[argyllcms] Re: "gamma key" Re: Photocamera calibration questions
- From: yan@xxxxxxxxxxxxxxxxx
- To: <argyllcms@xxxxxxxxxxxxx>
- Date: Mon, 23 May 2011 19:42:35 +0400
On Mon, 23 May 2011 12:16:31 +0200, Magnus Wallden wrote:
Of
course, there are a lot of profiling tools. You could use any tool
suitable for scanner.
By "only one" I mean the only one, that could
give normal result, not always ideal, but usable. I use them in
photography.
I have tried dcp2icc, with colorchecker passport
software.
Yes, it possible with dcp2icc utility to make matrix
profile and convert it to table .icc profile. Why not simple "matrix to
matrix" solution? I don't know, ask the author.
The result of this
conversion is ugly, I could send you an example if you want to chek
yourself :-)
> Hi !
>
> Not the only one, qpcard202 whit sw, this
gives as result an dcp file, and it is
> possible to convert this to an
icc whit dcp2icc.
>
>
http://www.qpcard.se/BizPart.aspx?tabId=31&prod=8&catId=1&tci=151 [2]
>
http://dcp2icc.sourceforge.net/ [3]
>
> //Magnus
>
> 2011/5/23
>
>> Hello! This would be answer to question 1
>>
>> I understood, that
coolproof is not the ideal tool to create profiles from raw sensor. In
fact, it's the only one.
>>
>> When you write "by hand" you mean, that
I need to write another tool by myself. Of course, it's kind of an aswer
:-) But this is rather hard thing. And, on the other hand, I think that
only slighter code changes are needed.
>>
>> Coolproof in matrix+gamma
mode reads ".ti3" values and try to
>>
>> 1. Guess gamma values;
>>
>> 2. Find matrix for this gamma.
>>
>> The only key I need is "not
to guess gamma upon ti3 values, and assume gamma is [some value]". The
linearity of camera sensor is quite a good assumption. By analizing this
or that real data you get "gamma 2.18" or "gamma 2.21", and this number
would be not exactly 2.2 only beacause of camera noise.
>>
>>>
yan@xxxxxxxxxxxxxxxxx [1] wrote:> 1. IS IT POSSIBLE TO SET "GAMMA 2.2"
IN INPUT FOR COLPROF UTILITY SOMEHOW, NOT BY MAKING A LOT OF "AUXILIARY
COLOR CHARTS"?
>>> Hi, I'm not quite sure what you mean by this.
colorprof creates a profile that represents the behaviour of the device
as sampled by the .ti3 values. You can't "force" the profile model and
expect it to be a good fit to the samples. There is no point in doing
such a thing - the whole purpose of a profile is to represent the color
behaviour of the device. colprof is not the ideal tool to create
profiles from raw sensor data that's been converted to samples, since it
is doing a "best fit". You are probably better off creating a matrix and
curve model "by hand". Note that you (ideally) need both the spectral
sensitivity data, and the response data for each sensor. The curves
should then translate the sensor responses to linear light, and the
matrix convert to XYZ space
>>>
>>> Graeme Gill.
>
> --
>
> Magnus
Wallden
> Klockarevägen 29
> 331 52 Värnamo
>
> +46 (0) 370 10478
>
+46 (0)703 33 45 19
> magnus.wallden@xxxxxxxxx [5]
Links:
------
[1] mailto:yan@xxxxxxxxxxxxxxxxx
[2]
http://www.qpcard.se/BizPart.aspx?tabId=31|+|amp|+|prod=8|+|amp|+|catId=1|+|amp|+|tci=151
[3]
http://dcp2icc.sourceforge.net/
[4] mailto:yan@xxxxxxxxxxxxxxxxx
[5]
mailto:magnus.wallden@xxxxxxxxx
Other related posts:
