Re: tuning ambient occlusion dark/bright input using a custom curve?

["Tim Leydecker" <BauerOink(at)gmx.de>]

> Good luck for your work, it looks promising  !

Thanks for the thumbs up. That´s what I´m trying to make
myself believe in on a daily basis. Oh, no. For years now...

The funny thing about all this is the repetitiveness.
I went through all this occluded/photorealism stuff
years ago allready but can´t remember how I did it.

http://www.hafenlola.com/images/cubes_large.jpg
(an image from my homepage, nothing new really)

Did that one using Jeremy Pronk´s ambient occlusion shaders
(~dirtmap from Daniel Rind) using Maya, maybe I should at
least dig out the compositing...no shameless plug intended, btw.

Cheers

tim

P.S: Thanks again for the great help. I´ll try to remember it.


----- Original Message ----- 
From: "guillaume laforge" <guillaume.laforge.3d(at)gmail.com>
To: <XSI(at)Softimage.COM>
Sent: Sunday, April 30, 2006 7:44 PM
Subject: Re: tuning ambient occlusion dark/bright input using a custom curve?


Forget my last post with mixing two occlusions shaders. Of course it doesn't
work the same as only one "Sampled
environment" occlusion shader as the environment color won't spread the same
with a super small  "maximum distance" :-p.

Good luck for your work, it looks promising  !

Cheers.


On 4/30/06, Tim Leydecker <BauerOink(at)gmx.de> wrote:
> Hi Guillaume,
>
> thanks for the extra tips! I`ll probably best describe what I´m
> currently doing - should be easier to understand why I desperately
> seek to preserve as much tonal range as possible.
>
> I´m rendering a SSS pass with one (it´s several later) strong spot
> casting pretty soft area shadow, had originally used FG to get
> bouncelight in the areas not directly lit. Great results. In a nutshell:
>
> I take that SSS passand put a black2mid grey occlusion layer ontop,
> in  *difference mode*. Sounds weird, looks weird but suddenly
> comes great when you add another of the first SSS layer ontop,
> this time masked with a normalmappass (taking, the two directions
> opposite the camera) and then using the third normaldirection, the
> one rectangular to camera as a linear add (~35%) to simulate rimlight.
> Basically, the normalmapstuff is a placeholder for the scenelights that
> will be added later. I put my flat colors inbetween (often using mult)
> to get an idea of the look and mood. Since I still have no reflections
> and hardly any specular (there may be some in the SSS), my tonal
> range will be pretty much flat in the upper RGB range, ~110-255.
>
> I had posted the picture before, first one was posted before:
> http://www.hafenlola.com/downloads/Ley_buddy72.jpg (FG)
>
> Unfortunately I have ceramic tiles in the backround, which is
> becoming a problem for FG and AA, I´ve now spent the last three
> days optimizing the mesh (changing the curvatures ) to get smooth
> but not too pronounced results when swapping out FG with ambient
> occlusion, finally got it but still need to make better use of the
> tonalrange
> for the shaded areas that shouldn´t become *too ambient* ...
>
> http://www.hafenlola.com/downloads/Ley_buddy92.jpg
>
> (uff, I hate ceramic tiles as much as flickering now, ignore the
> glitches).
>
> So, simultaneously, I fiddled with a simple phong to get the areas
> not directly lit and use that now to drive the influence of the
> ambient occlusion, using a sampled environment to light up
> the shaded areas based on their luminance values.
>
> To make a long email short, I use subtile occlusion to describe the
> shape of objects, slap the shaded areas derived from a phong ontop,
> mix al this together with SSS, put a bit of fake bounce into the shaded
> areas (the unlit areas, therefore "shadow" doesn´t help) and hope to
> get more dynamic range when I later add specular and reflections or
> by making better use of the existing layers.
>
> Cheers
>
> tim
>
> P.S: Please not the images don´t represent final look or detail but
> are stand-ins.
>
>
> ----- Original Message -----
> From: "guillaume laforge" <guillaume.laforge.3d(at)gmail.com>
> To: <XSI(at)Softimage.COM>
> Sent: Sunday, April 30, 2006 3:00 PM
> Subject: Re: tuning ambient occlusion dark/bright input using a custom
> curve?
>
>
> Sorry, I didn't notice you need it with "Sampled environment". As you need
> the rvb value, adding a color corrector after the occlusion node will
> work.
>
> You can color correct the environment map too if needed ( in fxtree for
> example ).
>
> If it really doesn't do the job you can break the occlusion and the
> environment color into two occlusion nodes.
>
> Set the first node to "occlusion using bent normal" and plug it  in the
> gradient. Set the second node to "Sampled environment" with a super small
> "max distance" ( 0.001 for example ). Mix the two nodes in multiply mode.
>
> It should be a little bit slower than a single occlusion node in "Sampled
> environment" but not too much with this small "max distance".
>
> However I never feel the need to use such render tree, it is just an idea
> ;-).
>
> Cheers,
>
>
>
>
>
>
>
>
>
>
>
> On 4/30/06, Tim Leydecker <BauerOink(at)gmx.de> wrote:
>
> > Hi Guillaume,
> >
> > changing the interpolation type alone (linear/cubic) helps in
> > getting a "softer" tonalmapping but kills the colors when I
> > pipe in an environment sampled AO.
> >
> > It seems just about any node that´ll result in the autocreation
> > of an extra Color2scalar node will do this, e.g. automatically
> > convert to my input greyscale...
> >
> > I´m still not an expert user in XSI, my impression may therefrore
> > be based on limited experience but I still find the available nodes
> > for the rendertree *limiting* - compared to my (still also limited)
> > experience with complex shading networks in Maya hypershade.
> >
> > In respect to the mib_ambient_occlusion, I should have asked if
> > there is an internal interpolation/conversion added to tonemapping
> > between Bright and Dark color, e.g. a gammacorrection, that could
> > be switched off?
> >
> > Cheers
> >
> > tim
> >
> > P.S: I´ve spent waaaay to much time on all this, didn´t expect to
> > find so many differences in the way shaders need to be set up in
> > XSI compared to Maya, e.g. getting used to the workflow.
> >
> >
> >
> >
> >
> >
> > ----- Original Message -----
> > From: "guillaume laforge" <guillaume.laforge.3d(at)gmail.com>
> > To: <XSI(at)Softimage.COM>
> > Sent: Sunday, April 30, 2006 9:30 AM
> > Subject: Re: tuning ambient occlusion dark/bright input using a custom
> > curve?
> >
> >
> > Hi Tim,
> >
> > Nodes > Mixers > Gradient. Set it to black and white and plug your
> > occlusion
> > shader in the gradient input.
> > Then you can change the "mid color" slider or add some colors in this
> > gradient to control the dark/bright from occlusion as you want.
> >
> > Cheers.
> >
> >
> > On 4/29/06, Tim Leydecker <BauerOink(at)gmx.de> wrote:
> > >
> > > Hi folks,
> > >
> > > I´m currently trying both mib_ambient_occlusion and the
> > > XSI 5.01 XSIAmbientOcclusion shader as a lightshader
> > > (e.g. sampling an environment texture) as described here,
> > >
> > > http://www.lamrug.org/resources/doc/occlusion_tutorial.pdf
> > >
> > > setup basically similar to pages 9 and 19, exept that I use
> > > a tiny 8bit file to reduce the amount of sampling rays
> > > required for clean results, also because I don´t have
> > > access to a fully functional *.exr creation/manipulation app.
> > >
> > > the results are quite fine allready (Thanks again to Harry Bardak)
> > > but now I´d like to improve the  tonemapping between Bright
> > > color and Dark color, like using a gradient/curve to soften the
> > > effect in heavily occluded areas *without* simply turning up
> > > the Dark color value. Is there a way to do this without piping
> > > a colorcorrection node inbetween the AO_out and (material)ambient?
> > >
> > > I don´t want to chop into my tonalrange by stretching things out
> > > but would like to have extra steps...or is there enough room for
> > > CC in the rendertree (e.g. 32/64bit calculation)?
> > >
> > > To get an idea what I look for, please compare the images of
> > > mib_fg_occlusion and mib_ambient_occlusion on page 19
> > > of the above *.pdf (Thanks for putting it together guys!).
> > >
> > > You´ll notice the FG occlusion looks more *elegant* compared
> > > to the mib_occlsion looking like it´s been *autocontrasted*.
> > >
> > > Any help appreciated, as I said, my results are rewarding allready
> > > (given my level of experience and ressources) but I still try to
> > > improve...
> > >
> > > Cheers
> > >
> > >
> > > tim
> > >
> > >
> > >
> > >
> > >
> > > environmet
> > >
> > >
> > > ---
> > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in
> > body:
> > > unsubscribe xsi
> > >
> >
> >
> >
> > --
> > Guillaume Laforge
> > freelance TD | cg Artist
> >
> >
> > ---
> > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in
> body:
> > unsubscribe xsi
> >
>
>
>
> --
> Guillaume Laforge
> freelance TD | cg Artist
>
>
> ---
> Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
> unsubscribe xsi



--
Guillaume Laforge
freelance TD | cg Artist


---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi