[argyllcms] Re: White Point

  • From: Elle Stone <ellestone@xxxxxxxxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 21 Sep 2015 16:37:51 -0400

On 09/21/2015 03:19 PM, Hening Bettermann wrote:

Elle Stone:
>Is there a reason to think Hening's camera is saving significantly nonlinear
results in the raw file?
My current camera is a Sony a7r,

FWIW (not much!) my current camera is a Sony a7 and I used "-am" to make a general purpose profile.

and the reason I re-reconsider my camera profile is that I read that the a7r it
is compressing extreme values, so I thought the linearity was in question. But
these extreme values may not fall within the range of the CC patches.

Hmm, yes. I'm not too happy about the compression. But that's a "file size" compression that can produce artifacts under certain circumstances, and can't be fixed with a camera profile:


Unless there's another type of compression also going on with the Sonys (please say no!)?

-am versus -as:

As I understand it, -am presumes linear data, whereas -as does not. Can it do
any harm to use -as, meaning not to make a presumption? Will -as not handle the
data correctly in case they turn out to be linear? I get better Delta E values
with -as.

You could make several profiles, with "-am" and with "-as", using various "-qx" settings, and compare the xicclu graphs vs. the reported deltas. I did that once. With the "-as" type profile, the xicclu graph changed shape depending on the in-camera exposure plus exposure compensation used to interpolate the raw file, which seems not reasonable for a camera input profile. And the higher the quality setting ("-as -qh") the squigglier the graph, which also seems not reasonable for a camera input profile. But I did those experiments awhile back, not sure what would happen with recent versions of ArgyllCMS.

IMHO, low deltas aren't all that important for a general purpose camera input profile. Making a very low delta profile is asking for more accuracy than the target chart+lighting+technique is likely to merit, and you run the risk of introducing odd color shifts.

For something like Ben's art reproduction, it's another matter entirely. But he uses controlled studio lighting and target charts with lots of patches, and I think he's even made some custom target charts for the pigments he photographs - Ben, yes?


Other related posts: