I don't thinkg that's how Darktable works... Darktable doesn't clip
highlights in its pipeline...
Take a look at these screenshots. The image passes through the Exposure
module first, then through the Tone Curve.
1) Starting poing. Exposure 0, Default linear tone curve
https://www.dropbox.com/s/j2apphtvbny6qvd/Darktable_1.png
2) I push the exposure to 4ev. It "burns" almost the whole image
https://www.dropbox.com/s/4inix9hv0l6k34m/Darktable_2.png
3) I use the Tone curve to reduce the "range". You see that the image looks
like 1), but internally, it was burnt at some point.
https://www.dropbox.com/s/uv29enu2l436vcu/Darktable_3.png
So, the point of highlight reconstruction is to recover burned data in the
RAW file. Darktable won't loose information, at least not unless you do
some really crazy things.
On Thu, Jun 6, 2013 at 6:14 PM, Chris Siebenmann <[email protected]> wrote:
> | Burned highlights is bad, because it means that information is
> | lost. There are methods for reconstructing that information
> | back. Rawtherapee, which I just installed to test this, has a method
> | called "Color propagation". From what I understand, it uses surounding
> | pixels in the burned area to reconstruct the burned components.
> |
> | For example, this is a sample picture I tested with.
> [...]
> |
> | CR2: https://www.dropbox.com/s/fkzymeor14ypmfx/IMG_0450.CR2
>
> This is a RAW file with clipped data in one or more channels. Clipped
> data cannot be recovered (the data is gone) but in some situations it
> can be invented via interpolation from other available data.
>
> Data that is intact in the RAW file but that is then blown in
> postprocessing is different. The original data is there in the RAW and
> is simply being curved or processed away. That means that it can be
> processed *back* and fully recovered, at least in theory; the challenge
> is to do it easily without huge amounts of manual manipulation and also
> to do it in a way that visually natural and appealing.
>
> (For example, going to a linear base curve will recover most or all
> of those blown highlights in a single action but generally makes your
> picture look terrible. Going to a linear base curve with this CR2 shows
> that, well, the data is gone, it's hard-clipped.)
>
> I am specifically interested in recovering highlights that are intact
> in the RAW but blown in post-processing. Interpolating RAW-clipped data
> is a hard enough problem that I am happy for the darktable developers to
> say 'well, there's nothing we can really do to help except possibly make
> the area go white instead of having wacky colour casts'[*].
>
> (The current git tip darktable actually gives the clipped area a colour
> cast by default but then curves it away to nearly pure white. As usual,
> you can see this clearly by looking at the pre-base-curve version
> through the history stack.)
>
> - cks
> [*: I would be happy to see a highlights reconstruction option for 'if
> any channel is clipped, set the pixel to white'. I might even turn
> it on all the time.
> ]
>
--
http://www.julianmenendez.es
------------------------------------------------------------------------------
How ServiceNow helps IT people transform IT departments:
1. A cloud service to automate IT design, transition and operations
2. Dashboards that offer high-level views of enterprise services
3. A single system of record for all IT processes
http://p.sf.net/sfu/servicenow-d2d-j
_______________________________________________
Darktable-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/darktable-users
