Pages: 1 2 [3]   Go Down

Author Topic: Curves over sliders  (Read 17275 times)

stamper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5882
Re: Curves over sliders
« Reply #40 on: November 11, 2012, 05:31:39 am »

But the images should be low in noise, right?  That's the goal of using HDR, to maximise SNR and thereby reduce noise.

That is one consideration. The best description that I have read was by Harold Davis. I am paraphrasing him.

The goal is to combine two images that can be used as a background layer in Photoshop for further processing and the result is one that can't be processed well without the HDR being used and it shouldn't be obvious that it started off as HDR, as so many examples unfortunately do. In a nutshell the grunge look isn't or shouldn't be the final goal of HDR.

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Curves over sliders
« Reply #41 on: November 11, 2012, 07:57:01 am »

Are you talking about the highlight recovery section of GL's article?  If so, then no it 's not truly recovering the blown channel(s).  It's recreating or replacing data from other information in the image.  This can be done in PS as well with saturation masks.  The fact remains that you cannot truly recover data from a blown channel because there is no data to recover.  And recreating or replacing isn't the same.

Aren't we getting into semantics, Bob.

You were wondering about "highlight recovery" which S. Tindeman states as part of the processing steps to generate a scene-referred rendition, and I pointed to the use of the term to describe the repair of clipped channels (single channels, by emulating the surrounding colors), just the way G. Luijk uses the term and explains it in his DCRAW tutorial.

We may consider finding a different name while the missing information is not the Raw capture, however, said "highlight recovery" still fits to the idea of scene-reconstruction while the information was in the scene.

Notwithstanding the above there are (of course) Photoshop techniques which attempt to do the same, as a kind of late correction.  Aside from mentioned use of a Saturation mask for targeted adjustments, there is a quite cool technique called "The Impossible Retouch" (second example on the skin tones of the two girls). Although it requires changing to Lab mode, which will certainly distract some fellows around here from adopting it.

Peter

--
« Last Edit: November 11, 2012, 08:01:50 am by Peter_DL »
Logged

RFPhotography

  • Guest
Re: Curves over sliders
« Reply #42 on: November 12, 2012, 08:29:10 am »

That is one consideration. The best description that I have read was by Harold Davis. I am paraphrasing him.

The goal is to combine two images that can be used as a background layer in Photoshop for further processing and the result is one that can't be processed well without the HDR being used and it shouldn't be obvious that it started off as HDR, as so many examples unfortunately do. In a nutshell the grunge look isn't or shouldn't be the final goal of HDR.

I think that's a fairly limiting definition of what HDR is and I don't think injecting his personal opinion about the 'look' of the image is warranted for an accurate definition.

Quote
Aren't we getting into semantics, Bob

No, I don't really think so.  But it depends on how broad any one person is willing to make a definition, I suppose.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Curves over sliders
« Reply #43 on: November 12, 2012, 11:51:00 am »

It may not be scientifically or mathematically exactly the same as luminosity, but we are playing horseshoes here, aren't we

Sure seems that way to me, after starting with post 1 here and moving on....

The raw is never affected so what you end up with is either ideal for you or maybe you need to use another converter. This idea of trying to get "scene referred" data out of a converter then use Photoshop or another converter seems to be a boat-load of work. I sure wish someone would post a raw file, some exact settings and procedures such that I could see that setting say ACR/LR one way to produce this "scene referred" TIFF and subsequent edits in Photoshop provide better/faster/cheaper results then just moving the sliders in the converter.

If we take the same raw file in ACR and another converter, and we set the two such that someone says they are both scene referred, do the two converters thus provide a visual and numeric match to each other?
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

stamper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5882
Re: Curves over sliders
« Reply #44 on: November 13, 2012, 04:13:47 am »

Quote BobFisher.

I think that's a fairly limiting definition of what HDR is and I don't think injecting his personal opinion about the 'look' of the image is warranted for an accurate definition.

Unquote

Bob it is all about personal opinion. The vast majority of the opinions that I have read on here - and other sites - is that the grunge look isn't liked and it gives the process a bad reputation. The photographers who do use it properly tend to get ignored because as stated their use of it isn't obvious. What is your definition?

RFPhotography

  • Guest
Re: Curves over sliders
« Reply #45 on: November 13, 2012, 07:16:32 am »

It would depend on whether we're talking about a very technical definition or a looser one.

From a very technical standpoint - and speaking strictly with respect to photography - I'd say that HDR a process which combines two or more images made at different exposure settings in specialised software to create a 32 bit high dynamic range, intermediary image, in order to maximise signal to noise ratio across the brightness range of the image.

What you won't see in that definition is any mention of tonemapping or any other process to bring the file back to something useable because that useable image is no longer HDR.

In a less technical sense, and again speaking strictly with respect to photography, I'd say that HDR is the combination of two or more images made at different exposure settings with the purpose of creating a final image that has a broader range of brightness than the camera is capable of capturing in a single image and optimising the signal to noise ratio across the brightness range of the image.

Again, no mention of what that final image looks like.  Why?  Because that's an artistic choice.  I don't like the very overprocessed look much either, although I will use it to some degree depending on the subject matter.  But it artistic choices have nothing whatever to do with what an hdr image is or isn't.

Some would say that newer cameras like the D800 with its 'tested' 14+ stops of drange are also HDR.  I disagree with that position for a couple of reasons.  First, I think that inherent in the process is that you're capturing a wider range than the camera can capture 'normally'.  If you're working with a single exposure then I believe you're working with the 'normal' brightness range of the camera.  Second, if you're working with a single exposure you're not optimising SNR across the brightness range of the image.  You're still going to be dealing with greater evidence of noise in the shadows which will be made more visible if those areas are pushed around much at all.
Logged

stamper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5882
Re: Curves over sliders
« Reply #46 on: November 13, 2012, 07:55:09 am »

Bob I don't see much difference between. The first is my quote and the second yours.

The goal is to combine two images that can be used as a background layer in Photoshop for further processing and the result is one that can't be processed well without the HDR being used.

and

I'd say that HDR is the combination of two or more images made at different exposure settings with the purpose of creating a final image that has a broader range of brightness than the camera is capable of capturing in a single image and optimising the signal to noise ratio across the brightness range of the image.


However over the years of reading about HDR - I have two books - I can't remember the noise issue being taken into account. Is this an issue that possibly concerns you and possibly not others? :)

RFPhotography

  • Guest
Re: Curves over sliders
« Reply #47 on: November 13, 2012, 08:17:19 am »

Well, you're being selective in your comparison.  The part you left out is the part where that other definition continues on to opine about the processing of the images and what one should or, rather, shouldn't do in generating the final result.  I'd also disagree with the idea that the goal is (always) a single background layer in Photoshop.  That, again, goes into the editing of the image.  But to that end, I'd suggest that creating a single background layer in PS is likely not the goal in a lot of cases.  Rather I think a lot of people would create the combined image via layers and masks so that each layer can then be edited separately as may be needed to optimise the final result.  Sorry, but I see large differences between the two.

As far as noise, look back earlier in this thread.  Bart and I both brought up the idea of maximising SNR with HDR.  Is noise an issue just for me?  I'd hardly think so.  Truth be told, I'm probably one of the more tolerant of image noise on this board.  Unless one is going for a specific 'look' then maximising SNR is the goal of pretty much any photograph.  It's the whole reason ETTR came into vogue.  HDR - whether the 32 bit form or the (manual or automated) blended image form - is, for all intents and purposes, an extreme form of ETTR. 

Logged

stamper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5882
Re: Curves over sliders
« Reply #48 on: November 13, 2012, 08:51:18 am »

Quote

It's the whole reason ETTR came into vogue

Unquote

I think Michael's original article didn't emphasis the noise issue, only capturing as much information as possible?

RFPhotography

  • Guest
Re: Curves over sliders
« Reply #49 on: November 13, 2012, 08:59:07 am »

Quote

It's the whole reason ETTR came into vogue

Unquote

I think Michael's original article didn't emphasis the noise issue, only capturing as much information as possible?



That's the whole reason for doing it!  More information, more signal, less evidence of noise.
Logged

jrsforums

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1288
Re: Curves over sliders
« Reply #50 on: November 13, 2012, 10:08:49 am »

Quote

It's the whole reason ETTR came into vogue

Unquote

I think Michael's original article didn't emphasis the noise issue, only capturing as much information as possible?



I believe Michael was taken to task on this by some of the sensor technocrats.  I think he now supports the fact of improved SNR.
Logged
John

RFPhotography

  • Guest
Re: Curves over sliders
« Reply #51 on: November 13, 2012, 11:02:19 am »

I don't think that Michael not mentioning the benefit of reduced evidence of noise by using ETTR in his original, now 9 year old, article can be taken as a suggestion that he doesn't think it's important.  But I also don't think that's what he did.  Very clearly the headline on the article is "Maximizing S/N Ratio in Digital Photography".  I don't see how that can be taken as anything but recognising the benefits of ETTR wrt reducing the evidence of noise.  He discusses this explicitly in the section of the article titled "The Lesson".  Further, in the section titled "A Test" he discusses what could/should be seen by close examination of shadow areas between a 'normal' exposure and an ETTR exposure. 

I'm just not sure how it's possible to read that article and not come to the conclusion that the benefit if ETTR is increasing the SNR and reducing the evidence of noise.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Curves over sliders
« Reply #52 on: November 13, 2012, 06:39:48 pm »


It's the whole reason ETTR came into vogue

Unquote

I think Michael's original article didn't emphasis the noise issue, only capturing as much information as possible?


I believe Michael was taken to task on this by some of the sensor technocrats.  I think he now supports the fact of improved SNR.

Yes, I remember those discussions well. In the original article, Michael spent an inordinate amount of time discussing the number of levels in the brightest stop of a digital capture, and in the ensuing flame wars Michael's sidekicks (Schewe and Fraser) stressed the number of levels in the brightest f/stop of an exposure as a rationale for ETTR, suggesting that improved tonality would ensue from ETTR. This evoked a reasoned scientific response by Emil Martinic in his masterful treatise on noise and dynamic range, where he specifically referred to Michael's article and pointed out that the number of levels has little to do with the rationale of ETTR but that the SNR is what is improved and the benefit in in the shadows more than the highlights.

That is all remote history (in terms of digital imaging) and by now I think that all agree on the SNR rationale. In any event, we owe Michael gratitude for introducing the concept of ETTR to many of us.

Regards,

Bill
Logged

jrsforums

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1288
Re: Curves over sliders
« Reply #53 on: November 13, 2012, 09:11:32 pm »

Yes, I remember those discussions well. In the original article, Michael spent an inordinate amount of time discussing the number of levels in the brightest stop of a digital capture, and in the ensuing flame wars Michael's sidekicks (Schewe and Fraser) stressed the number of levels in the brightest f/stop of an exposure as a rationale for ETTR, suggesting that improved tonality would ensue from ETTR. This evoked a reasoned scientific response by Emil Martinic in his masterful treatise on noise and dynamic range, where he specifically referred to Michael's article and pointed out that the number of levels has little to do with the rationale of ETTR but that the SNR is what is improved and the benefit in in the shadows more than the highlights.

That is all remote history (in terms of digital imaging) and by now I think that all agree on the SNR rationale. In any event, we owe Michael gratitude for introducing the concept of ETTR to many of us.

Regards,

Bill

Well said Bill. 

Michael realistically set us on the path of the digital (RAW) zone system, where the important tone is lightest significant tone.  This is exposed where it will just barely not clip, then let all the other tones fall where they will.....adjust to taste in the raw converter.

John
Logged
John

donbga

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 454
Re: Curves over sliders
« Reply #54 on: November 13, 2012, 09:19:05 pm »

I do however do go the the Lens Correction panel and pick my lens presets first;
 

According to a well known LR guru one may wish to delay applying Lens Correction until the last step. When Lens Correction is turned on there is a heavy computational penalty each time the image is redrawn.

Best,

Donbga
Logged

RFPhotography

  • Guest
Re: Curves over sliders
« Reply #55 on: November 13, 2012, 10:20:24 pm »

According to a well known LR guru one may wish to delay applying Lens Correction until the last step. When Lens Correction is turned on there is a heavy computational penalty each time the image is redrawn.

Best,

Donbga

Adobe designed the top down workflow in the Develop module for a reason.  It's the order that the steps are recommended to be completed.
Logged
Pages: 1 2 [3]   Go Up