Re: [Lcms-user] Inverting a LUT?

[Gerhard F. <nos...@gm...>]

Gerhard Fuernkranz schrieb:

> Mike Russell schrieb:
>
>> Marti wrote:
>>
>>> That is indeed a very difficult topic. For properly inverting a LUT 
>>> you have to know the gamut boundaries, and then proceed differently 
>>> inside or outside gamut. ..
>>>
>> Thanks for your answer. This explains why some profiles have one 
>> direction, and then just include TRC's for the other direction.
>>
> For input profiles, the ICC spec does not requre the presence of B2A 
> tables, since it is not assumed that a scanner profile is ever used 
> for PCS -> device transformations. But LUT-based display and output 
> profiles should contain both, A2B and B2A tables. If they don't, then 
> they actually do not comply with the ICC spec. Of course, if a display 
> or output profile is only 

... used for PCS -> device transformations, then the A2B tables won't be 
used.

(sorry, I must have accidentally deleted some text before sending the 
message).
-Gerhard

>> Is this a generally unsolved problem, or do proprietary solutions exist?
>>
> I fully agree with Marti, that this is not an easy task. But 
> basically, any profiler must somehow solve this kind of problem. If 
> you want to look at an example, see http://www.argyllcms.com/
>
> Regards,
> Gerhard