47 replies to this topic

[Fred Nirque]

The best DSLR I have been able to use regularly is the D3s. Its low-light capability is legendary, and up until recently I thought its DR was all I'd ever need - it seemed to always find detail in shadows and highlight blowouts didn't seem to be a problem, but then I got the X-Pro1. Not that it is a replacement for the D3s, but as I've mentioned here before, it struck me that I was seeing more shadow detail but particularly more highlight recovery ability. Still awaiting DXO testing to confirm this (if they ever get around to it) yesterday I took both the D3s and Fuji with fisheye lenses attached for a walk in a paddock just after sunrise, and coincidentally took a similar view with both cameras one just after the other looking straight at the sun.

The result made me start to seriously contemplate whether in fact I should be thinking of trading the D3s for the D4, and in fact just how well the Fuji will rate in the DXO DR test should that ever happen.

D3s:


X-Pro1:


For consistency both were processed in ACR with the highlight recovery slider adjusted until just before colour was lost altogether and the obvious "blown" look started to appear. The Samyang 8mm also gives a wider view than the Tokina 10-17 @ 16mm, which I didn't crop to match as that has no effect on this comparison.

Edited by Fred Nirque, 26 November 2012 - 22:54 .

[schwett]

that is an impressive result. i'm not sure the d4 would be so much better at this particular test than the d3s. dxo (if you take stock in such things...) says it has only 1 stop more DR than the d3s. they claim the d800/e have 2.3 stops more... but they also say the d7000 does, almost, which doesn't really jibe with what i saw when i owned that camera.

i'm curious if both cameras metered the scene the same - as you note it's easier to recover the shadows than the highlights sometimes.

[Fred Nirque]

I didn't have to adjust the exposure or brightness in ACR, so the program must have thought the two were exposed the same. I can't check in EXIF because the Fuji doesn't recorf the F/stops on 3rd party lenses.

Perhaps I should have included a 100% sample - it's a bit hard to appreciate things with such a large AOV. So here they are, D3s first:

D3s


X-Pro1:

[Ron Scubadiver]

I think that the DXO test might not make sense with the Fuji sensor because it is so different. As it is, I am not sure how much sense it makes at all. Their numbers would suggest a 1/3 stop improvement from the D700 to the D800 in high ISO noise, but in real image quality it is much more.

[Fred Nirque]

Something I will mention here and anywhere else that those interested might see it is that the Fuji is bordering on lousy compared with the D3s in high ISO, long exposures (10+secs) . While the NR is effective, it takes at least as long as the exposure itself afterwards, during which time the camera is locked up solid. By comparison the D3s produces a better result in about a second or two. So while in low light at high ISO (up to 6400) with shorter exposures (less than 2 secs) the Fuji does very well, things go belly-up quickly after that as the exposure length increases.

[pluton]

This is an encouraging result, but I wonder how much of the difference is---or could be--- lens related.

[Fred Nirque]

On the basis that I had noticed how much better the Fuji files recovered highlights than the D3s could regardless of lenses (and often having used the same lens), this just confirmed things on a close one-on-one comparison. Ideally I would like to put the Fuji up against the D4 with everything identical, but I don't have a D4 and I'm not a tester so that leaves that on the scrap heap. I hear the D4 is very good in this area, I'm interested in how much better than the Fuji, if in fact it is.

[Alaun]

"For consistency both were processed in ACR with the highlight recovery slider adjusted until just before colour was lost altogether and the obvious "blown" look started to appear."

But is this the correct way in ACR? As I understand the new concept, first the exposure should be adjusted then the highlighst and the white (mostly, just in the order, the sliders are arranged).

Werner

[Fred Nirque]

Things were adjusted, in order, correctly, as can be seen by the rest of the image in both case being a very close match in everything else. As it transpired the exposure in both didn't need adjusting - they were both OK as can be seen by the rest of the shot in both cases.

The highlight slider was adjusted as described to recover as much as possible of the highlights after that - anything past this was completely blown. That's what I meant by "for consistency" - I simply used the same method for that particular slider in both pictures to judge when the effective maximum recovery had been achieved before colour was lost completely.

I am not a tester, a reviewer or an author. I am a photographer, pure and simple, and I put this up as a point of interest to illustrate comments I have made before that I've noticed during my workday life as a photographer that in normal day to day situations when I've used both the Fuji and the D3s on the same shoot that the Fuji has displayed greater latitude in its ability to recover highlights in particular. This was the first time I actually took virtually the same shot with both cameras, and I thought that it just might be of interest to others who might be curious as to the practical performance as far as extreme DR is concerned in an actual shooting situation using two different cameras which bear out observations I had made previously. That's all.

As far as DR goes, it doesn't get much more extreme than shooting straight at the sun.

[Ann]

That was about to be my question too, Werner!

ACR 7.x has specifically been designed for the User to start by setting the Exposure slider and then to work down the list from the top.

"Default" settings need to be set to your own personal Defaults or not used at all because there is absolutely no way for Adobe to know how an individual shoots or what their personal preferences for imaging might be so any preset Default settings were probably designed with the casual snapshotter in mind?!

In this case, my guess is that either the Nikon shot was over-exposed or the ACR Exp. slider needed to be set back and the shadow slider then moved to the right and the HL Slider to the left?

Edited by Ann, 26 November 2012 - 04:28 .

[schwett]

That was about to be my question too, Werner!

ACR 7.x has specifically been designed for the User to start by setting the Exposure slider and then to work down the list from the top.

"Default" settings need to be set to your own personal Defaults or not used at all because there is absolutely no way for Adobe to know how an individual shoots or what their personal preferences for imaging might be so any preset Default settings were probably designed with the casual snapshotter in mind?!

but if the exposure slider didn't need to be adjusted, it's a moot point. most users of ACR will pursue an iterative process, starting at the top and going down but then also bouncing around. i often find, for example, that after adjusting vibrance and saturation (which are at the *bottom*) that one has to go back and readjust highlights, blacks, and sometimes exposure.

the difference here is really striking. in my experience with ACR when something is as fully blown out as the example above, pulling back exposure and highlights just turns it grey and slightly softens the edge. the result from the x-pro1 is impressive, if somewhat mystifying.

[Fred Nirque]

Gee bluddy zuz.

I know how to use ACR, F.F.S.

Can we please return to the highlight recovery abilities of the X-Trans sensor, being the subject of the original post, and not whether the author is too bloody stupid to know how to use ACR properly?

[PedroS]

Nice writing Fred, as usual...

Not saying the opposite, but the two first pictures aren't, for me, comparable, as the sun is much brighter in the first vs the second. So it's normal to show more shadows detail.

[Michael Erlewine]

Not wanting to play the devil's advocate, Fred, but IMO the camera for you, aside from your losing interest in having one, is the D800E. That is just my two cents, but in many ways the kinds of things we photograph are similar, and the D800E is by far the best camera I have ever owned, despite whatever shortcomings people point out about it.

[Andrea B.]

I have these Dpreview dynamic range charts for the XPro and some Nikons. Please don't use these charts to beat anyone over the head with! They show that all the cams are very good these days. A couple of bars on either end do not provide a license to crow.

In the first chart, all in-camera dynamic range extenders are set to OFF.
In the second chart, all in-camera dynamic range extenders are set to the maximum possible.

Now, these charts don't necessarily mean that there isn't more in the Fuji or the Nikons than what is shown here. It is possible that in the "right" editor there can be amazing recovery of highlights & shadows, who knows. Still it is all very interesting.

[The uploader is missing and I have no way to post the charts. Sorry about that. I'll try again later.]

No DR extension.


With DR extension.
NOTE: The original chart has been replaced with a corrected chart showing the XPro set to the highest DR extension. I did not originally realize that there were multiple settings for DR on the XPro. Now the XPro looks even better at highlight DR - as if we did not already realize that from Fred's work.
Both the XPro and the D4 go "off the chart" on the right.

[Fred Nirque]

I think you have just demonstrated rather succinctly exactly what I was trying to point out before things degenerated into a debate about my lack of understanding of exposure, processing and camera preference, Andrea. I thought that using pictures might be a good way of illustrating my point, but I guess people have been so influenced by "reviews" and "tests" that an actual real-life photography situation takes second row to graphs or studio setups.

Unfortunately a graph or studio setup rarely duplicates exactly what happens in the field, which is why I generally ignore them as being gospel, but I think my point was that while the D3s sensor is still good, technology has moved on again and a small, cheap camera such as the Fuji now betters it in DR performance in practical use.

So as I said to summarize my initial intro to the post, before it degenerated into a critique questioning a whole lot of things that were irrelevant to what I was noting:

The result made me start to seriously contemplate whether in fact I should be thinking of trading the D3s for the D4

Which I do believe is basically in agreement with your post.

It is also true that no matter what processor or how many sliders you push in whatever direction and whatever order, that sometimes things get blown to a point where there is simply nothing to recover:

D3s


X-Pro1

Edited by Fred Nirque, 26 November 2012 - 22:54 .

[schwett]

I have these Dpreview dynamic range charts for the XPro and some Nikons. Please don't use these charts to beat anyone over the head with! They show that all the cams are very good these days. A couple of bars on either end do not provide a license to crow.

In the first chart, all in-camera dynamic range extenders are set to OFF.
In the second chart, all in-camera dynamic range extenders are set to the maximum possible.

Now, these charts don't necessarily mean that there isn't more in the Fuji or the Nikons than what is shown here. It is possible that in the "right" editor there can be amazing recovery of highlights & shadows, who knows. Still it is all very interesting.

andrea - that's really interesting. could you add the D3s to it? i'm surprised how "low" it's DR is as reported elsewhere considering how great it does in low light.

[willl]

andrea - that's really interesting. could you add the D3s to it? i'm surprised how "low" it's DR is as reported elsewhere considering how great it does in low light.

You should be able to do it yourself, just goto this dpreview.com review and plug in the values yourself - http://www.dpreview....s/nikon-d600/14

Edit: Looks like the D3s isn't one of the available options.

Edited by willl, 27 November 2012 - 02:10 .

[Fred Nirque]

andrea - that's really interesting. could you add the D3s to it? i'm surprised how "low" it's DR is as reported elsewhere considering how great it does in low light.

The D3s DR isn't "low" - it's right up there with the best of the cameras which were new in 2008. It's just that now things have advanced, so there are cameras which are better, largely due to trickery such as in the X-Pro1 where setting "Auto" or DR 400%" invokes the camera to reduce the sensor sensitivity (ISO) in the brightest areas and boost the sensitivity the darkest areas according to how the light hits the sensor. The result of this is some additional additional noise in shadow areas, in fact this can clearly be seen in my poppy field shot taken earlier the same morning as the shots above - just after the sun rose, only blocked by the flower itself, the greenery below was in the deepest possible shadow, which auto DR has boosted by effectively increasing ISO in those areas, and the resultant noise can easily be seen even on the web-size 1400px wide version.

The other thing I like with the Fuji is just how film-like the noise, where it occurs, can be - it looks more like irregular film grain than the usual regimented digital noise.

Without this trickery, short of doing an HDR bracket (not really possible because of the breeze blowing) this would have been rendered almost black. I know this because two years ago I tried a similar shot with the D3s and that is exactly what happened, with the sky and flower properly exposed, the shadows were all but featureless black, even after a valiant effort in ACR. The shot was OK all the same, but looked totally different to what I was after. That shot I turned into a dramatic silhouette because lifting the shadows resulted in far too much noise, and exposing for the mid-tones to reduce that effect ended up really blowing out the sky behind the flower (bold type used to indicate that yes, I did that deliberately, and this is not an attempt to compare processing or how to do it), and is why this time around I chose the Fuji for a different result:

2010, D3s


2012, X-Pro1

[Eb Mueller]

According to DxO mark, there is 1 stop difference in dynamic range, "landscape," in D3s vs D4. It seems to me that graphs I've seen published in NG indicated about the same. So, 1 stop is worth how many $ more?