Hi,

yes, I also can confirm that RT really excels in highlight 
reconstruction and is ahead of darktable (*). IMHO this has two aspects:

1) RT controls all this from one single panel. So there are only few 
clicks needed to get to the desired result.

2) Color reconstruction of blown-out highlights works very well.

I think the first aspect is a question of UI philosophy. Highlight 
reconstruction involves different steps (dealing with clipped channels 
in pre-demosaic, negative exposure correction, color reconstruction) 
which in the case of darktable involves different modules. I think there 
are good arguments for our way of working.

The second aspect is certainly something we need to work on. RT has a 
method called "color propagation" which seems to work better than our 
"color reconstruction" (which already has been a big step forward).

I am not sure if we can get much better within the current "highlight 
recovery" module - maybe this is job for a separate module that works on 
demosaiced data.

Ulrich

(*) Driven by the example image provided by Istvan I have worked a bit 
on the shadows&highlights module. You can find my work in PR 866. This 
further helps in correcting blown out highlights but still lacks a 
convincing color recovery.

Am 22.02.2015 um 19:28 schrieb Markus Jung:
> Am 22.02.2015 um 19:18 schrieb KOVÁCS István:
>> I've tried dcraw -E (just dump the raw pixels without
>> demosaicing/highlight recovery etc.), then used a 2x2 convolution
>> matrix in the Gimp to join the 2x2 R-G-B-R blocks into a single pixel.
>> The result is at
>> http://tech.kovacs-telekes.org/files/DSC_4684/DSC_4684-dcraw_-E.jpg -
>> lots of highlight details are available, but I have been unable to get
>> usable highlights from either darktable, digikam, dcraw (used -b 0.5
>> to reduce exposure by 1EV, and tried all highlight recovery modes of
>> dcraw).
>>
>> Anyone?
>
> I tried to recover them with darktable, too. The best results have been
> archived by using the new highlight recovery mode "reconstruct color"
> but RawTherapee did a noticeable better job.
> I guess darktable lacks some more enhanced highlight recovery algorithm.
> Maybe one of the devs will notice your example and has some more ideas.
>
> Regards,
> Markus
>

------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
_______________________________________________
Darktable-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/darktable-users

Reply via email to