[argyllcms] Re: Camera calibration: LUT only as good as matrix?
- From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Thu, 4 Jul 2013 08:55:28 -0700
On Jul 4, 2013, at 8:46 AM, Iliah Borg <iliah.i.borg@xxxxxxxxx> wrote:
> On Jul 4, 2013, at 11:38 AM, Ben Goren wrote:
>
>> On Jul 4, 2013, at 8:31 AM, Iliah Borg <iliah.i.borg@xxxxxxxxx> wrote:
>>
>>> Imagine blue filter in CFA allows red. Now if we have a lot of red in the
>>> light, the blue response is different, and blue chromaticity is determined
>>> as different.
>>> We are down to filter selectivity issue.
>>
>> As Dr. Berns has demonstrated, this can be exploited if you can make
>> multiple exposures -- some with filters and some without.
>
> It was in use well before his demo, and well before digital too; and of
> course I know a lot of folks who do it routinely, and use it myself a lot.
> But it is a different approach to the whole workflow, and it works best with
> monochrome sensors.
Right, but the advantage to putting filters in front of the lens of a modern
DSLR is that you can get three times as many filter readings from each shot. If
you want nine different channels, that's eight or nine different filters with a
monochrome sensor and nine different exposures...but it's only two filters and
three exposures with a DSLR.
> When multispectral CFA will be available on commercial basis we will enjoy a
> lot of frustration, starting right with demosaicking.
That would be a very welcome frustration to have to deal with, I should think!
...but, if my idea pans out, demosaicing will be of no more concern for me with
the multi-filter workflow than in an unfiltered workflow.
b&
Other related posts:
