"Samuel B." <stb### [at] hotmailcom> wrote:
> I've been wondering how one might
> make procedural craters and your post is making me think about it again...


Hi Sam,

You might check out the work of Sebastian Lague:
https://www.youtube.com/watch?v=lctXaT9pxA0

and related

https://www.reddit.com/r/proceduralgeneration/comments/l4ml53/generating_lunar_craters/

https://github.com/smcameron/space-nerds-in-space/blob/master/crater.c#L146

https://www.blendswap.com/blend/17050

I gave it a whirl, but the isosurface version was too challenging to debug after
a long day, but the normal worked "better" - still hella buggy.

Maybe you'll be able to puzzle out something better.

- Bill

Post a reply to this message

Attachments:
Download 'proceduralcrater.png' (860 KB)

Preview of image 'proceduralcrater.png'

Op 12-1-2024 om 04:00 schreef Samuel B.:
> Hi Thomas,
> 
> Your render looks good! It feels like it should be on an album cover for
> something featured on Hearts of Space :D
> 
> Are the craters procedural, or a texture map? I've been wondering how one might
> make procedural craters and your post is making me think about it again...
> 
You are not far wrong ;-) I had one of those covers from the SF magazine 
Analog in mind (somewhen 1966 - 1972) which I was reading at that time 
(does not make me feel younger...) and I have half a mind to continue in 
that particular vein.

The craters are generated by an uv_mapped bump_map. As I rarely do 
uv_mapping directly on primitive POV objects like here, I got some 
surprises which I need to investigate further. I shall certainly be back 
here with some questions. The documentation only gave me partial, for my 
purpose unsatisfactory answers.

Procedural craters might be an interesting option indeed. No idea at 
present on how to go about it...

-- 
Thomas

Post a reply to this message

Op 12-1-2024 om 06:49 schreef Bald Eagle:
> "Samuel B." <stb### [at] hotmailcom> wrote:
>> I've been wondering how one might
>> make procedural craters and your post is making me think about it again...
> 
> 
> Hi Sam,
> 
> You might check out the work of Sebastian Lague:
> https://www.youtube.com/watch?v=lctXaT9pxA0
> 
> and related
> 
>
https://www.reddit.com/r/proceduralgeneration/comments/l4ml53/generating_lunar_craters/
> 
> https://github.com/smcameron/space-nerds-in-space/blob/master/crater.c#L146
> 
> https://www.blendswap.com/blend/17050
> 
> I gave it a whirl, but the isosurface version was too challenging to debug after
> a long day, but the normal worked "better" - still hella buggy.
> 
> Maybe you'll be able to puzzle out something better.
> 
Not bad indeed, although the image looks a bit more like "seals" than 
anything else ;-) but is might be a start. I shall visit those sites next.

-- 
Thomas

Post a reply to this message

"Bald Eagle" <cre### [at] netscapenet> wrote:
> "Samuel B." <stb### [at] hotmailcom> wrote:
> > I've been wondering how one might
> > make procedural craters and your post is making me think about it again...
>
>
> Hi Sam,
>
> You might check out the work of Sebastian Lague:
> https://www.youtube.com/watch?v=lctXaT9pxA0
>
> and related
>
>
https://www.reddit.com/r/proceduralgeneration/comments/l4ml53/generating_lunar_craters/
>
> https://github.com/smcameron/space-nerds-in-space/blob/master/crater.c#L146
>
> https://www.blendswap.com/blend/17050

Thanks for the links. I only skimmed Sebastian's video, but I got the gist of
what I need to focus on if/when I get something going myself.

> I gave it a whirl, but the isosurface version was too challenging to debug after
> a long day, but the normal worked "better" - still hella buggy.

I'm not sure what's happening with that, but experimental things rarely work the
first time (or after long days).

> Maybe you'll be able to puzzle out something better.
>
> - Bill

I'm thinking of using layered form x crackle patterns. I found some smooth
minumum/maximum functions that might help with the blending...

Sam

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:
> Op 12-1-2024 om 04:00 schreef Samuel B.:
> > Hi Thomas,
> >
> > Your render looks good! It feels like it should be on an album cover for
> > something featured on Hearts of Space :D
> >
> > Are the craters procedural, or a texture map? I've been wondering how one might
> > make procedural craters and your post is making me think about it again...
> >
> You are not far wrong ;-) I had one of those covers from the SF magazine
> Analog in mind (somewhen 1966 - 1972) which I was reading at that time
> (does not make me feel younger...) and I have half a mind to continue in
> that particular vein.

Is Analog even still being published? Some of those stories were great.

> The craters are generated by an uv_mapped bump_map. As I rarely do
> uv_mapping directly on primitive POV objects like here, I got some
> surprises which I need to investigate further. I shall certainly be back
> here with some questions. The documentation only gave me partial, for my
> purpose unsatisfactory answers.

The default uv mapping is going to be different for every primitive, of course,
and some stretching is guaranteed in any case.

I'm thinking that for complex objects, triplanar mapping might be the best thing
to use. (Triplanar mapping in POV-Ray would probably use 3 nested slope maps,
with each axis getting one planar map). There /is/ a blending issue when using
triplanar mapping, but it can be mitigated somewhat. It's one of those things I
haven't gotten around to yet (I was going to use something similar to depict
claymation-style fingerprints on objects), but if I make any headway I'll post
it.

> Procedural craters might be an interesting option indeed. No idea at
> present on how to go about it...
>
> --
> Thomas

I'm thinking crackle form x patterns as a basis, but it's going to take a bit of
work...

Sam

Post a reply to this message

"Samuel B." <stb### [at] hotmailcom> wrote:

> I'm thinking that for complex objects, triplanar mapping might be the best thing
> to use. (Triplanar mapping in POV-Ray would probably use 3 nested slope maps,
> with each axis getting one planar map). There /is/ a blending issue when using
> triplanar mapping, but it can be mitigated somewhat. It's one of those things I
> haven't gotten around to yet (I was going to use something similar to depict
> claymation-style fingerprints on objects), but if I make any headway I'll post
> it.

I was _just_ reading this article ... last night, in fact.
At work.
Because I had gotten all of my work done already.

:D

http://iquilezles.org/articles/biplanar/

Post a reply to this message

Op 13/01/2024 om 02:30 schreef Samuel B.:
> 
> Is Analog even still being published? Some of those stories were great.
> 
It seems to be:

https://www.analogsf.com/


> The default uv mapping is going to be different for every primitive, of course,
> and some stretching is guaranteed in any case.
> 
Yes, I noticed that.

> I'm thinking that for complex objects, triplanar mapping might be the best thing
> to use. (Triplanar mapping in POV-Ray would probably use 3 nested slope maps,
> with each axis getting one planar map). There /is/ a blending issue when using
> triplanar mapping, but it can be mitigated somewhat. It's one of those things I
> haven't gotten around to yet (I was going to use something similar to depict
> claymation-style fingerprints on objects), but if I make any headway I'll post
> it.
> 
> 
> I'm thinking crackle form x patterns as a basis, but it's going to take a bit of
> work...
> 
Exciting info indeed! I shall study these ideas also; I don't like to 
sit back and wait for others to do the dirty jobs :-)

-- 
Thomas

Post a reply to this message

"Bald Eagle" <cre### [at] netscapenet> wrote:
> "Samuel B." <stb### [at] hotmailcom> wrote:
>
> > I'm thinking that for complex objects, triplanar mapping might be the best thing
> > to use. (Triplanar mapping in POV-Ray would probably use 3 nested slope maps,
> > with each axis getting one planar map). There /is/ a blending issue when using
> > triplanar mapping, but it can be mitigated somewhat. It's one of those things I
> > haven't gotten around to yet (I was going to use something similar to depict
> > claymation-style fingerprints on objects), but if I make any headway I'll post
> > it.
>
> I was _just_ reading this article ... last night, in fact.
> At work.
> Because I had gotten all of my work done already.
>
> :D
>
> http://iquilezles.org/articles/biplanar/

Hmm, I seem to remember seeing something about that before. I haven't looked at
the code very closely, but using two texture samples instead of three seems a
bit like magic. I wonder if it's more expensive than doing it the standard
way...

Sam

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:
> Op 13/01/2024 om 02:30 schreef Samuel B.:
> >
> > Is Analog even still being published? Some of those stories were great.
> >
> It seems to be:
>
> https://www.analogsf.com/

Ah cool. I'm a bit tempted to get a subscription for my mom (which would
eventually make its way here as well). I should probably check with the library
before doing that, though :D

> > The default uv mapping is going to be different for every primitive, of course,
> > and some stretching is guaranteed in any case.
> >
> Yes, I noticed that.
>
> > I'm thinking that for complex objects, triplanar mapping might be the best thing
> > to use. (Triplanar mapping in POV-Ray would probably use 3 nested slope maps,
> > with each axis getting one planar map). There /is/ a blending issue when using
> > triplanar mapping, but it can be mitigated somewhat. It's one of those things I
> > haven't gotten around to yet (I was going to use something similar to depict
> > claymation-style fingerprints on objects), but if I make any headway I'll post
> > it.
> >
> > I'm thinking crackle form x patterns as a basis, but it's going to take a bit of
> > work...
> >
> Exciting info indeed! I shall study these ideas also; I don't like to
> sit back and wait for others to do the dirty jobs :-)
>
> --
> Thomas

That's the spirit!

I managed to get one thing done today: the triplanar mapping
(https://news.povray.org/povray.binaries.images/thread/%3Cweb.65a4705751388f9716bed5696e741498%40news.povray.org%3E/).
It's very basic and doesn't have a sharpness setting (rate of transition between
normals).

The craters idea will have to wait for another day.

Sam

Post a reply to this message

Op 15/01/2024 om 00:50 schreef Samuel B.:
> Thomas de Groot <tho### [at] degrootorg> wrote:
>> Op 13/01/2024 om 02:30 schreef Samuel B.:
>>>
>>> Is Analog even still being published? Some of those stories were great.
>>>
>> It seems to be:
>>
>> https://www.analogsf.com/
> 
> Ah cool. I'm a bit tempted to get a subscription for my mom (which would
> eventually make its way here as well). I should probably check with the library
> before doing that, though :D
> 
<grin>
It is amazing that they still do exist after such a long time (1930, if 
I am correct). Analog was indeed my first introduction to the (serious) 
SF literature.

>>> The default uv mapping is going to be different for every primitive, of course,
>>> and some stretching is guaranteed in any case.
>>>
>> Yes, I noticed that.
>>
>>> I'm thinking that for complex objects, triplanar mapping might be the best thing
>>> to use. (Triplanar mapping in POV-Ray would probably use 3 nested slope maps,
>>> with each axis getting one planar map). There /is/ a blending issue when using
>>> triplanar mapping, but it can be mitigated somewhat. It's one of those things I
>>> haven't gotten around to yet (I was going to use something similar to depict
>>> claymation-style fingerprints on objects), but if I make any headway I'll post
>>> it.
>>>
>>> I'm thinking crackle form x patterns as a basis, but it's going to take a bit of
>>> work...
>>>
>> Exciting info indeed! I shall study these ideas also; I don't like to
>> sit back and wait for others to do the dirty jobs :-)
>>
>> --
>> Thomas
> 
> That's the spirit!
> 
> I managed to get one thing done today: the triplanar mapping
>
(https://news.povray.org/povray.binaries.images/thread/%3Cweb.65a4705751388f9716bed5696e741498%40news.povray.org%3E/).
> It's very basic and doesn't have a sharpness setting (rate of transition between
> normals).
> 
That is really looking good, Sam! I shall have to play with this.

Yesterday, I was exploring different issues more or less related to this 
and I experimented with the lunar crater image inside a cubic pattern, 
thinking that, perhaps, it would do something /similar/ to triplanar 
mapping. Of course it didn't, but it was a useful sidetrack because I 
could use it in an interesting way with a superellipsoid for instance, 
and even with the ovus object (which last one has a strange basic 
uv_mapping which I cannot agree with).

I had not got around to switch to pigment_pattern or image_pattern use, 
but it was on my ToDo.

> The craters idea will have to wait for another day.
> 
Of course. Still, a great progress already and worth a warm round of 
applause!

-- 
Thomas

Post a reply to this message