Am 9/21/2017 um 20:45 schrieb William F Pokorny:
> On 09/21/2017 06:15 AM, Ive wrote:
>> Am 9/20/2017 um 9:40 schrieb Mr:
>>
>> They are but the effect is too dim to be seen in this image.
>> While the headlights itself are modeled quite detailed and realistic I
>> don't think it is possible to get realistic head*lights* without using
>> IES light profiles, something POV-Ray does not support.
>>
>> -Ive
>
> I remember the VW van and bug headlights in the VWs of my youth as being
> VERY dim and so thought you'd gotten them about right for the environment!
>


> On the IES profiles. I took a run at some kind of support 5 years or
> more back on a mention of them in the newsgroups - prior to which I
> didn't know such files existed. Does a profile actually exist for a
> 1970s VW van headlight? ;-)
>
> Perhaps it was unfortunate choices in light fixture manufactures and IES
> profiles, but with the samples I grabbed, I ended frustrated by the
> differing standards over time, the seeming lack of full compliance to
> any of them and the frequent, probable errors in the data. A task that
> started simple per first look, turned into real work with a view to
> endless support. I quit. I got to where I was able to render a few
> reasonable looking IES profiles via light enclosing, hollow spheres with
> image mapped transparency(1).
>
> Today, are the other rendering tools supporting real IES manufacturer
> profiles (if so, what does that really mean...(3)) or are they using
> some already converted set of spot light like variations for a
> particular tool or standard being 'called' IES lights in the tool? I'm
> thinking it easier to support the latter sort of 'IES stand-in' in
> POV-Ray as that could perhaps be a smallish set of spherical image maps
> folks could use in some semi-standard light enclosing set up?
>
> Bill P.
>
> (1) - Passable for most, essentially spreading, IES intensity profiles
> such as a light and fixture against a building's siding say. This faking
> is not OK for light&fixture profiles which have focusing or additive
> areas as one moves away from the light in 3D. Such faking as that needs
> more complicated, light associated, 3D pigments/media densities for all
> objects and media around the light(s). This too is doable with functions
> or DF3s/functions in POV-Ray and it'd now be easier with function based
> user_defined {} pigments in 3.8. Though not IES related, I've played
> with projected_through no_image-isosurface-fragments as a way to shape
> 3d scene light intensity in interesting ways not requiring modification
> to in scene textures / media. Maybe such a technique could play an IES
> roll too...
>
> The IES files contain a smallish number of samples taken around a
> certain light source and light fixture which get interpolated for the
> resultant, in scene, profile.
>
> To 'accurately' model a light(2) and fixture one must shoot photons from
> a light source inside a good representation of the actual light fixture
> - perhaps also calibrate/fit to actual IES measured locations for the
> fixture. I'd argue POV-Ray can implement the photons method quite well -
> if one wants to burn the time and effort for the precise effect.
>
> (2) - As you know, you need too some overall system like lightsys
> (spectral rendering?) for the sources and intensities indicated in the
> IES file about which you know far, far more than me :-).
>
> (3) - For one, does such implementation grab too the light fixture's
> model? Asking because in my very limited bit of playing, I had trouble
> with the environment corrupting the result of the IES profiled light due
> not having the actual fixture (parts corresponding to
> black/blocked-by-the-fixture light of the profile). This was partly me
> not being able to easily sort out good sphere enclosed light position
> and orientations relative to the fixture given the actual light bulb's
> shape was not the shape of my spherical stand-in. All caused me to
> wonder whether folks using IES profiles were just after some rough
> effect more than any kind of exact one. If so, maybe the format issues
> don't matter all that much and sometimes sloppy IES files - so long as
> you can parse and interpret them somehow - are OK in practice. Or, maybe
> the IES implementation in other tools is much different than anything I
> considered?

 From what I can tell, especially aimed at architectural 
pre-visualisation, there is meanwhile a huge collection of IES profiles 
with and without fixtures from every lighting manufacturer available. 
Including headlights for cars - but surely not for an old VW van ;)
I'm using IES profiles in Luxrender and Iray with ease and all works 
very well - just out of the box and without any impact on render time - 
so I actually never did think much about it ;)

-Ive

Post a reply to this message

Am 9/23/2017 um 14:01 schrieb Kenneth:
> Thomas de Groot <tho### [at] degrootorg> wrote:
>> On 23-9-2017 10:58, clipka wrote:
>
>>> And yes, POV-Ray should be able to do it.
>
>>
>> Aha! That is an interesting notion... :-) So mie haze or mie murky
>> should do the trick. Something to look into...
>
> Yes indeed! I didn't realize that POV-Ray had that blurring effect already
> available.
>
> Although...I've's beautiful scene here is already using one or both of those
> scattering types (I assume), yet it's hard to tell if there is actual blurring
> around the streetlights. Perhaps the haze isn't 'thick' enough (or else the
> effect is very subtle, in his particular set-up.) The only way to be sure is to
> do some experimenting! :-)
>

I used rayleigh scattering for an overall atmospheric effect in 
combination with mie murky for the mist. A faint light blurring *is* 
there but note that POV-Ray's media will only blur light sources and not 
any object seen through the mist.

-Ive

Post a reply to this message

Ive <ive### [at] lilysoftorg> wrote:

>
> I used rayleigh scattering for an overall atmospheric effect in
> combination with mie murky for the mist. A faint light blurring *is*
> there but note that POV-Ray's media will only blur light sources and not
> any object seen through the mist.
>

I guess that means the light's look_like object is not included in the blurring
:-(  The effect I'm thinking of would necessarily need to include that. Oh well,
it was an interesting idea anyway.

As an alternative, I've been thinking about several different ways of
accomplishing the same thing-- all of which are basically 'cheats.' (The easiest
one involves simply post-processing the render in Photoshop or the like.) One
other wild idea would be for POV-ray to be re-written(!) so that its camera
focal_blur could be applied only to individual OBJECTS. But that's not the way
focal_blur works, of course... :-(

Post a reply to this message

"Kenneth" <kdw### [at] gmailcom> wrote:

> As an alternative, I've been thinking about several different ways of
> accomplishing the same thing--

What if you constructed an [oversized] object-shaped container and filled it
with media, or lots of little transparent spheres with an ior to make a cloud
that diffracts the light?
Or a slightly oversized version of the same object with filter/transmit applied
to give a "ghost" around the edges??

I'm sure there's some clever way to incorporate Perlin noise that might give an
indistinct appearance to an object.

What about literally blurring it with a lens placed into the scene between the
object and the camera?

Post a reply to this message

On 09/23/2017 08:19 AM, Ive wrote:
> 
>  From what I can tell, especially aimed at architectural 
> pre-visualisation, 

So these could likely be a little sloppy. Good to know.

there is meanwhile a huge collection of IES profiles
with and without fixtures from every lighting manufacturer available.

Yes, even when I played with these years ago, there were a lot of 
profiles out there. They seemed to be provided for every new fixture by 
major manufactures and the first version of the standard dates back to 
the 90s as I remember.

Do you have a link or two providing both MODEL and profile? I just had 
the thought with POV-Ray we could perhaps create our own 'IES profiles' 
- do the intensity measurements ourselves - if we had an accurate model. 
Suppose we'd not even have to bother with IES formats or files 
internally with that approach.

> Including headlights for cars - but surely not for an old VW van ;)
> I'm using IES profiles in Luxrender and Iray with ease and all works 
> very well - just out of the box and without any impact on render time - 
> so I actually never did think much about it ;)
>   Cool, you've used them in other tools! So when you use them, what do 
you know, or could you quickly determine, about what are you in fact using?

- Is the profile information added as extra information into a given 
light source modulating the source intensity as was done in a megapov 
patch in the early 2000s?

- Is it a complete lighting rig with predefined media around the light?

Asking because per-calculating emitting media as part of a 'light 
profile' might be doable and perform reasonably well. In other words 
faking the light/media interaction in the full render where one wants 
the profile represented in media but not the media as an accurate light 
source.

- Do either of implementations put an extra transparent surface with a 
transfer function in play as I was trying?

- Are the implementations otherwise decoupled from / not-tangled with 
the other objects in the scene? I'd guess, yes, but if not would be nice 
to know what other methods are being employed.

- If those two tools have optional script outputs, might we get a look 
at code snipets from both tools with example use?

- Also wonder if full spherical profiles are common or if it is 
primarily spot/directional lights?

I recall a free standing fixture having light bleeding through the back 
side such that the profile isn't the half sphere/horizontal angle sort. 
It had some holes on the topside/backside (venting for heat I suppose?) 
and the sampling looked to have caught some but not all the backside, 
peep hole light bolts. In this case the profile represented the reality 
poorly around part of the fixture. Suppose the support position could be 
the IES is what it is, but on seeing this I wondered whether a good IES 
implementation would offer options to perhaps force symmetry across an 
axis or something.

- Do the two tools you use with IES support 'fix-up/clean-up' options on 
import?

Wonder how it can take zero extra time given extra calculations or 
surfaces must be involved... I guess I could see how 'adc bailout' might 
happen earlier for some rays at the surface of a transparency mapped 
sphere/cylinder, but that approach generates extra rays at the surface 
and transmission calculations overall. If IES is handled in the light 
code itself, I guess it could better prune the surface-look-to-light 
rays as I believe happens today with POV-Ray's spotlights. Though I've 
not spent much time with the light source code.

Question to self: Might the profile itself affect the adapative AA 
algorithm?

I searched just now to see if since I played with IES files, perhaps 
others had come up with some public canned solutions and found this:

https://seblagarde.wordpress.com/tag/ies-parser/

but little else. Well, also one c sharp parser for one version of the 
standard, but looked like no back end for what got parsed.

No. I've not been able to find my own previous aborted attempt even in a 
couple backups...

One last question. Do these other ray tracing tools implement EULUMDAT, 
.ldt, support too?

Thanks for your time and any further information you can offer. I've 
opened up a github issue:

    https://github.com/POV-Ray/povray/issues/322

where we can roll(1) up links, discussion and information. Perhaps one 
of us will take another run at 'IES support' at some point. For now I 
have to get back to some other work.

Bill P.

(1) - On reading your response, my eye caught my previous misuse of 
'roll' for 'role.' :-)

Post a reply to this message

"Bald Eagle" <cre### [at] netscapenet> wrote:
> "Kenneth" <kdw### [at] gmailcom> wrote:
>
> > As an alternative, I've been thinking about several different ways of
> > accomplishing the same thing--
>
> What if you constructed an [oversized] object-shaped container and filled it
> with...lots of little transparent spheres with an ior to make a cloud
> that diffracts the light?

That's an intriguing idea-- sort of like duplicating the many *tiny* water
droplets that constitute thick fog. With variable sized-spheres or different
IORs. Food for thought :-)
>
> Or a slightly oversized version of the same object with filter/transmit applied
> to give a "ghost" around the edges?

Certainly possible, although I *think* the slightly larger object would still
look distinct/sharp. I'd have to experiment to find out.
>
> What about literally blurring it with a lens placed into the scene between the
> object and the camera?
>
I actually don't know if a POV-Ray lens-type object will cause 'blurring' the
way a real lens does (which is a combination of the lens itself and the eye's
inability to focus on the resulting 'aerial image.') I've never tried testing
such a thing in POV-Ray. I *suspect* that its camera will always keep all rays
sharply 'focused' (so to speak).

Post a reply to this message

"Kenneth" <kdw### [at] gmailcom> wrote:
>
> I actually don't know if a POV-Ray lens-type object will cause 'blurring' the
> way a real lens does (which is a combination of the lens itself and the eye's
> inability to focus on the resulting 'aerial image.') I've never tried testing
> such a thing in POV-Ray. I *suspect* that its camera will always keep all rays
> sharply 'focused' (so to speak).

.....because the camera types (excluding focal-blur of course) are what might be
termed 'ideal pinhole cameras'. In photographic terms, that's like a lens with
an infinitely small aperture-- where *everything* is always in focus, no matter
how close or far away. Our eyes can't do that. :-(

Post a reply to this message

William F Pokorny <ano### [at] anonymousorg> wrote:
> On 09/23/2017 08:19 AM, Ive wrote:
> >
> >  From what I can tell, especially aimed at architectural
> > pre-visualisation,
>
> So these could likely be a little sloppy. Good to know.

two open source available implementations:
http://www.luxrender.net/wiki/LuxRender_Lighting#IES_data

If in doubt, a good compromise to not go as far as full IES support is also to
look at Blender's custom curves light attenuation where a spline commands the
blending of a spotlight from center to edge of cone:

https://docs.blender.org/manual/ja/dev/render/blender_render/lighting/lights/light_attenuation.html#custom-curve

Post a reply to this message

"Kenneth" <kdw### [at] gmailcom> wrote:

> > Or a slightly oversized version of the same object with filter/transmit applied
> > to give a "ghost" around the edges?
>
> Certainly possible, although I *think* the slightly larger object would still
> look distinct/sharp. I'd have to experiment to find out.


Well, all of the suggestions were with the idea that there would be a fog/haze
in between the fake-blur and the camera to help hide the kludge.

It also occurred to me that one might be able to make an object slightly ---
hairy - to soften the edges.  Maybe do _that_ with a transparent+ior or
transmissive texture and see how that goes.

For the oversized transmissive object, I was thinking that perhaps there's a way
to soften / completely fade out the edges.
Interestingly, while looking for hair macro, I looked over at Rune's website and
found his Pattern Control Trick - "Change the color_map of a pigment along a
pattern"  Perhaps this could be of some use in this instance.

Post a reply to this message

On 09/25/2017 04:38 AM, Mr wrote:
> William F Pokorny <ano### [at] anonymousorg> wrote:
>> On 09/23/2017 08:19 AM, Ive wrote:
>>>
>>>   From what I can tell, especially aimed at architectural
>>> pre-visualisation,
>>
>> So these could likely be a little sloppy. Good to know.
> 
> two open source available implementations:
> http://www.luxrender.net/wiki/LuxRender_Lighting#IES_data
> 
> If in doubt, a good compromise to not go as far as full IES support is also to
> look at Blender's custom curves light attenuation where a spline commands the
> blending of a spotlight from center to edge of cone:
> 
>
https://docs.blender.org/manual/ja/dev/render/blender_render/lighting/lights/light_attenuation.html#custom-curve
> 
> 
> 
Thanks for these references.

Starting with the Blender reference, you can accomplish the custom 
curves sort of thing - and more - already in POV-Ray though in a less 
pre-packaged way. See the attached 3.8 scene for a quick example.

On the luxrender reference.

- So the LuxRender code itself is open source with a license compatible 
with POV-Ray's AGPLv3 license?

- Assuming a little about how the luxrender mesh light sources actually 
work and seeing their warning regarding IES light intensity corruption 
due them... One advantage of having some enclosing shape for 
non-spherical lights while handing the spherical IES light profile is 
that it gives you an adjustable transmission channel over the surface of 
the shape which can compensate for less-than-correct IES fading for 
non-point/centered source lights. Though, I have to think with some 
work, the same sort of adjustments/compensations could be part of any 
mesh light implementation using IES profiles and it more a matter it has 
not yet been done.

Bill P.

Post a reply to this message

Attachments:
Download 'utf-8' (2 KB)