and so I preferentially use those cameras in such conditions. ...Show more →
In addition to the red vs blue light, the other factor is that dim natural light outdoors is also generally very diffused. So these are not high contrast scenes with very dark shadows. Indoor low-light scenarios often have areas with deep shadows without much light and that is where the biggest amount of noise lurks. Deep shadow part of an ISO12800 shot would have more noise than mid-gray part of an ISO 25600 shot.
I dont see why 12800 is a problem, really, on a ff camera made in the last 10+ years. I haven't found 6400 to be a problem most of the time on aps, and ff >1 stop better
You need good high iso pp though. I didn't just immediately run it through and get good results. Took some time to figure it out, for me , software varies too
I didn't find a bit of underexposure to be much problem
This is ISO 25600, yes it did loose some detail but I'll take it! Amazing to be able to show an image at that high an ISO, yrs ago it would have looked like $hit to put it bluntly!! Yes some topaz was used!!
NIKON Z 9NIKKOR Z 800mm f/6.3 VR S lens800mmf/7.11/4000s25600 ISO-0.3 EV
