Yes, agreed, I was having the same ideas while playing with it. It looks
like the DNG has different clipping values for each channel, and LR seems
to automatically fill in the information better that DT's highlight
reconstruction module in this case.
It seems your best option is to desaturate the picture, maybe use a mask to
limit the effect to the sky or to the highlights.

Check the xmp attached. It's a very quick test but I think it's promising.
The idea is to:
- disable highlight reconstruction
- change exposure to avoid clipping
- use color channel mixer to copy the red channel to the green and blue,
but using a parametric mask to limit the effect to the blown out sky

The last step can be done also using B&W or color correction modules (the
difference being the position in the pipeline). And the mask can surely be
perfected. But I'm short of time and happy enough that it worked :)

Best regards,
Guillermo


On Sun, Apr 8, 2018 at 10:19 PM, Chris <bydersjo...@gmail.com> wrote:

> Fascinating! So dt is clipping somewhere somehow, or the dng format 'pro'
> adobe somehow? :p
>
> So, just taking this further and assuming the detail I was looking for was
> in the red channel.. I was able to desaturate the reds... but obviously
> this would not work if I had someone wearing red in the photograph. Proof
> that the detail exists though, albeit in an odd channel for clouds.
>
> https://www.dropbox.com/s/slbor6tg6ymzd73/darkt_07.jpg?dl=0
>
> Chris
>
> On 2018/04/08 06:16 PM, Peter Cripps wrote:
>
> And, 'for the sake of science', I downloaded the .dng and opened it in Lr
> and dt.
>
> The histograms are quite different, even allowing for log vs linear on the
> y-axes. Lr shows a bunch of high intensity pixels on the far right hand
> side; these are missing in dt. This is before applying any sliders.
>
> Again, I have seen similar results on several other images.
>
>
>
> On 04/08/2018 05:54 PM, Chris wrote:
>
> Hi Guillermo,
>
> For the sake of science here is the .dng file I was using for both the DT
> & LR examples earlier.
>
> https://www.dropbox.com/s/cavd1o7z8r5ao0n/_MG_1496.dng?dl=0
> Busy playing with it now paying more attention to the histogram.
>
> Chris
>
>
>
> ____________________________________________________________________________
> darktable user mailing list to unsubscribe send a mail to
> darktable-user+unsubscr...@lists.darktable.org
>
>
>
> ____________________________________________________________________________
> darktable user mailing list to unsubscribe send a mail to
> darktable-user+unsubscr...@lists.darktable.org
>

____________________________________________________________________________
darktable user mailing list
to unsubscribe send a mail to darktable-user+unsubscr...@lists.darktable.org

Attachment: _MG_1496.dng.xmp
Description: Binary data

Reply via email to