Pages: [1]   Go Down

Author Topic: Raw histograms  (Read 3011 times)

wolfnowl

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5824
    • M&M's Musings
Raw histograms
« on: December 09, 2012, 08:53:08 pm »

Photographer Jim Kasson has a blog called 'The Last Word'; starting on December 4th he began a series of posts on getting 'true' histogram matches.  I know enough to understand what he wrote, but probably not enough to comment on it with any authority.  Thought it might generate some (hopefully civil) discussion here. There are about a dozen posts on the topic so far.  The oldest post is 'Raw histograms' which may be found here: http://blog.kasson.com/?m=201212&paged=2 and the others move forward in time from there.

Mike.
Logged
If your mind is attuned t

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Raw histograms
« Reply #1 on: December 10, 2012, 04:03:29 am »

UniWB, raw histograms, matching LR/ACR to raw histogram - everything was discussed so many times here and there... so what he is trying to do exactly ? replace google search ?
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Raw histograms
« Reply #2 on: December 10, 2012, 12:00:01 pm »

UniWB, raw histograms, matching LR/ACR to raw histogram - everything was discussed so many times here and there... so what he is trying to do exactly ? replace google search ?

Agreed. There are a few areas we could nitpick but nothing said there hasn't been discussed in better/greater detail on LuLa.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

bill t.

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3011
    • http://www.unit16.net
Re: Raw histograms
« Reply #3 on: December 10, 2012, 07:45:32 pm »

I kinda like the f-stop quantified graphs.  Would be nice if LR made available a tick mark a zero offset in the histogram data so I could keep an eye on what I was actually doing in the field.  Would add an absolute reference to something that is now presented as completely relative.
Logged

NikoJorj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1082
    • http://nikojorj.free.fr/
Re: Raw histograms
« Reply #4 on: December 11, 2012, 03:45:47 am »

I kinda like the f-stop quantified graphs. 
Me too... but we definitely need those in camera rather than in post, don't we? (3 years later, almost 10 years since Michael's original article, and still nothing but a cloud of dust in the sun and the green grass).

Btw, the article linked in the firtst post is nice in that it tries to synthetize the matter, but still gives me a feeling to try to reinvent warm water (old french saying for rediscovering well-known things, as mixing hot and cold makes warm water, and spot metering can be used for ETTR as long as it's pointed on a not too saturated object).
Logged
Nicolas from Grenoble
A small gallery

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Raw histograms
« Reply #5 on: December 12, 2012, 08:59:47 am »

I kinda like the f-stop quantified graphs.  Would be nice if LR made available a tick mark a zero offset in the histogram data so I could keep an eye on what I was actually doing in the field.  Would add an absolute reference to something that is now presented as completely relative.

That would be nice, but LR with PV2012 is ill suited for that task because of its image adaptive processing with automatic highlight recovery--blown highlights may appear OK because they have been recovered. Furthermore, LR uses a BaselineExposure offset. It is +0.5 EV for the Nikon D3 and +0.35 EV for the D800. One must decrease the exposure in LR/ACR by this amount.

One can get a fair approximation of the raw histogram by using PV2010 with a linear tone curve (sliders on main tab set to zero and a linear point curve). Still, white balance has been applied, and blue or red highlights that are intact in the raw file may be clipped when WB is applied in LR/ACR. One can use UniWB (white balance multipliers all set to 1.0) to circumvent this problem.

As an example, here is a shot of a pink flower shot under florescent lighting with an approximate CCT of 2950K. The red channel is slightly short of clipping. The LR histogram using default settings with PV2012 is shown.



The raw histogram shows that the red channel is slightly more than 2 stops short of clipping. The spike in the histogram is from the WhiBal card included in the shot. The WB multipliers are 1.289 for the red and 2.176 for the blue, according to RawDigger.



To approximate the raw histogram, one can use PV2010 with a linear tone curve and apply the BaselineExposure correction of -0.35 EV. For WB, the  values from an exposure with UniWB from another shot can be applied. The results are shown.



The X-axis scale of the LR histogram is not well documented, but one can get an approximate value of how far the highlights are from clipping by increasing the exposure in LR so that the highlights begin to clip and note the difference in the exposure value. The result is 1.8 EV, roughly in agreement with the RawDigger.



This is perhaps reinventing warm water, but the results may be of interest to some forum members. Likewise, I found the articles deference by the OP to be useful. They collect in one place information that may be scattered about on the internet or in the LuLa forums.

Regards,

Bill
Logged

bill t.

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3011
    • http://www.unit16.net
Re: Raw histograms
« Reply #6 on: December 13, 2012, 02:56:53 am »

Bill, thanks for that well prepared and very interesting post!
Logged
Pages: [1]   Go Up