Jump to content


Receive a FREE CAMERA BAG from Think Tank Photo

Photo
- - - - -

Highlight recovery


  • Please log in to reply
47 replies to this topic

#21 Andrea B.

Andrea B.

    Lost in the Desert

  • Administrators
  • 11,034 posts
  • LocationNuovo Jersey
  • Edit my pics?:Yes

Posted 27 November 2012 - 03:38

In the end a good shot makes the case, and I'd say the case is amply made with that stunning Poppy shot.
The D4 may squeak out a little more DR with a little less noise when using its ADL mode.
But all we really need is enough DR - which the XPro has.
So Fred, I'm not really sure you need a D4, although it would certainly be nice to have.

You know, I'm *still* interested in the XPro (or possibly the XE1) for use with my remaining Leica 50 Summicron. But I gotta wait until the demosaicing problem is cured.

Andrea B.
UltravioletPhotography.com


#22 Eb Mueller

Eb Mueller

    Advanced Member

  • Life Member
  • 1,185 posts
  • LocationChilliwack, BC, Canada
  • Edit my pics?:Yes

Posted 27 November 2012 - 03:50

In any case, if DR was the priority, then the D800E would be the most efficient. The range, from D3S - D4 - D800E is 12 - 13.1 - 14.3 EV, which makes the D800E a relative bargain. I presume the XPro is even more of a bargain, once conversion software troubles are resolved. BTW, Fred, I attempted a similar photographic idea, (low angle floral against sun,) albeit not with fisheye, but lupines rather than poppies - instead I used a little bit of fill flash to augment the D3S' range. Not the same, but just saying, here. Note the 3rd image.

Edited by Eb Mueller, 27 November 2012 - 03:51 .

Eb Mueller
British Columbia, Canada
http://www.pbase.com/emueller

#23 Alan7140

Alan7140

    Advanced Member

  • Life Member
  • 5,314 posts
  • LocationTasmania, Australia
  • Edit my pics?:Ask Me

Posted 27 November 2012 - 03:57

You know, I'm *still* interested in the XPro (or possibly the XE1) for use with my remaining Leica 50 Summicron. But I gotta wait until the demosaicing problem is cured.


Well, the poppy shot was processed in Helicon Filter which uses dcraw, and while it doesn't do the X-E1 yet that can't be too far away. Helicon Filter is very easy to use, it just lacks some of the ultra-fine finesse of ACR, but the results, as you can see on schwett's thread here: http://www.fotozones..._40#entry371834 are very good indeed compared with the ACR mess.

Helicon filter will do Nikons and Canons etc as well using dcraw if you so specify in the prefs under the RAW tab, so you might want to try the fully functional 30-day trial to see if it's functional enough for you in the interim to Adobe/Fuji coming up with a more professional solution.

#24 yunfat

yunfat

    Advanced Member

  • Life Member
  • 2,821 posts

Posted 28 November 2012 - 01:55

Just get an S5. :good:

I still marvel at my S5 files. I don't care what DX0 or anyone else says, nothing will ever touch that camera for highlight recovery.

As to a D4 or not. I would wait for the D4X considering your type of work fred.

Sometime in April we are gonna get a high megapixel piece from Nikon.

#25 Alan7140

Alan7140

    Advanced Member

  • Life Member
  • 5,314 posts
  • LocationTasmania, Australia
  • Edit my pics?:Ask Me

Posted 28 November 2012 - 02:46

I'm thinking the experience with the S5 has been blended into the design of the X-Trans sensor - the ability to produce colour is on par with the files I've seen from the S3 & S5, so the likelihood of the X-series being in a similar league is likely.

I'm still trying to understand the (as usual) off-the-wall implementation of the DR extension facility - and why I'm holding fire on that graph posted by Andrea which tested the Fuji at "DR100". The DR extender (for want of a better word) only offers 100% at 200 ISO, at 400 ISO the choice is opened to 200%, and at 800 ISO it becomes 400%. Old school thinking says that the DR of a sensor reduces as ISO increases, so at 800 ISO it's obvious that the highlight/underexpose-shadow/boost function would kick into overdrive, but also it is reasonable to assume that with the base ISO set higher there is more leeway for these functions to work in reducing or boosting on either side of a "normal" mid-tone at, say, 800 ISO.

If that's the case, then testing at DR100% would not be giving the greatest DR range.

As I've said, I'm a photographer, not a reviewer, tester or technocrat, so I might be way off beam in my understanding of how this works, but given the weird way in which Fuji designs things anything's possible.

The fact that Hasselblad chose a Fuji camera and lens system to rebadge as their flagship camera line is not lost on me, either.

#26 schwett

schwett

    Advanced Member

  • Life Member
  • 1,171 posts
  • LocationSan Francisco
  • Edit my pics?:Ask Me

Posted 28 November 2012 - 04:51

the more important graph, imo, is the one without the software trickery. no ADL, no dr200%+ etc. it's like trying to assess a sensor's noise characteristics with noise reduction turned on.....
http://photo.sfmthd.org/ [under construction]

#27 Andrea B.

Andrea B.

    Lost in the Desert

  • Administrators
  • 11,034 posts
  • LocationNuovo Jersey
  • Edit my pics?:Yes

Posted 28 November 2012 - 20:04

Fred, here is a nice Dpreview graph showing the XPro's dynamic range at the different settings. Not being familiar with the XPro, I didn't realize there were multiple advanced settings on this DR thing. Looks pretty good to me!! :D

I have gone back and replaced the original chart I posted with an upgraded version showing the XPro set to its highest dyamic range D400. Everyone will note in the corrected chart that the XPro "goes off the chart" on the right just like the Nikon D4 does. It is only on the left (shadows) that the D4 shows somewhat more range.

This chart shows 4 of the XPro's dynamic range modes.
Screen Shot 2012-11-28 at 3.01.40 PM.png

Andrea B.
UltravioletPhotography.com


#28 Alan7140

Alan7140

    Advanced Member

  • Life Member
  • 5,314 posts
  • LocationTasmania, Australia
  • Edit my pics?:Ask Me

Posted 28 November 2012 - 20:28

Thanks for that, Andrea - that better reflects what I've been seeing in actual photographs as far as highlight recovery goes - 400% is the more effective setting.The more "film-like" curve profile and shoulder has been commented on by a few reviewers now, something that the 400% setting confirms, and yet another confirmation of the "filmic" (for want of a better word) results that I've been seeing in photographs I've taken at the higher DR enhancement setting.

It would also tend to explain the outstanding B&W conversion results possible, keeping highlight separation at a reduced contrast ratio while maintaining a strong contrast ratio through the mid-tones to avoid the "muddy greys" (which Carolyn has mentioned on occasion), and tapering off in the shadows again. In fact that curve looks very much like a fully processed very fine grain B&W film curve.

As for the toe of the curve - the extension to the left is pretty meaningless once the curve flattens out - there'll be no further separation of tones in this area. What I can't understand is why the Fuji curve stops at -6 and the others drift further in a straight, toneless line. In film parlance that would indicate a longer threshold enabling the printing of a denser black by exposing that part of the image onto the paper for a longer proportion of time and allowing a greater build-up of silver grains to develop than in an image with a short threshold, but as to how that translates to digital printing I'm not sure as all that would imply is dumping more black ink into that area, but ink having its own D-Max means that this would be of little benefit once that density is reached (i.e 100% black will remain that colour even if multiple layers of 100% black were added to it).

So if that long toe just means less noise in the blacks, it is certainly not a deal breaker and blown highlights to me are a far greater problem than slightly more noise in deep shadow areas.

Edited by Fred Nirque, 28 November 2012 - 21:40 .


#29 schwett

schwett

    Advanced Member

  • Life Member
  • 1,171 posts
  • LocationSan Francisco
  • Edit my pics?:Ask Me

Posted 29 November 2012 - 01:20

Thanks for that, Andrea - that better reflects what I've been seeing in actual photographs as far as highlight recovery goes - 400% is the more effective setting.The more "film-like" curve profile and shoulder has been commented on by a few reviewers now, something that the 400% setting confirms, and yet another confirmation of the "filmic" (for want of a better word) results that I've been seeing in photographs I've taken at the higher DR enhancement setting.

It would also tend to explain the outstanding B&W conversion results possible, keeping highlight separation at a reduced contrast ratio while maintaining a strong contrast ratio through the mid-tones to avoid the "muddy greys" (which Carolyn has mentioned on occasion), and tapering off in the shadows again. In fact that curve looks very much like a fully processed very fine grain B&W film curve.

As for the toe of the curve - the extension to the left is pretty meaningless once the curve flattens out - there'll be no further separation of tones in this area. What I can't understand is why the Fuji curve stops at -6 and the others drift further in a straight, toneless line. In film parlance that would indicate a longer threshold enabling the printing of a denser black by exposing that part of the image onto the paper for a longer proportion of time and allowing a greater build-up of silver grains to develop than in an image with a short threshold, but as to how that translates to digital printing I'm not sure as all that would imply is dumping more black ink into that area, but ink having its own D-Max means that this would be of little benefit once that density is reached (i.e 100% black will remain that colour even if multiple layers of 100% black were added to it).

So if that long toe just means less noise in the blacks, it is certainly not a deal breaker and blown highlights to me are a far greater problem than slightly more noise in deep shadow areas.


i could be wrong here - but i'm pretty sure those dynamic range extension modes are just like nikon's active d-lighting - they are a software adjustment to a slightly different original exposure than would otherwise be selected. that's why the camera wants the flexibility to select a different ISO internally and i suspect that's why fuji locks it out at base ISO.

if you shoot raw, you'll probably see that the ADR 100 200 400 settings have no (positive) effect on the output when compared to simply underexposing by a stop or two and then boosting the exposure later. This is the old school of digital highlight protection. There's now a lot of talk about "ETTR" since the downside of this approach is more noise in the shadows.......
http://photo.sfmthd.org/ [under construction]

#30 Andrea B.

Andrea B.

    Lost in the Desert

  • Administrators
  • 11,034 posts
  • LocationNuovo Jersey
  • Edit my pics?:Yes

Posted 29 November 2012 - 06:05

From experiment, I've confirmed that Nikon's in-camera Active D-Lighting gives a result different than would be obtained by underexposing and then simply boosting exposure in the editor. I've also observed that the in-camera ADL gives a different result than would be obtained by using Nx2's D-Lighting. This in-camera difference is because in addition to the slight underexposure, Nikon also applies a tone curve that lifts shadows. I don't believe that underexposure and tone curves are the only trickery going on with this in-camera ADL. I think there could be some noise masking or highlight recovery going on also.

Now, could I simply underexpose and figure out my own tone curve, noise masking and highlight recovery in an editor that would mimic the in-camera ADL to some extent? Well, probably. But the point is - why should I have to when it is already built into the camera????

I think it would surprise me if Fuji didn't also also have some interesting in-camera trickery going on with its DR modes in the XPro. I have read that their way of extending DR involves ISO, but I haven't yet found any details about other tweaks. Again, why try to extend DR manually in an editor when you can get it "free" in camera? (...said the lazy photographer me...).

The potential downside of underexposure may have been noise back in the "olden days" of DSLRs. But these days when ADL is used in a D4 or D600 or D800, there is little danger of inducing much noise from a small 1/6 - 1/3 underexposure.

BTW, Nikon's ADL in Aperture mode alters speed (of course), but when using Auto-ISO in Manual mode, it is the ISO that altered (similar to the XPro).

To be extra sure of not inducing noise when using Nikon ADL or Fuji DR, I guess we should perhaps also push the exposure to the right without hitting the wall (except maybe in the red channel). Might be overkill, however. And then there would be a lot of work in the editor in the opposite direction to bring all that hdr-like brightness back under control !!

Andrea B.
UltravioletPhotography.com


#31 Ann

Ann

    Professional Photographer

  • Life Member
  • 9,823 posts
  • LocationNew York State
  • Edit my pics?:No

Posted 29 November 2012 - 06:05

If you want greater DR, don't use either in-camera Landscape or Vivid Picture Controls. Better still, just ignore Picture Controls entirely and shoot RAW.

#32 schwett

schwett

    Advanced Member

  • Life Member
  • 1,171 posts
  • LocationSan Francisco
  • Edit my pics?:Ask Me

Posted 29 November 2012 - 06:20

...The potential downside of underexposure may have been noise back in the "olden days" of DSLRs. But these days when ADL is used in a D4 or D600 or D800, there is little danger of inducing much noise from a small 1/6 - 1/3 underexposure...


fuji's version seems to underexpose by two stops at the 400% setting. shadow noise is inevitable at that ISO (at least as i use the camera it is) but perhaps it's worth the price.

as to why you'd do it yourself instead of letting the camera do it........... it only work it's magic on JPGs, just like Nikon's, unless you use capture. i'm not sure which i'd rather be subject to - nikon capture, "silkypix," or a good swift kick in the teeth!
http://photo.sfmthd.org/ [under construction]

#33 Andrea B.

Andrea B.

    Lost in the Desert

  • Administrators
  • 11,034 posts
  • LocationNuovo Jersey
  • Edit my pics?:Yes

Posted 29 November 2012 - 06:33

In a Nikon, both Landscape and Vivid increase contrast globally and thus cut dynamic range. The Neutral and Portrait settings have max DR with the Portrait range having a tad less in the highlights. Picture Controls must be set on any Nikon even when shooting Raw. It absolutely does matter which one you select when shooting Raws because these profiles will have an effect on the exposure you choose.

In the XPro, IIRC, it is the Provia profile which has the most DR - and the Velvia which has the least because of increased contrasts and saturation.

ADDED: Schwett, I think the XPro lowers ISO when set to a DR mode?? Thus noise is not induced. I'm trying to find some details about this, but it is getting late, and I'm tired from cleaning up the last of the hurricane mess in our yard.

[[[ASIDE: I can use *any* converter/editor, but I like NX2. There's still nothing better for bringing out the best in a Nikon file. It is such an *easy* converter that I don't understand why people complain about it. But to each his own! I don't think it is necessary to get into converters here in this thread. FWIW, I've recently learned the Photo Ninja converter, and it is almost as good as NX2 for bringing out detail in a Nikon Raw.]]]

Andrea B.
UltravioletPhotography.com


#34 Ann

Ann

    Professional Photographer

  • Life Member
  • 9,823 posts
  • LocationNew York State
  • Edit my pics?:No

Posted 29 November 2012 - 06:58

I don't use Nikon's Converter; shoot 14-bit RAW; have set my Picture Controls to be as neutralized and unsharpened as possible and then simply ignore them entirely and I have no trouble in getting good exposures by using appropriate metering techniques (often in Manual mode) in every kind of situation from starlight to high noon!

I have no interest in using automated settings or processing of any kind and have found that I can easily process 1000 RAW shots in a day without any need for Picture Controls or Presets or Effects Filters!

#35 afx

afx

    Advanced Member

  • Moderators
  • 5,317 posts
  • LocationMunich, Germany
  • Edit my pics?:Yes

Posted 29 November 2012 - 07:03

Picture Controls must be set on any Nikon even when shooting Raw. It absolutely does matter which one you select when shooting Raws because these profiles will have an effect on the exposure you choose.

Interesting idea.
Can you back it up with a source that proves this?

cheers
afx
"Patriotism is the virtue of the vicious" - Oscar Wilde
"They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety." - Benjamin Franklin
My latest book: The AfterShot Survival Guide  
sRGB clipping sucks and Adobe RGB is just as bad  Still no clue how to take decent pictures though, see afximages.com ;-(

#36 schwett

schwett

    Advanced Member

  • Life Member
  • 1,171 posts
  • LocationSan Francisco
  • Edit my pics?:Ask Me

Posted 29 November 2012 - 07:07

...
ADDED: Schwett, I think the XPro lowers ISO when set to a DR mode?? Thus noise is not induced. I'm trying to find some details about this, but it is getting late, and I'm tired from cleaning up the last of the hurricane mess in our yard.
...


strangely, it's the opposite. if you set the camera to 'auto iso 800' you can activate DR100, 200, or 400. if you set it to iso 200, you can't pick anything but DR100. with iso set to auto 800, i took a photo of a white envelope under a bright light next to a couple black batteries on my desk. the resulting exposures :

at DR100, the camera selected ISO 200, 1/420, f/1.4 (it was on aperture priority)
at DR400, the camera selected ISO 800, 1/1800, f/1.4 (it was on aperture priority).

so, some obvious things. the exposure is pretty close - it's under by less than a third of a stop - but it's absolutely shooting at a much higher ISO. if you set the ISO simply to 800 (not auto 800), you get two quite similar exposures - 1/1800 and 1/1900 - so it's not under or overexposing by as much as is widely reported on the internet!

it's pretty obvious that it's actually shooting at ISO 800 and not performing some magic feat where certain pixels are exposed differently than others. the camera does, after all, have a mechanical shutter.

Posted Image

my conclusion - if one is scientifically comparing the dynamic range of these cameras, you need to do it with all the software trickery turned off. you can achieve the same thing in your raw processor of choice, and with more control.
http://photo.sfmthd.org/ [under construction]

#37 schwett

schwett

    Advanced Member

  • Life Member
  • 1,171 posts
  • LocationSan Francisco
  • Edit my pics?:Ask Me

Posted 29 November 2012 - 07:19

one final note - i believe those dpreview charts look like they are ONLY measuring jpg output (if you go to the various reviews you'll see the explanatory text notes this). thus the impressive results fred has achieved in highlight recovery from the x-pro1 raw file are not at all represented by these charts, nor is the real potential quality of the sensor being measured. what's being measured is the jpg engine, and the trickery it might or might not employ.
http://photo.sfmthd.org/ [under construction]

#38 Andrea B.

Andrea B.

    Lost in the Desert

  • Administrators
  • 11,034 posts
  • LocationNuovo Jersey
  • Edit my pics?:Yes

Posted 30 November 2012 - 01:52

Schwett, yes, the Dpreview charts measure jpegs. They are meant simply to show a *relative* comparison between the various cameras that might indicate how the Raws would play out. I posted them because we do not have any raw XPro DR measurements (such as those provided by DxOMark) to use in such a relative comparison. Anyway, I'm not out to prove anything here. Just trying to find out more because the XPro is an interesting camera. As I already said above, the photos from the XPro make the case well enough.

Andrea B.
UltravioletPhotography.com


#39 Andrea B.

Andrea B.

    Lost in the Desert

  • Administrators
  • 11,034 posts
  • LocationNuovo Jersey
  • Edit my pics?:Yes

Posted 30 November 2012 - 02:42

Andrea B.: Picture Controls must be set on any Nikon even when shooting Raw. It absolutely does matter which one you select when shooting Raws because these profiles will have an effect on the exposure you choose.

AFX: Interesting idea. Can you back it up with a source that proves this?

??
There's nothing to "prove". There is no way to turn Picture Controls to Off. "-)
And the second sentence is about exposure *choice*. Here's an example - when I'm shooting Landscape or Vivid, I usually choose to dial in some -EV because those Nikon reds hit the wall and pile up pretty quickly in those modes. So, in short, the Picture Controls help me see whether I need to override the meter in order to get the final look I want. I find them *very* helpful.




Andrea B.
UltravioletPhotography.com


#40 Alan7140

Alan7140

    Advanced Member

  • Life Member
  • 5,314 posts
  • LocationTasmania, Australia
  • Edit my pics?:Ask Me

Posted 30 November 2012 - 02:50

Until I actually see something coherent that makes sense from Fuji and isn't some badly translated floral marketing spiel, I'll go happily with what those graphs show in the difference between the way the Nikon DR enhancement and the Fuji version is handled, that being, of course, the film-like mid-tone contrast bias with the highlight roll-off in the Fuji DR400% example, as opposed to the common digital preference for maintaining equal tonal separation from threshold to blowout as displayed in the Nikon graphs.
Jpeg or not.

They do confirm what I've been seeing in the results I'm getting, and to be honest, that's all I care about.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users