[Bf-committers] Color correction in sequencer.

[Xavier Thomas]

Hi, its me...  again

It seems after all that "offset, gamma, gain" can be concerted to "lift,
gamma, gain" to get same result so do_version will do fine for bug
correction.

For UI changes I will wait to get more in touch with blender code.


2008/12/11 Xavier Thomas <xavier.thomas.1980 at gmail.com>

> they are not really intuitives, but they works like "Levels" in gimp
> Lift must set the black level
> Gain the white
> and Gamme the mid point
>
> The problem here is that the the values are manipulated for :
> - "1,1,1" to be "no change value"
> - Values between 0 and 1 to darken the image
> - An "inverse" button tho permit values > 1 wich brighten the image
> More than being unintuitives, it does not permit full control (inversing
> gain invers for all channels)
>
> What solution:
> - Curves will take big UI
> - Levels maybe also
>
> But it is not preferable to do this as an sequencer effect, This will
> permit the user to choose the effect acording to the type of correction he
> wants to make (Level, curves, ...)
>
> For primary color correction this is not really important as for me
> "Levels" and "Lift/Gamma/Gain" are the same with different way of setting
> the parameter and they can be done with curves.
>
> But if complex secondary collor correction (like changing saturation  and
> value acording to hue with a curve, ...) is going to make is way in blender
> the user must choose the one he wants
>
> Note that with nodes doing a "not so complex secondary color correction"
> (like the exemple below) requires:
> -color space conversion
> -doing a difference map with a constant value
> -using this map as factor for color
>
> It will need about 10 nodes and that is uaualy done in one in shake or with
> one effect in finalcut.
>
> I am interested in this kind of stuff and I can code it but as I said if is
> going to be used by nodes, sequencer and image editor ther must be a way to
> unify it.
>
> xavier
>
> 2008/12/10 Campbell Barton <ideasman42 at gmail.com>
>
> On Thu, Dec 11, 2008 at 9:33 AM, Xavier Thomas
>> <xavier.thomas.1980 at gmail.com> wrote:
>> > Yes, that is what I ment but is it ok to add an UI parameter "ofset"
>> with is
>> > button "Inv ofset" at the side of the Lift gamma and gain parameters
>> because
>> > the UI will get really charged.
>> >
>> > And what about an image treatment library, because I already have made
>> this
>> > function in my keyer node and crating a color correction node with these
>> > controls might be usefull also: this will result in a lot of code
>> > duplication.
>> >
>> >
>> > 2008/12/10 Dalai Felinto <dfelinto at gmail.com>
>> >>
>> >> Hi Xavier.
>> >>
>> >> Can't you do a do_version()?
>> >> In readfile.c you can see a lot of parameters that need to be updated
>> >> over Blender version.
>> >> If what you need is a simple offset for old blender files work in the
>> >> same way as new blender files (with this bug fixed), this should work.
>> >>
>> >> I didn't see the bug or this part of the code, but do_version()
>> >> sometimes is the better way to go.
>> >>
>> >> Cheers,
>> >> Dalai
>> >> http://blenderecia.orgfree.com
>> >>
>> >> 2008/12/10 Xavier Thomas <xavier.thomas.1980 at gmail.com>:
>> >> > Hello,
>> >> >
>> >> > Lately, I was trying to implicate me in blender developement,  and
>> >> > begining
>> >> > with small things (after the keyr node) I  desided to correct this
>> bug:
>> >> >
>> >> >
>> http://projects.blender.org/tracker/index.php?func=detail&aid=18010&group_id=9&atid=125
>> >> >
>> >> > Basicly the lift parameter of color correction in sequencer works
>> more
>> >> > like
>> >> > a "value" color correction than a "lift" (exept that it work on RGB
>> >> > channel
>> >> > separately so I think "ofset" is more apropriate then "value")
>> >> >
>> >> > Fixing it would be simple but old files that use this parameter will
>> >> > have a
>> >> > diferent render result, and after discussing with Matt Ebb on IRC it
>> >> > seems
>> >> > that this is not acceptable.
>> >> > So the only solution to have a proper color correction in the
>> sequencer
>> >> > is
>> >> > to add an "ofset" parameter. Is this OK ? (maybe the UI will get too
>> >> > compact)
>> >> >
>> >> > Further, why do not put this kind of function in an image library so
>> it
>> >> > will
>> >> > be accesible form other part of the program (nodes and maybe image
>> >> > editor)
>> >> > like the quick blur function in the nodes CMP_util.h
>> >> >
>> >> > Sorry to disturb, I know lot of you are really buzzy but I prefer to
>> get
>> >> > good directions before coding.
>> >> >
>> >> > Thanks
>> >> >
>> >> > Xavier
>> >> >
>> >> > _______________________________________________
>> >> > Bf-committers mailing list
>> >> > Bf-committers at blender.org
>> >> > http://lists.blender.org/mailman/listinfo/bf-committers
>> >> >
>> >> >
>> >> _______________________________________________
>> >> Bf-committers mailing list
>> >> Bf-committers at blender.org
>> >> http://lists.blender.org/mailman/listinfo/bf-committers
>> >
>> >
>> > _______________________________________________
>> > Bf-committers mailing list
>> > Bf-committers at blender.org
>> > http://lists.blender.org/mailman/listinfo/bf-committers
>>
>> Recently I tried making sense of the sequencer's color functions and
>> found them quite confusing. Even after tweaking them for 5min I don't
>> really understand what they do exactly.
>>
>> Id much prefer convert these values into either, gimp/photoshop color
>> 'Levels' or 'Curves'.
>> Curves are already used in other parts of blender so I'ts probably not
>> that hard to make this change,
>>
>> Peter, if your reading, Could these settings be converted to curves?
>> (even a close approximation would be ok I think)
>>
>> --
>> - Campbell
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-committers
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/bf-committers/attachments/20081211/7f091d7e/attachment.htm