dxoMark Nikon ISO numbers

astroNikon · Oct 2, 2014

Don't ask why I did this.
I was just curious, as it shows the evolution of ISO sensitivity of the FX and DX line.
a 25% difference in numbers represents a 1/3 stop performance advantage.

Listings are Ordered by the ISO ratings
ISO = DxO ISO Low LIght test results
ISO Range = the normal ISO range
Extended = extended ISO, normally identified as "Lo" & "Hi"
DR = Dynamic Range
MP = megapixels
Year = Year released
F Pts = Focus Point / Cross Type focus point
xx = Sensor Module / Image processor / Frames Per Minute in FULL size mode
=== FX Full Frame ===

Nikon	Canon	ISO DxO	ISO Range	Extended	DR	MP	Year	F Pts	xx
Df $2749	.	3279	100-12,800	50-204,800	13.1	16.2	2013	39/9	Multi-CAM 4800 Expeed 3 5.5 fps
D3s $5200	.	3253	200-12,800	100-102,400	12	12.1	2009	51/15	Multi-CAM 3500FX Expeed 2 9 fps
D4s $6500	.	3074	100-25,600	50-409,600	13.3	16.2	2014	51/15	Multi-CAM 3500FX Expeed 4 11 fps
D600 $2099	.	2980	100-6400	50-12,800	14.2	24	2012	39/11	Multi-CAM 4800 Expeed 3 5.5 fps
D800e $3300	.	2979	. 100–6400	50–25600	14.3	36	2012	51/15	Multi-CAM 3500FX Expeed 3 4 fps
D4 $6000	.	2965	. 100–12800	50–204800	13.1	16.2	2012	51/15	Multi-CAM3500FX Expeed 3 10 fps
D750 $2300	.	2956	100–12800.	50 to 51,200	14.5	24.3	2014	51/	Expeed 4a 6.5 fps
D610	.	2925	100 to 6400	50–25600	14.4	24.3	2013	39/9	Multi-CAM 4800 Expeed 3 6 fps
D810 $3300	.	2853	64–12,800	32-51,200	.	36.3	2014	51/15	Multi-CAM 3500FX Expeed 4 5 fps
D800 $3000	..	2853	100–6400.	50–25600.	14.4	36	.2012	51/15	Multi-CAM3500FX Expeed 3 4 fps
.	1dx $6800	2786	100–51,200.	50-204,800	11.8	18.1	2011	61	2xDIGIC5+ DIGIC 4 12 fps
.	6d $1899	2340	100–25600.	100–102,400	12.1	20.2	2012	11	DIGIC 5+ 4.5 fps
D700	.	2303	.	.	12.2	12	2008
.	5d mk III	2293	.	.	11.7	23	2012
D3 $5000	.	2290	.	.	12.2	.	2007
D3X	.	1992	.	.	13.7	24	2008
.	5d mk II	1815	.	.	11.9	21	2008
.	1ds mk III	1663	.	.	12	21	2007
.	1ds mk II	1460	.	.	11.3	17	2004
.	5d	1368	.	.	11.1
.	1d mk IV	1320	.	.	12	15	2009
.	1d mk III	1078	.	.	11.7	21	2007
.	1d mk II	1003	.	.	11.1	8	2004
.	1d mk IIN	975	.	.	11.2	8	2005
.	1ds	954	.	.	11	11	2002
.	1dc	xx	.	.	xx	17	2012
.	1dx	xx	.	.	xx	18	2011

[TBODY] [/TBODY]

===DX Crop ===

Nikon	Canon	ISO	ISO Range	Extended	DR	MP	Year	F Pts	xx
d3300	.	1385	100 – 12,800.	100 - 25,600	12.8	24.2	2014	11	Expeed 4 5 fps
d5300	.	1338	100–12800.	100–25600.	13.9	24.1	2013 $800	39	Multi-CAM 4800DX Expeed 4 5 fps
d5200	.	1284	100–6400	100–25600	13.9	24.1	2012	39	Multi-CAM 4800DX EXPEED 3 5 fps
d7100	.	1256	.	.	13.7	24	2013	51
d5100	.	1183	.	.	.	16	2011
d7000 $1199	.	1167	.	.	.	16	2010	Multi-CAM 4800DX
d3200	.	1131	.	.	.	24	2012
.	7D mk II	1082	.	.	11.8
d90 $999	.	977	.	.	.	12	2008
.	70D	926	xx	xx	11.6
d3100	.	919	.	.	.	14	2010
d5000 $730	.	868	.	.	.	12	2009
.	7D	854	xx	xx	11.7
.	100D	843	xx	xx	11.3
.	60D	813	xx	xx	11.5
.	600D	793	xx	xx	11.5
d300s	.	787
.	550D	784	xx	xx	11.5
.	1100D	755	xx	xx	11
.	30D	736	xx	xx	10.8
.	1200D	724	xx	xx	11.3
.	650D	722	xx	xx	11.2
.	20D	721	xx	xx	11
.	1000D	719	xx	xx	10.9
.	40D	703	xx	xx	11.3
.	50D	696	xx	xx	11.4
.	450D	692	xx	xx	10.8
.	700D	681	xx	xx	11.2
d300 $1800	.	679	.	.	12	12	2007
.	400D	664	xx	xx	11
.	500D	663	xx	xx	11.5
.	350D	637	xx	xx	10.8
d200 $1699	.	583	X	10	2005
.	10D	571	xx	Xx	10.9
d3000	.	563	.	.	11.1	10	2009
d60	.	562	.	.	.	.	2008
d40	.	561	6	2006
d50	.	560	6	2005
.	300D	544	xx	Xx	10.8
d70s	.	529
d70	.	529	.	.	.	6	2004
d80	.	524	10	2006
d40x	.	516	.	.	.	10	2007
d2XS $4700	.	489	12	2006
d2X $5000	.	476	12	2004
d2H $5000	.	352	4.1	2003
D100 $1999	.	.	6	2002
D2HS #3500	4	2005
D1H $5000	.	.	.	.	.	2.7	2001
D1X	5.3	2001
D1	2.7	1999

[TBODY] [/TBODY]

shadowlands · Oct 2, 2014

Thanks for sharing. Nice to see. Learned a lot. D90 still got it doing on, for it's age....

Solarflare · Oct 2, 2014

FYI, the High ISO number of DxOMark is actually pretty awful as a rating.

It only tells you at what ISO the camera will drop below a certain more or less randomly chosen performance (dynamic range, color depth).

It does NOT tell you how exatly the camera will behave before or after that point, and it will NOT tell you how exactly it will behave at all. It wont tell you if the camera keeps giving you any dynamic range, if the camera keeps good colors even in high ISO, it wont tell you what kind of noise will be prevalent (since there are more obtrusive and less obtrusive kinds), it wont tell you if there will be banding problems, and it wont tell you until which point the camera will boost the performance natively, i.e. through an analog amplifier before the A/D converter, and so on and so forth.

So for example, judging from example pictures I saw on the net, the Nikon D800 and D810 seem to have a problem to keep natural colors, more so than sensors of less high pixel counts, when it comes to high ISO. That means they are actually quite a bit worse as High ISO cameras, than a D4s, Df, or the current king of High ISO, the Sony A7s, despite of having a quite good DxOMark rating in respect to High ISO, not much behind the D4s and Df.

The D3s gets a great high ISO rating - but it only does amplify natively up to 12k. The D4s amplifies natively up to 25k and has an advantage over the D3s in that regard. The Sony A7s amplifies up to 100k.

So yeah, the DxOMark number gives you a nice general idea, but if you really want best performance in high ISO, dont rely on it blindly.

KmH · Oct 2, 2014

Randomly chosen? I don't think so. ISO sensitivity - DxOMark
Please link us to a better, non-subjective rating system than what DXOMark uses.

Test result reliability - DxOMark
The SNR indicates how much noise is present in an image compared to the actual information (signal). The higher the SNR value, the better the image looks, because details aren't drowned by noise. SNR strength is given in dB, which is a logarithmic scale: an increase of 6 dB corresponds to doubling the SNR, which equates to half the noise for the same signal.

An SNR value of 30dB means excellent image quality. Thus low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits.

A difference in low-light ISO of 25% represents 1/3 EV and is only slightly noticeable.

Derrel · Oct 2, 2014

Thom Hogan published a piece on September 29, 2014, using DxO Mark's sensor ratings over the past decade, to show the long, steady progress in sensor performance on Nikon's professional and "prosumer" DX-format cameras, beginning with the D2x in 2004, and ending with the D7100 in 2013. He graphs both the overall sensor performance scores, as well as DxO Mark's High ISO test results, for eight different Nikon bodies. You can read his article here: Sensors Are a Moving Target | byThom | Thom Hogan

PaulWog · Oct 3, 2014

Interesting.

I think a lot of people get hung up on these dx vs fx numbers... I still do. But then when I look at what I often shoot at, and I realize that anything slower than your fastest aperture on dx is fairly transferable to fx.

Example, if you CAN shoot at f1.8, but say you want f2.8 on a DX sensor, then the equivalent shot on FX is shot at f4, rounded approximately (same depth of field). So, you've got similar noise performance at the same exposure.

Only reason I bring this up, other than it mattering to me, is there's a lot of beginners around here who forget that fact. I mean, if you've got a slow piece of glass (150-600), or budget glass that is slow (18-55), that's different... But then that's talking budget, size, or reach needs, rather than fx.

These numbers don't matter much to me though. I'd prefer just to look at raw image sample comparisons with equivalent processing.

Solarflare · Oct 3, 2014

Well - thats simply wrong. And if it was actually true, who would be so stupid to pay so much more money for larger sensors, anyway ?

You run both the DX and FX sensor at base ISO, more often than not, as well as any other sensor really, because quite often there is MORE than enough light around for that, which is then thrown away by using a shutter speed faster than would be necessary. And even if there isnt enough light, you can often make it have enough light, by using a flash. And even if that doesnt work, you can limit yourself to use maximum aperture and shallow depth of field, and by accepting this limit you'll still have more light than with the smaller sensor. So yeah, a larger sensor DOES give you more reserves.

Thus you need to get to really large sensor sizes (large format really) before your argument is truely an issue (and even then large format offers you tricks with your focal plane, so even then it might still not be true). Technically it might also appear at really high demands for shutter speed - i.e. sports - but then you have no choice than to work at maximum aperture anyway, so that argument fails again.

In fact for some scenarios, like studio photography, you have full control over the light and will ALWAYS and ONLY work at base ISO. Thats when these medium format cameras that do not work well on anything close to base ISO, but work really, really well when used this way, are so popular for pro studio work.

Thus, a larger sensor gives you more possibilities, and in most scenarios, you will get better performance from it. Again, if that wasnt the case, nobody would spend so much extra money for a full frame or, even much worse, medium format camera.

Also, the performance of todays sensor are pretty much at the physical limit of whats possible at all, with current technology. Which the graph of Hogan shows, but shows only VERY VERY poorly, because he hasnt bothered to actually, oh I dunno, MAKE A PROPER TIME SCALE. Its realy just a list of of cameras, sorted by time, instead. But just check the difference between the D7000 and the D7100 - its minimal compared to what happened before, and yet theres three years between those two. So yeah, unless technology advances in some way, I dont expect to much more progress.

Things they can still improve:

- dark noise - Depends upon the temperature, thus you could cool that circuit.
- readout noise - Again, you could cool that circuit to reduce this.
- bayer filter - if you could get rid of that, you wouldnt throw away 50% or more of the light. You wont get less noise though, you just could operate with less light.
- efficiency - if you could raise that, again you could get same results with less light. Again you wont benefit in other respects, though

Things they cannot improve:

- Photon noise (not sure how the english word is really). Its because light is is actually made from single particles of a certain energy - photons - and they are random. Thus the less light there is, the more random variation is in the light. Thus this noise depends upon size of the pixels. The better you get with dark noise and readout noise, the more important this kind of noise gets.

PaulWog · Oct 3, 2014

You totally missed the point... it completely went over your head.

I would like to emphasize what I said: "Only reason I bring this up, other than it mattering to me, is there's a lot of beginners around here who forget that fact." There are particular qualifiers in what I wrote... And I'm *clearly* not talking about studio work, full frame work, high-end professional work, comparable work in which absolute base ISO performance insanely important, etc...

You preface your response poorly... your response comes off as stuck up because of that. You skipped over all my points, and I'm not quite sure if you even read what I said... You kind-of came up with a strawman argument on so many levels... Your very first point is that FX has more reserves. My very first point was concluded by "... anything slower than your fastest aperture on dx is fairly transferable to fx." If you had read that line, why would you even produce such a pompous, irrelevant response? I was speaking to the average joe...

I know you're providing some technically good points, I just don't like the whole "You're completely wrong" start to your response... that's just pure misinformation on your part. I am in fact not completely wrong, if you read what I actually have written.

Vince.1551 · Oct 3, 2014

Here it goes again lol

Sent from my iPhone using Tapatalk

jaomul · Oct 3, 2014

There just numbers

hamlet · Oct 3, 2014

I actually have come to accept noise as part of the sacrifice to getting my shot. Before when i was so worried about it, i missed a lot of shots or shots that were ruined by thin dof or slow shutter speed. Now i just don't worry about it anymore. Just get the correct settings for my shot and let the iso fall where it may.

astroNikon · Oct 3, 2014

hamlet said:
I actually have come to accept noise as part of the sacrifice to getting my shot. Before when i was so worried about it, i missed a lot of shots or shots that were ruined by thin dof or slow shutter speed. Now i just don't worry about it anymore. Just get the correct settings for my shot and let the iso fall where it may.

You are learning grasshopper ... or tribble .. hamster

Solarflare · Oct 7, 2014

PaulWog said:
I would like to emphasize what I said: "Only reason I bring this up, other than it mattering to me, is there's a lot of beginners around here who forget that fact."

Aha.

So you wanted to "teach" people things that arent true - and I disrupted your game.

Sorry about that, I guess.

PaulWog · Oct 7, 2014

Solarflare said:
PaulWog said:

I would like to emphasize what I said: "Only reason I bring this up, other than it mattering to me, is there's a lot of beginners around here who forget that fact."

Click to expand...

Aha.

So you wanted to "teach" people things that arent true - and I disrupted your game.

Sorry about that, I guess.

You're in a really antagonizing mood...

I directly addressed what you said and explained why your response is irrelevant. Go shoot at f1.8 at 35mm on a D7100 and compare that with 50mm at f2.8 on a D610. Frame it identically. Tell me they don't have similar depth of field, field of view, and noise, as I said. Or you can keep cupping your ears and saying "it's not true, it's not true"... doesn't change the facts...

Re-read my original post. It's not wrong. It's entirely correct. The only distinction that could be made (to be more clear, and I believe I already was since I made that distinction) is that the post is directed at non-professionals (ie. the guy who is deciding what he wants to buy). I make no incorrect claims. If you want to keep arguing to the contrary, then... be grumpy.