Re: Dosch HDRI maps

[Andy Jones <andy(at)thefront.com>]

I'm not totally clear on what you're using your reflection map for.  It 
sounds like you're using it for IBL, in which case I think it's possible 
that the over-brightness could be coming from whatever method of IBL 
you're using.  Maybe it's fixed in this version (I'm not using XSI 
Materials anymore, so I don't know what happens off the top of my head), 
but Final Gathering used to produce overly-bright results, due to the 
missing factor of 1/pi in the calculations.

Since you mentioned it, lately, I've had really fantastic results with 
linear space workflow (and gradually, I think I'm even winning over our 
compositors -- particularly now that we've discovered "guide layers" in 
AE), so everything I'm doing is now biased towards trying to get renders 
that treat light in a physically-realistic and consistent manner.  Then 
we can do any "cheating" we want to do in the comp.  It's just easier 
this way, since everything plays nicely together.  Thinking back to 
before we adopted linear workflow, we constantly had problems with 
balancing lights and contributions from different light sources.  And 
reflections never looked quite real (you never got that sort of image 
duality you get in real life, where the diffuse color and the reflected 
color are both perceptually present).  I don't mean for this to turn 
into a rant about the benefits of linear space workflow, but I do think 
it has a lot to do with what sort of look you're going to get out of 
using high dynamic range environment maps for lighting and reflections.  
So, I guess what I'm saying is, it's really hard to disagree or agree 
with what you're doing without knowing if you're working with linearized 
textures and applying a gamma correction after-the-fact.  It sounds like 
maybe you're working in PC gamma space -- particularly if you're using 
16-bit renders.  Is that the case?  If so, having to apply a gamma of 
2.2 to the maps before rendering would make sense.  I've seen a lot of 
people leave linearized reflections from .hdr or .exr files in their 
renders without ever doing a gamma correction, and it works sort of 
okay, since what they're typically going for is bright values that 
"punch through" a partially-reflective material, but I would think 
diffuse light contributions would produce pretty bad results without 
texture linearization and image gamma-correction (probably in the form 
of overly-hard, constrasty lighting, as you mention).

Going back to the original thread, this is actually a big part of my 
beef with the computer-generated Dosch images -- they appear to be 
living mostly in video space before any gamma is applied, and therefore 
don't have good range.  Also, maybe I mentioned this, but having 
completely black materials in an exr file to be used for environment 
maps is pretty much just insane, and they do it all over the place.

I'm hoping that the eventual result of this thread will be guidelines 
for a toolkit for generating a set of environment maps for a given 
scene, which I will then build in my nonexistent spare time.  Currently, 
the process is a little too slow, and we end up re-using environment 
maps when we really shouldn't.  Ideally, what I'd like is to have a 
one-click solution, where I can place an environment-recording source in 
a scene and optionally generate a single environmnet map or time-driven 
sequence of environment maps.  This would hopefully include several 
levels of blurriness, as well as a "fast-lookup diffuse ball" 
environment map (allowing for an IBL look and feel with a single texture 
lookup).  Probably, this toolkit will also involve a few shaders for 
coordinating raycast types between material shaders and environment 
shaders, so that in a shader setup, you can control which environment 
map gets used.  So, that being said, if anyone has ideas for tools 
they'd like to see, I'd love to hear them.

Thanks,
Andy

Tim Leydecker wrote:

> Following up on bogus intensity shifts, seems
> I get expected results when I use correct gamma,
> e.g. "display gamma affects rendering" together
> with a 2.2 gamma gives a way smoother and
> more natural look than sticking to the 1.0 gamma
> and  having to tune down the materials due to
> the more pronounced contrast biasing to the brights.
>
> The above is in no way accurate and shouldn´t
> be mixed up with working in a "linear space workflow"
> but helps alot in getting believable environmental lighting
> with controlable diffuse component, as a rule of thumb,
> gamma around 1 (for the *.hdr) will be hard, reducing
> the soft, bouncy lighting component while a higher gamma
> will result in smoother transitions and a quite softer look,
> pronouncing the mid values, washing out darks as well
> as toning down the hotspot areas. In a way it´s like
> going from a sunlight to an overcast sky lighting setup.
> Personally, I prefer to default to the more diffuse look
> and "S-Curve" things later in post, esp. for 16+bit renderings.
>
> Cheers
>
> tim
>
> ----- Original Message ----- From: "Tim Leydecker" <BauerOink(at)gmx.de>
> To: <XSI(at)Softimage.COM>
> Sent: Tuesday, September 19, 2006 12:04 PM
> Subject: Re: Dosch HDRI maps
>
>
> > Hi Andy,
> >
> > sorry for the delayed response, I lost track.
> >
> > Yeah, getting good UV´s on the sphere to rendermap from
> > has proven bitchy, I´ve still got some stretching at the poles
> > currently but decided to live with it for now.
> >
> > Regarding the intensity shift, factor 2 or 4 are ballparkfigures
> > I´ll dial in to account for the fallof that seems to happen when
> > using the rendermapped file as source for the enviromentshader
> > together with FG. From a physically accurate POV one could
> > argue this "must be" correct, due to conversion happening from
> >
> > light>rendermapshere>envshader>lit object
> >
> > but this could also be a complete bogus and I had just set my
> > diffuse values wrong initially...
> >
> >
> > Cheers
> >
> > tim
> >
> > P.S: I just learned yesterday that at least Mayas materials have
> > a factor 3.14 (pi) too high defaultvalue for the diffuse, althought
> > the irradiance attribute has been fixed for mR (both XSI+Maya),
> > this seems to have slipped through, maybe something to do with that...
> >
> >
> > ----- Original Message ----- From: "Andy Jones" <andy(at)thefront.com>
> > To: <XSI(at)Softimage.COM>
> > Sent: Saturday, September 16, 2006 11:38 PM
> > Subject: Re: Dosch HDRI maps
> >
> >
> > > Actually, now that you mention it, I don't know that I've ever 
> > > generated a 32-bit reflection map from scene geometry.  I have taken 
> > > existing reflection maps and blurred them using shaders and texture 
> > > baking to allow for optimized convergence in shaders that 
> > > multi-sample the environment (since just blurring a reflection map 
> > > typically produces uneven blur).  I guess to do a straight-up 
> > > reflection map, I would probably just throw a reflection shader on a 
> > > sphere and rendermap to a spherical projection.  I suppose you could 
> > > do a cubic cross, but AFAIK, you'd have to do some manual uv'ing or 
> > > something.  I think at one point, we wrote a script to spherize an 
> > > object, so you could cubic-project on a cube, then subdivide a bunch 
> > > of times and spherize it.  So you'd basically have a 
> > > perfectly-cubically-divisible sphere.  Otherwise, the edges aren't 
> > > really in the right place, I think.  You could also do a cubic map 
> > > by rendering out 90 degree cameras in 6 directions --  I've done 
> > > that for quick&dirty GL stuff before.
> > > I think in the past, I've usually gone straight to .map, and then 
> > > used the .map files directly, without re-exposing.  I don't really 
> > > see why you should have to adjust for intensity in a general sense 
> > > -- do you have any idea why that is?
> > >
> > > -Andy
> > >
> > >
> > >
> > > Tim Leydecker wrote:
> > >
> > > > Sorry, I´m using *.ct, not *.map. *.ct works.
> > > > Convert it using imf_disp.exe from a Maya install.
> > > > (XSI one will work the same or better,depends on build).
> > > >
> > > > Cheers
> > > >
> > > > tim
> > > >
> > > >
> > > > ----- Original Message ----- From: "Tim Leydecker" <BauerOink(at)gmx.de>
> > > > To: <XSI(at)Softimage.COM>
> > > > Sent: Saturday, September 16, 2006 10:11 AM
> > > > Subject: Re: Dosch HDRI maps
> > > >
> > > >
> > > > > How do you create your 32bit environmentmaps in xsi?
> > > > >
> > > > > custom shader? spherical ones or cubics?
> > > > >
> > > > > I currently bake a 100%reflective sphere to *.map texture
> > > > > and convert that to a *.hdr/*.exr since it seems using any
> > > > > other format produces incorrect results in XSI 5.1/5.11
> > > > > when selected a target for a baking. I think I logged it a bug
> > > > > allready, at least for *.exr. This approach lacks good
> > > > > filtering, thought. Also reuqires the resulting file to be
> > > > > normalized by factor 2 or 4 to get the proper intensity back.
> > > > >
> > > > > HDRshop is nice to have when comparing actual dynamic
> > > > > range of source and baked texture for such a matter and
> > > > > fitting the "intensity" to the former.
> > > > >
> > > > > Cheers
> > > > >
> > > > >
> > > > > tim
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > ----- Original Message ----- From: <andy(at)thefront.com>
> > > > > To: <XSI(at)Softimage.COM>
> > > > > Sent: Saturday, September 16, 2006 7:26 AM
> > > > > Subject: Re: Dosch HDRI maps
> > > > >
> > > > >
> > > > > > This is good to know.  So, I won't raise my arms and scream if 
> > > > > > anyone
> > > > > > tries to buy the Extreme High Res ones or the skies (although I will
> > > > > > mention that we already own the skies).  I still highly recommend 
> > > > > > that
> > > > > > others do so if anyone tries to buy "radiant skies" or "industrial
> > > > > > reflections."
> > > > > >
> > > > > > The project we're working on right now requires interior 
> > > > > > lighting.  And
> > > > > > actually, we were hoping to relight each shot individually.  Does 
> > > > > > anyone
> > > > > > have any good tricks for getting nice "soft-box" effects in 
> > > > > > reflections? And has anyone used the stuff from HDRI-Studio with 
> > > > > > any success?  It seems
> > > > > > like what we'd really like to have is just a model of a soft-box 
> > > > > > with a
> > > > > > proper re-exposable exr texture that I could duplicate and place 
> > > > > > in a
> > > > > > scene, then generate my own exr environment maps from it (or just 
> > > > > > render
> > > > > > it directly).
> > > > > >
> > > > > > -Andy
> > > > > >
> > > > > > On Fri, September 15, 2006 3:59 pm, Francois Lord wrote:
> > > > > >
> > > > > > > The Extreme High Res ones are useful when you need an 360 degrees
> > > > > > > environment for an SD shot, or when you need very crisp 
> > > > > > > reflections in HD.
> > > > > > > Creating your own HDR image with that quality is quite difficult.
> > > > > > > The skies are ok. It's definitely cheaper to buy them than to go 
> > > > > > > out and
> > > > > > > try to get the right sky for the shot you need... for 14 days in 
> > > > > > > a row.
> > > > > > >
> > > > > > > Kris Rivel wrote:
> > > > > > >
> > > > > > > > Well I can only speak on behalf of the "Extreme High Res" and 
> > > > > > > > standard
> > > > > > > > "Sky" libraries.  I found them very useful on several jobs where
> > > > > > > > creating or shooting our own HDRI was out of the question.  If 
> > > > > > > > you have
> > > > > > > > the time and money though, its worth making your own with a proper
> > > > > > > > setup.  I can highly recommend the guys at EPOP 
> > > > > > > > (http://www.epop.com/).
> > > > > > > > They use one of those great Spheron cameras
> > > > > > > > and they created stunning HDRI images for me in the past.
> > > > > > > >
> > > > > > > > Kris
> > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > > > Andy Jones wrote:
> > > > > > > >
> > > > > > > > > Yeah, I suppose we have too, though we've had just as much 
> > > > > > > > > luck just
> > > > > > > > > browsing through Turbo Squid.  But these exr's, specifically, are
> > > > > > > > > pretty weak, I think.
> > > > > > > > >
> > > > > > > > > -Andy
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > Joe Laffey wrote:
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > > On Thu, 14 Sep 2006, Andy Jones wrote:
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > > This is just an FYI for anyone who doesn't want to waste 
> > > > > > > > > > > $125 or
> > > > > > > > > > > so.  This is now the third time one of our project leads has
> > > > > > > > > > > decided to buy a package of Dosch HDRI environment maps for a
> > > > > > > > > > > project.  We now own skies, radiant skies, and industrial
> > > > > > > > > > > reflections.  Of these, it's really obvious that radiant skies
> > > > > > > > > > > and industrial reflections are just CG renders of
> > > > > > > > > > > poorly-constructed environments.  In the industrial 
> > > > > > > > > > > reflections,
> > > > > > > > > > > there are, for example, completely black materials (that don't
> > > > > > > > > > > ever change color no matter what exposure you use).  And 
> > > > > > > > > > > none of
> > > > > > > > > > > the interior lighting re-exposes in any sort of reasonable way.
> > > > > > > > > > > And the clouds
> > > > > > > > > > > in the radiant skies are really stupid-looking hand-painted
> > > > > > > > > > > chicken scratches.  The regular "skies" package is not as 
> > > > > > > > > > > bad, but
> > > > > > > > > > > I
> > > > > > > > > > > wouldn't say I'm particularly impressed by it, either. Anyway,
> > > > > > > > > > > just a heads-up about a lame product.  For interiors at least,
> > > > > > > > > > > just save yourself some money and make your own environment 
> > > > > > > > > > > maps
> > > > > > > > > > > of your scenes... If you don't have a scene, just throw some 
> > > > > > > > > > > cubes
> > > > > > > > > > > in there -- that's all Dosch did.
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > FWIW, I have gotten some models from Dosch that were more than
> > > > > > > > > > adequate for the price. So noth everything they have is lame. 
> > > > > > > > > > BUt,
> > > > > > > > > > like anything caveat emptor.
> > > > > > > > > >
> > > > > > > > > > --
> > > > > > > > > > Joe Laffey                |       Visual Effects for Film and 
> > > > > > > > > > Video
> > > > > > > > > > LAFFEY Computer Imaging   |
> > > > > > > > > > -------------------------------------
> > > > > > > > > > St. Louis, MO             |       Show Reel 
> > > > > > > > > > http://LAFFEY.tv/?e02809
> > > > > > > > > >  USA                       |
> > > > > > > > > > -------------------------------------
> > > > > > > > > > .                         |        -*- Digital Fusion Plugins 
> > > > > > > > > > -*-
> > > > > > > > > > -------------------------------------------------------------------- 
> > > > > > > > > >
> > > > > > > > > > ------
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Mail here will be rejected --> "Real Trap"
> > > > > > > > > > <r_trap(at)laffeycomputer.com>
> > > > > > > > > > ---
> > > > > > > > > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following 
> > > > > > > > > > text in
> > > > > > > > > >  body:
> > > > > > > > > > unsubscribe xsi
> > > > > > > > >
> > > > > > > > > ---
> > > > > > > > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following 
> > > > > > > > > text in
> > > > > > > > > body:
> > > > > > > > > unsubscribe xsi
> > > > > > > >
> > > > > > > > ---
> > > > > > > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following 
> > > > > > > > text in
> > > > > > > > body:
> > > > > > > > unsubscribe xsi
> > > > > > > ---
> > > > > > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following 
> > > > > > > text in body:
> > > > > > >  unsubscribe xsi
> > > > > >
> > > > > >
> > > > > > ---
> > > > > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text 
> > > > > > in body:
> > > > > > unsubscribe xsi
> > > > > ---
> > > > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text 
> > > > > in body:
> > > > > unsubscribe xsi
> > > >
> > > >
> > > > ---
> > > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text 
> > > > in body:
> > > > unsubscribe xsi
> > >
> > >
> > > ---
> > > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in 
> > > body:
> > > unsubscribe xsi
> >
> >
> > ---
> > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in 
> > body:
> > unsubscribe xsi
>
>
> ---
> Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in 
> body:
> unsubscribe xsi


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