Thomas de Groot <tho### [at] degrootorg> wrote:
> Op 26/04/2021 om 08:55 schreef Thomas de Groot:
> > Quick-and-Dirty.
> >
> > Yes, this is promising indeed! Good thinking!
> >
>
> ...and with an appropriate pigment_pattern.
>
> Later today I shall study those two sites and see if I can use the info.
>
> --
> Thomas

Exciting ! Now the tint looks spot on and better than the original.
About pattern, however, the fact that both Ive's and your own "quick and dirty"
versions feel more pleasing seems to confirm that the flakes scale may be
slightly too small, even for realistic measurements. Though I do understand that
Ive's scale is not acceptable to the rock specialist's expertise. Maybe a
compromise could be found if the noise depth is not yet too low and could be
reduced one level so as not to generate too small of a minimal cell?

Post a reply to this message

Op 26-4-2021 om 13:11 schreef Mr:
> Thomas de Groot <tho### [at] degrootorg> wrote:
>> Op 26/04/2021 om 08:55 schreef Thomas de Groot:
>>> Quick-and-Dirty.
>>>
>>> Yes, this is promising indeed! Good thinking!
>>>
>>
>> ...and with an appropriate pigment_pattern.
>>
>> Later today I shall study those two sites and see if I can use the info.
>>
>> --
>> Thomas
> 
> Exciting ! Now the tint looks spot on and better than the original.
> About pattern, however, the fact that both Ive's and your own "quick and dirty"
> versions feel more pleasing seems to confirm that the flakes scale may be
> slightly too small, even for realistic measurements. Though I do understand that
> Ive's scale is not acceptable to the rock specialist's expertise. Maybe a
> compromise could be found if the noise depth is not yet too low and could be
> reduced one level so as not to generate too small of a minimal cell?
> 
Finding the correct size/distribution of individual minerals is going to 
be a careful and slow process. Each pattern has its own range of scales; 
if I concentrate on /cells/ as in the latest image, the smallest should 
be a bit larger and the largest a bit smaller. ;-)

However, even in a RW granite, the very smallest minerals can only be 
distinguished through a microscope, forming  the basic matrix of the 
rock. Largest minerals can be really large, forming so-called 
/phenocrysts/. For those interested: 
https://en.wikipedia.org/wiki/Phenocryst .

-- 
Thomas

Post a reply to this message

This is a version of the frosted granite, with quartz veins, but also 
with ligh-coloured minerals in the matrix.

-- 
Thomas

Post a reply to this message

Attachments:
Download 'dakotagrains_test.jpg' (104 KB)

Preview of image 'dakotagrains_test.jpg'

Op 26/04/2021 om 12:47 schreef Bald Eagle:
> Thomas de Groot <tho### [at] degrootorg> wrote:
> 
> Thanks for making the render.  Looks better than my sphere - which looks like a
> cheap hard-rubber bowling ball, or the foam they use in sofa cushions.  :D
> 
> Needs 0.2-0.3 more turbulence. ;)
> 
>> The granite pattern is (almost) correct (without the turbulence).
> 
> I don't know what you mean.
> 
The original granite code contains turbulence. That is one of the 
reasons the original granite looks like a 'flow' pattern. The first 
thing I did was to comment out the turbulence; the granite became more 
'crispy' and closer to the real thing.

>> It also explains more comprehensively the original name: Mahogany. The
>> colour is similar to mahogany wood. Afaiac, I believe I should follow
>> that path in the first place and maybe disregard my own lighter version
>> (or not). Not sure yet. What do you think?
> 
> Well, there are myriad samples available at stonecontact-dot-com, all of which
> fall under the blanket/umbrella trade name of Dakota Mahogany and which have
> widely differing "color maps".
> 
I know. One of the problems is probably also that we do not know how 
much photoshopping has been applied to those photographs. However, I 
strongly believe that there is a link between the trade name and the 
deep hue of the mahogany wood. It would certainly help the sell.

> I think we need to alternate in our approach to this, and on the one hand try to
> formulaicly mimic the underlying patterns and sub-patterns, but on the other
> hand not forget that we can do some good ole' raytracing sleight-of-hand and do
> everything we can to fool the eye and mind into believing that they see
> something that is not there.
> 
Oh yes.

> One thing I thought of was to take a/some granite photos and run some
> edge-finding filters on them to see what pops out.
> 
> I'm curious about the rgb color mapping - the individual values - but also,
> given an individual grain region, what is the rgb variation across it?  How much
> secondary or tertiary pigment is patterned into the grain?
> 
That is a tedious exercise ;-) I did it years ago to get correct colour 
values for Mediterranean-style roof tiles (for which I used the cells 
pattern, btw). Can be done of course; I shall put it on my 
ToDo/ToExplore list.

> Is there a way to estimate the grain sizes from the photos?  Or maybe there is
> statistical data in the literature?
> 
That is more difficult. Literature tells me that grain sizes range from 
microscopic to about 10-15mm in most cases and within the same granite, 
with sometimes larger minerals (phenocrysts) in the more porphyric 
varieties. See my answer to Mr.

> I ran across a neat conformal mapping image with circle packing that looked like
> a good mathematical way to get discrete regions with good grain-size variation.
> 
> AFAIK, we're using straight Perlin noise right now for granite.   Doing floor()
> on each of the axes produces the cells {} pattern.  I'm thinking if we could
> split the difference.... maybe with the step trick and/or select ()
> 
> Also need to keep in mind that the base pattern might just be a guide, and the
> final "look" is going to likely be a result of layering textures/materials.
> Hard to visualize.
> 
> And lastly, there were the brick pattern and random hexagon color threads that
> employed methods we might be able to use here.
> 
All the above: yes; might need additional investigation. Concerning the 
hexagon pattern: many if not most of the granite minerals are hexagonal. 
That would mean it would be more appropriate for granites. However, I am 
afraid it would look 'circular' in practice.

>
http://news.povray.org/povray.general/thread/%3C5ae17c9c%40news.povray.org%3E/?mtop=422221&moff=10
> 
>
http://news.povray.org/povray.binaries.images/thread/%3Cweb.5ad633941828641ca47873e10%40news.povray.org%3E/?mtop=422589
> 

-- 
Thomas

Post a reply to this message

Op 26/04/2021 om 12:47 schreef Bald Eagle:

>
http://news.povray.org/povray.general/thread/%3C5ae17c9c%40news.povray.org%3E/?mtop=422221&moff=10
> 
>
http://news.povray.org/povray.binaries.images/thread/%3Cweb.5ad633941828641ca47873e10%40news.povray.org%3E/?mtop=422589
> 

Thanks for the reminder. I found test scene files I made back then from 
these posts.

So much we forget... The Search option of the POV-Ray site works 
adequately, but often I need something more sophisticated...

-- 
Thomas

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:
> This is a version of the frosted granite, with quartz veins, but also
> with ligh-coloured minerals in the matrix.
>
> --
> Thomas

To my profane eye this scale variation looks very natural! (though it remains to
be seen if it still holds from close up and when the pattern will be the real
one)

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:

> The original granite code contains turbulence. That is one of the
> reasons the original granite looks like a 'flow' pattern. The first
> thing I did was to comment out the turbulence; the granite became more
> 'crispy' and closer to the real thing.

Ah.  The pattern is also gradient noise based, which causes smooth blending
between the values, and creates a "roundness" - the rolling hills of a
heightfield.  I looked into modifying that, and I got something that's at least
good enough for some experiments to see if it's worth any more pursuit.


> That is a tedious exercise ;-) I did it years ago to get correct colour
> values for Mediterranean-style roof tiles (for which I used the cells
> pattern, btw). Can be done of course; I shall put it on my
> ToDo/ToExplore list.

Maybe eval_pigment and that histogram macro in .... math.inc?

> All the above: yes; might need additional investigation. Concerning the
> hexagon pattern: many if not most of the granite minerals are hexagonal.
> That would mean it would be more appropriate for granites. However, I am
> afraid it would look 'circular' in practice.

Yes, a slice down the side of a hexagonal basalt column would give a
rectangle...
Maybe layered patterns with all but the base having a mostly clear pigment in
the map.  That way we could just "sprinkle" a few different shapes in and not
have to do a lot of math / algorithms / pattern tweaking, and still get the
crisp grain boundaries.

Post a reply to this message

Attachments:
Download 'stepnoise.png' (229 KB)

Preview of image 'stepnoise.png'

"Mr" <nomail@nomail> wrote:

> To my profane eye this scale variation looks very natural! (though it remains to
> be seen if it still holds from close up and when the pattern will be the real
> one)

Yes, this reminds me to
a) comment on that:  It looks very VERY good.   The improvements really make a
huge difference.  I think that it's really a LOT better than the original.
Excellent work, Thomas.

and B)
Blender has a granite shader pack that maybe we can get some hints from  ;)
https://blendermarket.com/products/procedural-granite-shader-pack/

Post a reply to this message

Op 27-4-2021 om 12:49 schreef Bald Eagle:
> "Mr" <nomail@nomail> wrote:
> 
>> To my profane eye this scale variation looks very natural! (though it remains to
>> be seen if it still holds from close up and when the pattern will be the real
>> one)
> 
> Yes, this reminds me to
> a) comment on that:  It looks very VERY good.   The improvements really make a
> huge difference.  I think that it's really a LOT better than the original.
> Excellent work, Thomas.
> 
Thank you, sir.

I am /beginning/ to feel some satisfaction myself ;-) It is better to 
test on the frosted version than on the polished one. No surprise there.

Not yet there however...

> and B)
> Blender has a granite shader pack that maybe we can get some hints from  ;)
> https://blendermarket.com/products/procedural-granite-shader-pack/
> 
Thou shalt not bow to foreign idols! :-)

I shall have a peep... From the images shown, especially the Dakota 
Mahogany, it appears to me (gut feeling) that a 'granite' pattern was 
used; the minerals' shape appear somewhat like my initial texture.

-- 
Thomas

Post a reply to this message

Here's the half-noise half-cells pattern I worked out.

Adjusting Start and End to be both 0.5 gives results identical to cells.   Due
to the smoothstep, you can never get pure gradient noise.

I thought it was wasted time as it still has some gradient to it, but with
double-entry color_maps, you get a sharp delineation anyway...

Please "give it the treatment" and see if this produces something better or
worse than the turbulated cells.

(And your email bounced me.  :P)

Post a reply to this message

Attachments:
Download 'stepnoise.pov.txt' (3 KB)