D850

donaldejose

Yes, according to DSO Mark ISO improvement has not been one stop every time a new iteration comes out. But every iteration is not a new generation. By generations I am referring to the four year major cycles such as D1, D2, D3, D4, D5 not every new 2 year cycle model iteration. Also, DXoMark "sports" number is only one type of judgement. Let's look at those numbers.
D2H = 352
D3 = 2290 (more than a stop improvement)
D4= 2965 (about half a stop improvement)
Yes, it would be astonishing if DXoMark rated the D5 at 12000 ISO and I don't expert them to do so. However, I do expect many reviewers to claim they see more than a 1 stop improvement in high ISO between the D4 and the D5. I think that will be because Nikon has learned how to treat noise in dark shadows with stronger noise reduction while treating the noise in areas with a lot of detail much weaker. This gives the perception of a cleaner image because you don't see the noise in the large blocks of solid color where it is more obvious. Perhaps, it is my wishful thinking. We will see when production models are out for review.

Paperman

We can't really go back to the first years of DSLRs and expect that kind of improvement in sensor technology, can we ?

Plus, D3 to D4 is a 4 year period (the improvement is a more like 1/3 stops ) where as we are talking about a 2 ( maybe 2.5 ) year gap between the D810 of June 2014 and the D820.

I do agree with WestEndPhoto's opinion that the ISO improvement may mean something different for every user; it may be less in lower values and more in the higher end but that is of course very hard to measure. DxO figure gives an idea about when the image really starts to deteriorate.

Just saying I am not as optimistic as many here are . Digital tech has probably plateaued. I carry the same opinion about the D500 sensor - I think we will only see marginal difference from the D7200 ( less Mps will help though ).

spraynpray

Paperman said:

donaldejose said:

The D5 and D500 seem to be some sort of "breakthrough" on cleaner high ISO. ....... Normally, replacement camera models produce a stop or less of better high ISO but these two new models seem to be producing more than one stop of better high iSO. .....

No offense Donaldjose but there has never been a 1 stop / near 1 stop improvement in ISO performance over the years. We actually saw LESS THAN 1/10th stop improvement between models. L-) .

Copying what I wrote before :

D3s - ISO 3253 (2009)
D4 - ISO 2965 (2012)
D4s - ISO 3074 (2014)

Someone please factor in the Mp increase for me and we'll see what changed in 5-6 years.

and adding APS-C models

D7000 - ISO 1167 ( 2010 )
D7100 - ISO 1256 ( 2013 )
D7200 - ISO 1333 ( 2015 )

I wouldn't dream about that 1 stop difference

FWIW, my view is that I don't agree with you Paperman. I'll quickly re-state what I've said before:

I had no intention of selling my D7000 but when the D7100 came out, a fellow club member was kind enough to let me borrow her D7100 for an hour. During that hour I took about twenty images outside in the dark using the same subject, lens and settings. I was blown away by the difference. I will not and cannot talk numbers, because I have no faith in them but I can tell you that from body to body for the same exact settings and viewed on my PC at home, I saw a difference of almost two stops in the APPARENT noise in the raw files. The D7100 at 6400 ISO was virtually the same as the D7000 at 1600 ISO. The D7000 was more colour noise too which is harder to remove than the D7100's noise. I had no intention of buying a D7100 when I did the test, but I went straight out and bought one after that test.

I trust my own eyes more than anything else, certainly an apparently arbitrary number.

Another thing I have found is that comparing ISO improvements in good light is completely different to comparing ISO performance in very low light and as low light is why I use high ISO's, that is how I test it.

heartyfisher

Its always intellectually bothered me when you(spraynpray) said that the D7100 is 2 stops above the D7000 :-) I have the D7000 and now I have the D7200, I have not compared them, as high ISO is not what I use often as I always limit my ISO to as close to Base ISO as far as possible(by using all the techniques available to us since our Film days ...) However, I have now resolved it in my framework of understanding.

Like @paperman says, each camera has different noise characteristics as the ISO increases .. and it makes total sense to me that the "line drawn" by Spraynpray did move 2 stops between the D7000 and D7100 according to the personal criteria that he has. It did not for me (in the D7100 sample images that I have seen) but thats me. I will accept his opinion.

Did you know that some of the old cameras had documented jumps in degradation in factors of 160 iso .. ie there was a jump in noise after 160, 320, 640, etc .. so it was best to set the iso at those ISOs to get the cleanest image. ie 800 was way worst than you would expect vs 640. It could be that is what was seen by @spraynpray. eg lets say his D7000 had acceptable 800 ISO to him, not realising that it happened to be the first jump in noise boost after 640 ISO for that sensor. While 3200 ISO on the D7100 could be the last ISO before it makes the next Jump in noise boost.
So for him the D7100 did have a 2 stop advantage over the D7000.

As for me I set my auto ISO on my D7200 to stop increasing as 3200. And if needed I will manually go up to 12800. though I have probably done it only once..(not counting the times I set it as part of testing the camera). So I would say for me with regards to High ISO the D7200 is 2-4 stops better than the D7000 . But I don't use high ISO often enough to make it an important criteria.

donaldejose

I think spraynpray is correct here. DXoMark is great in that it is some sort of semi-objective number with which to compare two sensors assuming the testing is the same both times. However, it does remain their opinion on what level of noise is acceptable to most people. Your opinion and my opinion my be different. I even have different opinions on acceptable noise depending upon how large I may print the image. Generally with my D800 if printing poster size I do all I can to stay down to ISO 100 to 400. If I know the image will not be printed larger than 8x10 I will set auto ISO up to 3200. With my D750 I will set auto ISO up to two stops higher when I know most images will not be printed larger than 5x7. Expected end use greatly affects "acceptable" high ISO. Another factor is RAW vs JPEG. I often set my cameras to record RAW on one memory card and large fine JPEG on the other. Then I work first with the JPEG images and only go to a RAW image if I need it. Generally, if the exposure is good in the first place the room for change in the JPEG is sufficient and I don't need the additional latitude in the RAW file. This allows me to generally skip the lightroom step and work initially in Photoshop elements which I prefer. Surprisingly, I generally find that working with the RAW file is not needed and doesn't make a noticeable difference because all the changes I need to make can be accommodated within the latitude allowed by the JPEG file. It will be very interesting to see how DXoMark rates the sports ISO for the D5 compared to the D4 and the D500 compared to the D7200. If they rate a full stop difference that would be great. However, I highly suspect many people will, and have so far, claime they see astonishing gains in quality (not limited to noise) at high ISO. Whether or not Nikon will put this "magic sauce" into the D820 depends upon what is needed to create it and how that affects a 50+ megapixel sensor. If the same Expeed processor is put into the D5 and D820 that processor will have to be handling a much larger file in the D820 and this may impinge upon its ability to reduce noise at high ISO. My example would be the noise characteristics of the D3 vs the D3x. DXoMark rated them very similar but many people thought the D3x should not be set higher than ISO 800 while the D3 was good a full stop higher. Many factors are involved in the judgment of acceptable high ISO image quality. Time will tell and the reviews will be very interesting in this regard.

WestEndFoto

I assume every picture I take will be blown up big and critiqued by the harshest judges. I then shoot and process accordingly.

PB_PM

Paperman said:

No offense Donaldjose but there has never been a 1 stop / near 1 stop improvement in ISO performance over the years. We actually saw LESS THAN 1/10th stop improvement between models. L-) .

I wouldn't dream about that 1 stop difference

No offence Paperman, but you are very wrong.

Least we forget the jump from the D2H/D2X generation to the D3 generation. Or the leap from the D100 to the D200, or the D200 to the D300. In recent years the jumps have been less so, so in that respect you are right.

Artificial made up numbers from DXO mark are meaningless anyway.

donaldejose

WestEndFoto: That maybe fine for you but it sure would waste a lot of time for me. Sometimes I shoot 200 or more portrait photos in a session for a person. Adopting the philosophy of "good enough for the purpose the photo was wanted in the first place" saves a lot of time. I have stressed ultimate sharpness and ultimate image quality in the past and just saw a lot of my time and effort wasted. Now I shoot lots of "proofs." edit down to the "good ones," and post process only those. I have not had a person who was not happy with the 20 or 30 I sent them. Photos I shoot for "art" are different. Then I do adopt your philosophy.

WestEndFoto

donaldejose said:

WestEndFoto: That maybe fine for you but it sure would waste a lot of time for me. Sometimes I shoot 200 or more portrait photos in a session for a person. Adopting the philosophy of "good enough for the purpose the photo was wanted in the first place" saves a lot of time. I have stressed ultimate sharpness and ultimate image quality in the past and just saw a lot of my time and effort wasted. Now I shoot lots of "proofs." edit down to the "good ones," and post process only those. I have not had a person who was not happy with the 20 or 30 I sent them. Photos I shoot for "art" are different. Then I do adopt your philosophy.

Actually Donaldjose, I must concede that what I do in practice is closer to your approach than I implied. There are times that I adopt your philosophy. It is usually when I am doing something for somebody else rather than art. It was for this reason that I bought the 24-70 2.8E VR as I do a fair amount of shooting where I am not shooting for "Art". You won't see much of it posted on Flickr, but nevertheless, I do do it.

WestEndFoto

PB_PM said:

Paperman said:

No offense Donaldjose but there has never been a 1 stop / near 1 stop improvement in ISO performance over the years. We actually saw LESS THAN 1/10th stop improvement between models. L-) .

I wouldn't dream about that 1 stop difference

No offence Paperman, but you are very wrong.

Least we forget the jump from the D2H/D2X generation to the D3 generation. Or the leap from the D100 to the D200, or the D200 to the D300. In recent years the jumps have been less so, so in that respect you are right.

Artificial made up numbers from DXO mark are meaningless anyway.

I think it is going a bit to far to call DXO mark's numbers "artificial made up numbers". DXO may not explain their method that well which makes applying the scientific method to their claims difficult, but that does not invalidate them assuming you don't depend on only their results. I think that they have something to say.

donaldejose

WestEndPhoto and others: As someone who has learned how to produce quality work which can be printed larger than poster size it is quite disappointing to be asked for a "photo shoot" when you know the subject is just going to show the images on a cell phone or tablet and none will ever be printed larger than 5x7. The D3100 with its kit lens has more than enough ability to produce such images. Consequently, I have learned to "scale down" my effort in such situations. Recently I have even purchased soft filters for my zoom lenses in an effort to reduce their sharpness so I don't have to remove pimples and wrinkles in post processing! Zooms are still too sharp at f5.6. I think I may try to find a prime 85 or 105 mm lens (or around 50 to 150 range AF zoom) which is very soft at f5.6. Lensbaby maybe? I don't want to shoot at less than f4 because I want the DOF but I just want softness with a F4 or F5 depth of field like the old lenses had but with autofocus. When I am shooting for myself out comes the tripod and prime lenses!

Ironheart

To pile on a bit...

You can't compare ISO noise from camera to camera by itself, resolution matters. If you dig into the DxO scores, they are all relative to resolution, meaning they scale everything to an 8x12 printout at 300 dpi (this is what the standard DoF calculators do to BTW). It turns out that this is an 8mp image. So a D4 with a 16mp can be compared to a D810 with 36mp on a level playing field.

Any intelligent downsizing method (i.e. Photoshop, LR, etc...) will reduce noise by correlating signal to noise across the neighboring pixels. DxO talks about this in their methodology, and for all of the numbers you can look at absolute numbers or normalized ones. Except for ISO, the "sports" number doesn't seem to be normalized at all. Better to look at SNR where you can compare normalized (what they call "print") numbers. Also they say that a difference of 5 points on their overall score (which is logarithmic) equals 1/3 of a stop. Looking at that, the diff between D3 and D4 is 8, which is pretty close to a stop.

http://www.dxomark.com/Reviews/Detailed-computation-of-DxOMark-Sensor-normalization

The short answer is "you are all correct" because just like DoF or "sharpness" there is an element of "under what viewing conditions" and "what size and distance" you are looking. This means for one person there is a different threshold than another. DxO "sports" score is using a specific SNR of I think 18% but not normalized.

Put another way, viewing a D810 at 400% is like viewing a D4 at 1600%.

KnockKnock

I had a refurbished D7000 for about 2 weeks and eventually returned it due to AF issues. Took a lot of shots with it. When compared with the D7100 I ended up with I could see a clear improvement in low-light ISO performance. No, I wouldn't say it was 2-stops, but it was clearly more than ⅓ or ⅔ - probably around a 1-stop improvement in terms of noise and maybe more in terms of color "response." No, I don't have numbers to prove it, but just my eyes and hundreds of photos.

spraynpray

WestEndFoto said:

I assume every picture I take will be blown up big and critiqued by the harshest judges. I then shoot and process accordingly.

This is my mindset with my better shots too. Makes life hard doesn't it!

@KnockKnock and others: Note my comments about low light. For my use, low light is pretty close to no light, YMMV.

Msmoto

ISO... as one who came from film, I find on the D800E an ISO of 6400 is quite acceptable. But, each photographer has a different set of criteria for what their end goal is in recording an image. Mine is to capture the mood/atmosphere of an event,

Thus, in low light, some noise may be an asset not a liability in the final image. I suspect what we are seeing in the higher ISO's in the newer bodies is simply less noise, thus Nikon feels justified in rating the sensor as higher native ISO.

As to new bodies...I am still trying to learn the ones I have....

rmp

As we all have our different purposes and goals, I thought you may get a good laugh at this one. Once, long ago, I had a shot my wife wanted blown up to a 4 ft. by 6 ft. size to be hung on a wall in our home. When my lab tried it they called me to come over and look at it -- before they would print it. It looked terrible at that size. Today, many years later, I am still recovering from that stupid shot. Today, I want a tripod, lots of light, and the sharpest glass available. I laugh at my compulsiveness, then spray-and-pray. And yes, 90% of my shots are shown on the web.

donaldejose

What does the web display? For Nikon Rumors we are limited to 428x640 which is a little more than one fourth of a megabye and this is a photo forum! Good luck with the sharpest glass and a D810 sensor making any difference at that size.

WestEndFoto

donaldejose said:

What does the web display? For Nikon Rumors we are limited to 428x640 which is a little more than one fourth of a megabye and this is a photo forum! Good luck with the sharpest glass and a D810 sensor making any difference at that size.

I think a better question is, "What will the web display when my images are still relevant in the distant future?"

I assume that if your images are nice today, they will still be nice in 20 years?

After all the blood, sweat and money I have spent on this hobby, I would hate to turn an asset with a 50 year life into an asset with a 5 year life by shooting for today's standards, not tomorrow's. Especially when my equipment and skills permit me to shoot for tomorrow's.

rmp

I totally agree -- A D810 or a D7200 is much too much camera for a web picture. In fact a cellphone camera is good enough for web size pictures. However, once in a while I get a really good picture and then I want to print it really large. With a cellphone camera, large-blowups are not possible. So, I want as many megapixels, as much ISO, the sharpest lens available, as much light as I can get, and a tripod.

Nikon -- please bring out the D854 this year.

nek4life

There's more to a sensor than mega pixels though. Also retina displays are requiring larger photos, but I do agree with the spirit that a D810 is probably overkill for web images when strictly speaking in terms of mega pixels.

fiziks

From what I can tell, DXOMarks "Sports" rating seems to be where the sensor's ISO line crosses the 30db line on the SNR 18% chart using the "Print" setting. This is from a comparison of D3, D4 and D3X (Nikon's first 24MP sensor. Here's an exercise. Compare the D3x and the D90 on DXOMark. Click on "Measurements". Click on "SNR 18%". Then click on "Screen".)

donaldejose

I ordered a D500. When it arrives I will be able to determine whether or not it is more than sufficient for more than 90% of what I do. I just tested the latest DX sensor technology by ordering a refurbished D5500 and was very impressed. I am hoping for a D500 viewfinder that is more FX size than the smaller DX size so I will no longer feel the image I am looking at is cramped compared to an FX viewfinder image. I am hoping for an ability to enlarge to poster size at base ISO and not be able to tell the image was not shot on an FX size sensor. I am hoping for clean ISO up to 25,600 to be able to shoot indoor high school sports in dimly lit gyms in jpeg without having post process out the noise (I can increase saturation and noise reduction in camera if needed). I am hoping for a large enough buffer (shouldn't be a problem shooting jpeg) to not bump up against that limitation when shooting fast action sequences. I am hoping for top of the line focusing ability. I am hoping for good enough fps I will not want more for birds in flight. I am hoping Nikon and Sigma Art produce more fast DX lenses to match the D500. I am hoping for f1.8 to be the new f1.4 and f4 to be the new f2.8 because the new sensors don't need the additional light. I am hoping for essentially a $2000 D5 missing just a few unneeded fps, unneeded extreme ISO and the last 100,000 clicks of shutter life which I will never use anyway before new technology makes the D500 outdated. That is a lot to hope for. The one thing a D500 will never be able to do is replace a 50+ megapixel FX sensor for certain shooting such as landscapes. Thus, I am thinking the "perfect" pair for me may be a D500 and a D820. For others, the "perfect" pair may be a D500 and D5. And for still others the "perfect" pair may be a D7200 and D750. I think "perfect" includes a DX and FX body with the same control layout so you can move from one to the other seamlessly. I want to have two bodies set up the same with the same control layout so I know what each button does without looking and so I know what the settings are without looking. Less to look at, less to remember and more time focused on the image to be captured because the two tools are as identical as can be. I am thinking that DX should be the smallest sensor I use for anything beyond dedicated cell phone or web display. I have shot some things on a CX V2 for web use and it is fine but I don't want to use CX for anything I might later want to print larger than 8x10. I am thinking CX is dead, at least for me, because DX can be put into something as small as the Coolpix A and as light as the D5500. Thus, I can get both small size and light weight while still retaining the better image quality of DX. There is just no need for CX and its sensor limitations.

WestEndFoto

Well said Donaldejose. I suspect that you are mostly correct, including your comments about CX. Besides cost, I see no other reason for CX. It is not that convincing an upgrade for a cell phone user.

My perfect pair would be two D810s, or perhaps two D820s later this year. I suspect, however, that it will be a 820/800 combo, perhaps with a DF2 thrown in for "fun".

Ironheart

I can put 6 lenses, two bodies, and a bunch of gear into a 6lb camera bag. Try that with DX or FX
100% agree with everything @donaldejose has to say (usually the case) I would just add that weight is more of a consideration for when I choose my CX kit. Cost is laughable as I have $5k in CX lenses already
One last thing, when I think of a "perfect pair" or a "nice pair" my mind immediately goes to the cover of one of my favorite pink floyd albums:

(warning NSFW)

http://pinkfloydarchives.com/Discog/Turkey/LP/Comp/ANP/ANP2/FC.jpg

WestEndFoto

Ironheart, my second favorite pair type. But most of the time I get access to both if I have access to one.