Yes, as far as I can tell, the 
logic is as you've described below.

Also, one other thing I'd like to do in there ... for mip-maps resized with 
anything other than a box 
filter, I'd like to expose a flag that forces the use of the top-level 
mip for creating ALL lower-level mips (as opposed to successively 
smaller levels being used as the source for the next smaller level - 
which only makes sense when using a box filter). I think this was met 
with some disapproval when I proposed it a year or so ago - but it does 
make sense and I'd like to pursue it.



________________________________
 From: Larry Gritz <[email protected]>
To: Stephen Parker <[email protected]>; OpenImageIO developers 
<[email protected]> 
Sent: Tuesday, June 12, 2012 12:13 PM
Subject: Re: [Oiio-dev] maketx convert to linear before resize
 

How is this different than now?  Is it currently 

inputcolorspace -> outputcolorspace -> resize -> write?   

That would make the resize happen nonlinearly (bad) if the output space were 
not linear.


Makes sense to me.  Jeremy?

Note that you need to do the output color space conversion separately for each 
MIP level as you output it, because every resize needs to happen linearly.



On Jun 12, 2012, at 11:55 AM, Stephen Parker wrote:

I've been maintaining a local modification to maketx that converts data to 
linear colorspace prior to resizing, and then back out to sRGB. At the time I 
wrote it we were close to having OCIO support in OIIO, so I just maintained it 
as a local patch. Now that we have OCIO support, I'd like to change the logic 
of colorspace handling in maketx to:
>
>
>inputcolorspace -> linear -> resize -> outputcolorspace
>
>
>Is this ok? If so, I'll make the changes.
>
>
>
>-s

--
Larry Gritz
[email protected]
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to