fredmiranda.com Login
This topic is in the 2017 archive. You can read it and like its posts here. It is read only.

Archive 2017D850: Bill Claff's data shows significant High ISO DR improvement

  
 
snapsy
Offline

p.1 #1D850: Bill Claff's data shows significant High ISO DR improvement
Bill added more preliminary D850 today, this time read noise for much of the ISO range. Read noise corresponds to engineering DR, ie way DxO represents their DR measurements. The values indicate a nearly 1EV improvement in High ISO DR start at ISO 400, which matches Nikon's claim for the D850's 1-stop improvement in High ISO performance (although this is performance in the deeper tones/shadows vs an overall noise reduction that would be more useful).

D810 vs D850 Read Noise
Aug 27, 2017 at 03:34 PM
TAM63
Offline
Details
p.1 #2D850: Bill Claff's data shows significant High ISO DR improvement
Over on DPreview, he said to me "The shape and amount of per-pixel noise is like the D500 as opposed to the D810."

I sure hope that does not turn out to be the case, or it doesn't mean what I think it means.

On the D810, I almost never have to use luminance noise reduction (which reduces detail) - I can almost always get away with just color noise reduction.

The D500, noise seems to appear at very low ISOs, and is harder to get rid of.
Aug 27, 2017 at 03:45 PM
snapsy
Offline
Details
p.1 #3D850: Bill Claff's data shows significant High ISO DR improvement
TAM63 wrote:
Over on DPreview, he said to me "The shape and amount of per-pixel noise is like the D500 as opposed to the D810."

I sure hope that does not turn out to be the case, or it doesn't mean what I think it means.

On the D810, I almost never have to use luminance noise reduction (which reduces detail) - I can almost always get away with just color noise reduction.

The D500, noise seems to appear at very low ISOs, and is harder to get rid of.


He's probably referring to the dual ISO gain of the D500 (the knee you see at ISO 400 on the D850 graph), which matches current state-of-the-art low and high ISO DR balancing, like the A7rii/A7s sensors. It's a good thing The experience you had with the D500 is likely from its higher overall noise as a function of it being a DX sensor.
Aug 27, 2017 at 03:50 PM
TAM63
Offline
Details
p.1 #4D850: Bill Claff's data shows significant High ISO DR improvement
snapsy wrote:
TAM63 wroteOver on DPreview, he said to me "The shape and amount of per-pixel noise is like the D500 as opposed to the D810."

I sure hope that does not turn out to be the case, or it doesn't mean what I think it means.

On the D810, I almost never have to use luminance noise reduction (which reduces detail) - I can almost always get away with just color noise reduction.

The D500, noise seems to appear at very low ISOs, and is harder to get rid of.
He's probably referring to the dual ISO gain of the D500 (the knee you see at ISO 400 on the D850 graph), which matches current state-of-the-art low and high ISO DR balancing, like the A7rii/A7s sensors. It's a good thing The experience you had with the D500 is likely from its higher overall noise as a function of it being a DX sensor.


Thank you. I think I'm just going to decide this technical stuff is going right over my head
Aug 27, 2017 at 03:58 PM
arbitrage
Offline
Details
p.1 #5D850: Bill Claff's data shows significant High ISO DR improvement
I would guess that if you crop into D500 DX size the D850 may look very similar. Just like when I owned 5DSR and 7D2. Cropping the 5DSR to use it for reach down into 7D2 size or further looked just like a 7D2 file (i.e. not all that great). So basically you get a D500 (-1FPS and +$$) when you need reach and a nice FX sensor when you can use the full sensor or at least more than the DX area. But viewing the files at 1:1 in LR may look the same.
Aug 27, 2017 at 04:22 PM
Fred Miranda
Offline
Admin
Upload & Sell: On
Details
p.1 #6D850: Bill Claff's data shows significant High ISO DR improvement
snapsy wrote:
Bill added more preliminary D850 today, this time read noise for much of the ISO range. Read noise corresponds to engineering DR, ie way DxO represents their DR measurements. The values indicate a nearly 1EV improvement in High ISO DR start at ISO 400, which matches Nikon's claim for the D850's 1-stop improvement in High ISO performance (although this is performance in the deeper tones/shadows vs an overall noise reduction that would be more useful).

D810 vs D850 Read Noise


One stop improvement is a lot!
Much cleaner files from ISO 400 (dual ISO gain starts) until about 3200. It's almost the difference between going from APS-C to Full frame, except that both cameras are full frame!
Aug 27, 2017 at 04:42 PM
arbitrage
Offline
Details
p.1 #7D850: Bill Claff's data shows significant High ISO DR improvement
Except for the disclaimer below the graph that says this data is "not appropriate for comparing camera models".

But if we ignore that important disclaimer than we can see the D850 is about identical to 5D4 and still at least a stop worse than 1DX2 and D5 (more than a stop for the D5).
Aug 27, 2017 at 04:47 PM
TAM63
Offline
Details
p.1 #8D850: Bill Claff's data shows significant High ISO DR improvement
arbitrage wrote:
I would guess that if you crop into D500 DX size the D850 may look very similar. Just like when I owned 5DSR and 7D2. Cropping the 5DSR to use it for reach down into 7D2 size or further looked just like a 7D2 file (i.e. not all that great). So basically you get a D500 (-1FPS and +$$) when you need reach and a nice FX sensor when you can use the full sensor or at least more than the DX area. But viewing the files at 1:1 in LR may look the same.


Well if a cropped D850 shot looks like the D500, I am going to be heartbroken. Because I really disliked the D500 - both the noise, and just some harsh characteristic to the sensor. I sure hope you are not right on this, but I'm afraid you might be.

I can crop the heck out of the D810 and it still looks pretty good.
Aug 27, 2017 at 04:57 PM
protikh
Offline
Details
p.1 #9D850: Bill Claff's data shows significant High ISO DR improvement
TAM63 wrote:
arbitrage wroteI would guess that if you crop into D500 DX size the D850 may look very similar. Just like when I owned 5DSR and 7D2. Cropping the 5DSR to use it for reach down into 7D2 size or further looked just like a 7D2 file (i.e. not all that great). So basically you get a D500 (-1FPS and +$$) when you need reach and a nice FX sensor when you can use the full sensor or at least more than the DX area. But viewing the files at 1:1 in LR may look the same.
Well if a cropped D850 shot looks like the D500, I am going to be heartbroken. Because I really disliked the D500 - both the noise, and just some harsh characteristic to the sensor. I sure hope you are not right on this, but I'm afraid you might be.

I can crop the heck out of the D810 and it still looks pretty good.


When examining the ISO6400 and ISO12800 bird pictures here http://www.photographyblog.com/previews/nikon_d850_photos/ , the quality looks better than D500 and D810 to me. Also files responded very nicely to noise reduction.
But of course these are relatively low shutter speed photos and high shutter speed BIF picture might be different.
Aug 27, 2017 at 05:03 PM
TAM63
Offline
Details
p.1 #10D850: Bill Claff's data shows significant High ISO DR improvement
protikh wrote:
TAM63 wrote
arbitrage wroteI would guess that if you crop into D500 DX size the D850 may look very similar. Just like when I owned 5DSR and 7D2. Cropping the 5DSR to use it for reach down into 7D2 size or further looked just like a 7D2 file (i.e. not all that great). So basically you get a D500 (-1FPS and +$$) when you need reach and a nice FX sensor when you can use the full sensor or at least more than the DX area. But viewing the files at 1:1 in LR may look the same.
Well if a cropped D850 shot looks like the D500, I am going to be heartbroken. Because I really disliked the D500 - both the noise, and just some harsh characteristic to the sensor. I sure hope you are not right on this, but I'm afraid you might be.

I can crop the heck out of the D810 and it still looks pretty good.
When examining the ISO6400 and ISO12800 bird pictures here http://www.photographyblog.com/previews/nikon_d850_photos/ , the quality looks better than D500 and D810 to me. Also files responded very nicely to noise reduction.
But of course these are relatively low shutter speed photos and high shutter speed BIF picture might be different.


Thank you. Me personally, I'm not as concerned with what the high ISO shots look like, although that is a bonus. Low to middle ISO is what I really want to shoot.

For me (and others I know) I found the D500 got some sort of noise at quite low ISOs. And it was different noise from the D810. I was just not a fan of that sensor (I sold mine). If I'm going to end up with essentially the same photos if I crop the D850, I probably would pass.

Aug 27, 2017 at 05:07 PM
arbitrage
Offline
Details
p.1 #11D850: Bill Claff's data shows significant High ISO DR improvement
TAM63 wrote:
protikh wrote
TAM63 wrote
arbitrage wroteI would guess that if you crop into D500 DX size the D850 may look very similar. Just like when I owned 5DSR and 7D2. Cropping the 5DSR to use it for reach down into 7D2 size or further looked just like a 7D2 file (i.e. not all that great). So basically you get a D500 (-1FPS and +$$) when you need reach and a nice FX sensor when you can use the full sensor or at least more than the DX area. But viewing the files at 1:1 in LR may look the same.
Well if a cropped D850 shot looks like the D500, I am going to be heartbroken. Because I really disliked the D500 - both the noise, and just some harsh characteristic to the sensor. I sure hope you are not right on this, but I'm afraid you might be.

I can crop the heck out of the D810 and it still looks pretty good.
When examining the ISO6400 and ISO12800 bird pictures here http://www.photographyblog.com/previews/nikon_d850_photos/ , the quality looks better than D500 and D810 to me. Also files responded very nicely to noise reduction.
But of course these are relatively low shutter speed photos and high shutter speed BIF picture might be different.
Thank you. Me personally, I'm not as concerned with what the high ISO shots look like, although that is a bonus. Low to middle ISO is what I really want to shoot.

For me (and others I know) I found the D500 got some sort of noise at quite low ISOs. And it was different noise from the D810. I was just not a fan of that sensor (I sold mine). If I'm going to end up with essentially the same photos if I crop the D850, I probably would pass.



The D810 has bigger pixels so sometimes that will allow it to hold up to bigger crops. Certainly I seem to be able to get as good of quality cropping my 5D4 30MP files vs what I got with 5DSR 50MP files once you get into ISO 800 and up. Even the 20MP 1DX2 and 18MP 1DX files hold up better with bigger crops once you are getting up in the ISO range. There is always a balance and if one is using all the MPs to crop a lot you don't really gain IQ over a DX sensor with similar pixel density if the sensors are of similar generation and construction. Now the D850 has this BSI where the D500 doesn't but so far I haven't read anything that says it will make a big difference.

I think we will need to wait for a couple more weeks until sites start posting full RAW comparisons like DPReview so you can download RAWs and compare to D500 or whatever other camera.
Aug 27, 2017 at 05:18 PM
TAM63
Offline
Details
p.1 #12D850: Bill Claff's data shows significant High ISO DR improvement
arbitrage wrote:
TAM63 wrote
protikh wrote
TAM63 wrote
arbitrage wroteI would guess that if you crop into D500 DX size the D850 may look very similar. Just like when I owned 5DSR and 7D2. Cropping the 5DSR to use it for reach down into 7D2 size or further looked just like a 7D2 file (i.e. not all that great). So basically you get a D500 (-1FPS and +$$) when you need reach and a nice FX sensor when you can use the full sensor or at least more than the DX area. But viewing the files at 1:1 in LR may look the same.
Well if a cropped D850 shot looks like the D500, I am going to be heartbroken. Because I really disliked the D500 - both the noise, and just some harsh characteristic to the sensor. I sure hope you are not right on this, but I'm afraid you might be.

I can crop the heck out of the D810 and it still looks pretty good.
When examining the ISO6400 and ISO12800 bird pictures here http://www.photographyblog.com/previews/nikon_d850_photos/ , the quality looks better than D500 and D810 to me. Also files responded very nicely to noise reduction.
But of course these are relatively low shutter speed photos and high shutter speed BIF picture might be different.
Thank you. Me personally, I'm not as concerned with what the high ISO shots look like, although that is a bonus. Low to middle ISO is what I really want to shoot.

For me (and others I know) I found the D500 got some sort of noise at quite low ISOs. And it was different noise from the D810. I was just not a fan of that sensor (I sold mine). If I'm going to end up with essentially the same photos if I crop the D850, I probably would pass.
The D810 has bigger pixels so sometimes that will allow it to hold up to bigger crops. Certainly I seem to be able to get as good of quality cropping my 5D4 30MP files vs what I got with 5DSR 50MP files once you get into ISO 800 and up. Even the 20MP 1DX2 and 18MP 1DX files hold up better with bigger crops once you are getting up in the ISO range. There is always a balance and if one is using all the MPs to crop a lot you don't really gain IQ over a DX sensor with
...Show more →

It wasn't just how much noise - it was the type of noise, and just something else about that sensor I didn't like. Quote a few of us complained about the IQ. Hopefully the D850 will be different. I agree, we definitely need some nice RAWs to look at, time will tell.
Aug 27, 2017 at 05:32 PM
jhinkey
Offline
Details
p.1 #13D850: Bill Claff's data shows significant High ISO DR improvement
Hmmm . . . just plotted my D800 and A7RII on the same with D810 and D850. Now I know why my D800 always looks so good at base ISO compared to my A7RII when I have to really raise the shadows!
Aug 27, 2017 at 05:34 PM
Stoffer
Offline
Details
p.1 #14D850: Bill Claff's data shows significant High ISO DR improvement
snapsy wrote:
Bill added more preliminary D850 today, this time read noise for much of the ISO range. Read noise corresponds to engineering DR, ie way DxO represents their DR measurements. The values indicate a nearly 1EV improvement in High ISO DR start at ISO 400, which matches Nikon's claim for the D850's 1-stop improvement in High ISO performance (although this is performance in the deeper tones/shadows vs an overall noise reduction that would be more useful).

D810 vs D850 Read Noise


Excellent. With this data, and the fact that the sensor is BSI CMOS in combination with using copper, there is no doubt in my mind that this is very similar technology to the Sony A7r2 sensor. Which is a good thing.

It is tweaked to do native ISO 64 with a bit better DR there, and of course no phase detect AF on the sensor, so not a direct copy of the one in A7r2. And of course no 12-bit limitation for continuous shooting, so I wonder whether something is tweaked to the A/D converters too?
Aug 28, 2017 at 12:00 AM
jhinkey
Offline
Details
p.1 #15D850: Bill Claff's data shows significant High ISO DR improvement
Stoffer wrote:
snapsy wroteBill added more preliminary D850 today, this time read noise for much of the ISO range. Read noise corresponds to engineering DR, ie way DxO represents their DR measurements. The values indicate a nearly 1EV improvement in High ISO DR start at ISO 400, which matches Nikon's claim for the D850's 1-stop improvement in High ISO performance (although this is performance in the deeper tones/shadows vs an overall noise reduction that would be more useful).

D810 vs D850 Read Noise
Excellent. With this data, and the fact that the sensor is BSI CMOS in combination with using copper, there is no doubt in my mind that this is very similar technology to the Sony A7r2 sensor. Which is a good thing.

It is tweaked to do native ISO 64 with a bit better DR there, and of course no phase detect AF on the sensor, so not a direct copy of the one in A7r2. And of course no 12-bit limitation for continuous shooting, so I wonder whether something is tweaked to the A/D converters too?


How do you know there is no 12 bit limitation for continuous shooting?
Aug 28, 2017 at 12:24 AM
Stoffer
Offline
Details
p.1 #16D850: Bill Claff's data shows significant High ISO DR improvement
jhinkey wrote:
Stoffer wrote
snapsy wroteBill added more preliminary D850 today, this time read noise for much of the ISO range. Read noise corresponds to engineering DR, ie way DxO represents their DR measurements. The values indicate a nearly 1EV improvement in High ISO DR start at ISO 400, which matches Nikon's claim for the D850's 1-stop improvement in High ISO performance (although this is performance in the deeper tones/shadows vs an overall noise reduction that would be more useful).

D810 vs D850 Read Noise
Excellent. With this data, and the fact that the sensor is BSI CMOS in combination with using copper, there is no doubt in my mind that this is very similar technology to the Sony A7r2 sensor. Which is a good thing.

It is tweaked to do native ISO 64 with a bit better DR there, and of course no phase detect AF on the sensor, so not a direct copy of the one in A7r2. And of course no 12-bit limitation for continuous shooting, so I wonder whether something is tweaked to the A/D converters too?
How do you know there is no 12 bit limitation for continuous shooting?


Because of the marketing text about buffer depth: "Despite the heavy load, the camera is capable of continuous shooting for up to 51 frames*2 (body alone) even in 14-bit lossless compressed RAW".

We don't know, however, whether it drops down to 12-bit in live view (or quiet mode with electronic first curtain shutter), but it is worth to note that I haven't seen any footnotes indicating any limitations so far.

But I know why you're asking: We are both used to some kind of 12-bit limitations coming from Sony. Very important to read the fine print with Sony.
Aug 28, 2017 at 01:06 AM
Eric Calabros
Offline
Details
p.1 #17D850: Bill Claff's data shows significant High ISO DR improvement
Still half stop behind A7Rm2. But considering its a 9fps beast, that compromise seems inevitable.
Aug 28, 2017 at 04:55 AM
Mark_L
Offline
Details
p.1 #18D850: Bill Claff's data shows significant High ISO DR improvement
Key for me is what teh noise looks like. The D800 was about 1 stop above the D700 but the character of the noise was unobtrusive and grainy rather than blocky which made a massive difference. I agree with the poster above about never using luminance NR on the files. I really hope this character does not change.

Eric Calabros wrote:
Still half stop behind A7Rm2. But considering its a 9fps beast, that compromise seems inevitable.


That camera and sensor are years old though, I'd say this was fairly disappointing for their flagship IQ camera.
Aug 28, 2017 at 05:23 AM
JohnK007
Offline
Details
p.1 #19D850: Bill Claff's data shows significant High ISO DR improvement
Mark_L wrote:
That camera and sensor are years old though, I'd say this was fairly disappointing for their flagship IQ camera.


Same as the A7rII wasn't as good at Base ISO as the D810, even though the Sony was new at the time, and the D810 was 'years old' as well.

Aug 28, 2017 at 07:36 AM
Cliff L.
Offline
Details
p.1 #20D850: Bill Claff's data shows significant High ISO DR improvement
Eric Calabros wrote:
Still half stop behind A7Rm2.


That's the price you pay to get a camera with reliable low-light AF, I suppose...
Aug 28, 2017 at 07:46 AM
FM ForumsNikon ForumJoin Upload & Sell
  
 

You are not logged inLog inRegister