Steve Spencer
Online
• • • • • • •
Upload & Sell: On
|
 p.44 #4 · Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens! | |
zhangyue wrote:
Hi, Steve, Just have time to reply you. thanks for point out error on my side. I couldn't find information about Claff's DR is at pixel level or normalized. but I think you were right given the fact for the same camera has less DR at DX mode than FF. This is the 2nd time I am confused with Claff's measurement I think.
As D850 has better DR than other miniMF, I revisit the DXO sensor score and compare the two they have D850 and X1D. (they seems don't have GFX50 and 645Z)
Here are the comparison I got:
screen
https://farm5.staticflickr.com/4443/24153544538_39ee082e68_b.jpgScreen Shot 2017-10-28 at 11.34.03 PM by ZHNL, on Flickr
D850 is 13.55EV, X1D is 13.49EV
Print
https://farm5.staticflickr.com/4504/24153544838_83b1771fcf_b.jpgScreen Shot 2017-10-28 at 11.34.26 PM by ZHNL, on Flickr
For print, because X1D has more pixel, so it break even now with very minor advantage.
D850 is 14.81EV, X1D is 14.83EV, but it doesn't have high ISO capability of D850.
There should be negligible difference between DXO and Claff measurement difference other than DXO use 0dB as base and Claff use 20dB as base for SNR measurement.
Before DXO measure X1D sensor, actually there is only one place I can find technical measurement for both D850 and miniMF sony sensor: irental (Largest rental company in China). Their past measurement are all with match DXO and Claff since they all measure raw data from sensor.
Here is their results for dynamic range:
https://farm5.staticflickr.com/4494/38005212971_6ff2bc3324_b.jpg1504864437869429 by ZHNL, on Flickr
For understanding this graph, I need explain what they did for this score. They use -6eV(raw data) and down sampling the sensor, until SNR=30dB, which they feel a good threshold for nice IQ for landscape. D750 need downsize until to 10M pixel to get SNR=30dB, D850 downsize until 17.77M. D810 downsize until 15.6M and GFX need downsize until 12.3M. A7R 9.79M, A7RII 8.75M etc...... if you use D750 as reference as score 100, you can get score by using equivalent pixels/10M(D750 results)X100=score.
This looks quite technical and complicated, but fundamentally, just another way to calibrate between sensor by using 30dB as threshold.
We are purely doing technical discuss here. I don't want make a big deal out of DR advantage between cameras. The bottom line is I didn't make false claim about my original post about D850 sensor is score higher than any other miniMF other than phase IQ3100. DXO now of course score X1D higher than D850 though we don't know how that score is calculated.
D850's sensor performance is so good doesn't come with a surprise that ISO64 at 45+M pixels (allow more light(signal) to captured) will compensate 1.5 size advantage of miniMF sensor size. and we all know Nikon is the one really know how to push design limit for given sensor.
Sorry for the OT.
...Show more →
Michael,
Thanks for this helpful information. Now that DXO has the data for the X1D we have a second sight that measures dynamic range. Bill Claff's site had the Mini MF sensors having about a quarter of a stop more dynamic range. DXO has them with equal measurements (note the print scores at DXO are the one that measure accounting for downsizing of the files whereas the screen scores are at the per pixel level). If you use the screen scores you often get surprising results like the D750 score considerably higher than the D850. Put simply the screen score are only useful if you are looking at both are 100% on the screen and are ignoring that the higher megapixel sensor is magnified more. So, DXO has a result quite similar to Bill Claff, but Claff has the difference being larger. Neither site suggests that the D850 has higher dynamic range.
Now let's talk about the Chinese site. I would not call what they did a measure of dynamic range if I understand it correctly, because it seems to just focus on shadow noise and does not take into account highlight room as well. You need both to measure dynamic range. I think of it more as a measure of shadow noise or perhaps the related construct of read noise (or how clean the image processing is). One thing is sure about that chart, it is very different from either Claff's or DXO's measures of dynamic range. Whatever it is measuring if seems to be a different construct as it isn't very closely related to other measures of dynamic range.
So, no doubt the D850 has great dynamic range. I don't think we should quibble about whether it has equal dynamic range to the X1D (as DXO suggests) or whether it has a quarter of a stop less (as Bill Claff suggests), but in any event they are close. I think we just shouldn't say that it has more dynamic range. I think that goes for the miniMF systems as well, if they have more it is such a small amount more it would hardly be noticeable.
What is more likely to be noticeable is noise. DXO and Bill Claff both agree that the MiniMF systems do much better at high ISO. Bill Claff has the X1D as doing about a stop better at high ISO, wheres DXO has it as a little more than half a stop, but either way the miniMF systems measured by both sites show a clear advantage for the miniMF systems at high ISO.
Ok, let me note one more fly in the ointment, aspect ratio. Even though miniMF sensors are at least equal on dynamic range that might not be true depending on the aspect ratio you want to use. If you crop to 3 X 2 or a skinnier rectangle you are cropping the MiniMF sensor more (of course you aren't cropping the D850 at all for 3 X 2) and that will affect dynamic range and noise. In fact, at a 3 X 2 crop the miniMF sensor has 45.4 megapixels, and have a touch less dynamic range. So for crops of 3 X 2 or a skinnier rectangle, the D850 is going to have ever so slightly more resolution and ever so slightly more dynamic range. Conversely, if you shoot at 4 X 3 or squarer you are cropping the 3 X 2 FF 35mm image more. In fact, at this crop the D850 only has 40.6 megapixels compared to the 51.1 megapixels for the miniMF camera. Here the difference in both resolution and DR is bigger and begin to be noticeable especially on bigger prints.
So, I think the answer is not a simple one, and I think it is misleading to say something like the D850 has better dynamic range than miniMF sensors. If you shoot in a 3 X 2 aspect ratio or a skinnier rectangle there might be a small advantage for the D850. If you shoot in 4 X 3 or squarer their might be a small advantage for miniMF.
In my own shooting comparing the Fuji GFX and the Sony A7r II, I see a small advantage for the Fuji even in 3 X 2 for dynamic range and this difference is a little bigger if you crop 4 X 3 or squarer, but I would say at the same time unless you are very very picky, I don't think this difference is large enough to offer very much in practical use for the miniMF system. Differences in lenses can easily overwhelm this difference and personally I make my decision which I use based on how far I have to schlep the gear. If I don't have to carry things far, I tend to shoot landscapes with the Fuji, but I often choose the Sony because it is so much easier to cart around. My advice to people thinking of switching to miniMF for increased resolution or dynamic range is be cautious. You can easily spend a lot of money and get very little if any benefit, especially if you shoot 3 X 2 or a skinnier rectangle for most of your shots, as many landscape shooters do. Now after the next generation of miniMF sensors I expect this to change, but even then don't expect huge gains from miniMF unless you prefer to shoot primarily 4 X 3. If you crop to 3 X 2 the miniMF sensor in size is less than half a stop bigger than FF 35mm if the sensors are equally dense and convert light to digital signal in the same way, and if you crop to 4 X 3 the miniMF sensor is still less than a stop bigger if you equalize the sensors. Given that FF 35mm sensors will likely always be ahead in terms of technology, the actual gains are likely to be smaller as we see above when comparing the D850 to the X1D.
Obviously, for me I got a GFX so I think it is worth getting, but if I only shot landscapes I don't think I would have gotten it. It is a great landscape camera, but so is my SonyA7r II and now the Sony A7r III and the gains for the Fuji are very small and I expect even with the next generation of sensors in everyday use they will rarely show a noticeable difference.
|
