Chris; It appears odd to me that your image, that has so much highlight detail in the raw, should get 'blown' by the default base curve. It is relatively easy to correct using the tone-curve and setting the base-curve to lineal .... but that is hardly the point is it.
David On 13-06-06 11:40 AM, Chris Siebenmann wrote: > | I don't thinkg that's how Darktable works... Darktable doesn't clip > | highlights in its pipeline... > > My experience is that darktable will definitely blow highlights in > the final rendered image when they are not blown in the original > RAW. Depending on the image this may be blown in only some RGB channels > or in all of them. I think (but have not verified) that this process > does not distort colours (apart from sometimes taking colours to white) > but it does lose detail. > > (Since darktable uses high-precision floating point internally, I assume > that these values are never clipped during processing and thus remain > theoretically recoverable with the right manipulation.) > > If you want to see this, I'll point to my sample NEF: > > http://www.cs.toronto.edu/~cks/tmp/darktable/DSC_1101.NEF > > Look at the difference in output RGB values and detail in the lead > policeman's yellow jacket in a version with and without the base curve. > With the base curve applied, the darktable color picker says that spots > go to, eg, 255/255/122 when in the pre-base-curve version the same spot > is at 223/187/75. This wipes out a chunk of detail and makes the result > look glaringly bright. > > (I believe that part of the overall confusion about this is that many > RAW processors use 'highlight recovery' to refer to two different > processes: dealing with clipped channels in the RAW and 'recovering' > things that were unclipped in the RAW but which have become blown out > in processing. I believe that Darktable's 'highlight recovery' module > is specific to clipped RAW channels, which we can see from its very > early place in the processing pipeline; it takes effect even before > demosaicing[*].) > > - cks > [*: ... although after white balance is applied, which is its own issue, > since in theory white balance multipliers can push a channel over what > would normally be blown. If you want to see a close approximation > of the RAW channel data you need to manually set the white balance > multipliers to 1.0 (aka 'UniWB'). If you want to do this kind of > examination you may be better off with a program that looks at > the RAW file from a more numerical perspective. One such one that I > know about is Rawdigger, http://www.rawdigger.com/; it's a Windows/Mac > program but I've successfully run it under Wine in the past. I don't > know if there's a handy Linux equivalent. > ] > > ------------------------------------------------------------------------------ > 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 > Darktable-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/darktable-users > ------------------------------------------------------------------------------ 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 Darktable-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/darktable-users
