I have also put my teeth into Bill Pragnell's Mesh Approximator macros, 
revisiting the ruined place I had been working on earlier.

Here, I have *eroded* the column segments using the MeshShape() macro. It 
took some experimenting until the shape came out right, but I am rather 
satisfied with this. The basic shape is CSG. I experimented with isosurface 
but could not get the top and bottom surfaces right. They resulted in 
strange funnel-shaped surfaces.

Another difficulty is the base of the column which consists of a flat CSG of 
torus and cylinder. I have not (yet) been able to approximate that correctly 
with the macro.

Anyway, here is the state of the art.

Thomas

Post a reply to this message

Attachments:
Download 'IonianTempleRuins_3.jpg' (85 KB)

Preview of image 'IonianTempleRuins_3.jpg'

On Tue, 10 Jun 2008 11:40:52 +0200, "Thomas de Groot"
<t.d### [at] internlDOTnet> wrote:

>
>Anyway, here is the state of the art.

That is looking nicely weathered. 
I know that it is still a WIP and you're using it to test/demo but I
have a couple of suggestions.
I think that the gaps between the blocks are too large and the layout
of the palace (?) is wrong. Columns that apparent size would support a
larger open space between the rows IMO. Anyway the columns are looking
good. 
-- 

Regards
     Stephen

Post a reply to this message

"Stephen" <mcavoysAT@aolDOTcom> schreef in bericht 
news:rsks44dlcong1hmr487v235eu73un6d02d@4ax.com...
>
> That is looking nicely weathered.
> I know that it is still a WIP and you're using it to test/demo but I
> have a couple of suggestions.
> I think that the gaps between the blocks are too large and the layout
> of the palace (?) is wrong. Columns that apparent size would support a
> larger open space between the rows IMO. Anyway the columns are looking
> good.
> -- 
>
Thank you indeed, Stephen.

I am aware of the gap between the blocks. I have already reduced it but 
still not enough. It appears from the weathering macro itself as the central 
axis is slightly thicker than the edges. I have now compensated 
(automatically) for the basis of the block, but I need to do this also for 
the top, in order for trace() to do a correct job.

About your second comment: It might depend on the type of space involved. 
The present setting would be rather claustrophobic, depending on the size 
relationship with human beings. I am not too concerned about this now, but I 
shall keep this in mind for further development.

Thomas

Post a reply to this message

Not to blow my own trumpet too blatantly, but I think this looks much better!
Although that's partly because you've also got the lichen texturing just right
- it looks really realistic. How do the render times compare?

"Thomas de Groot" <t.d### [at] internlDOTnet> wrote:
> took some experimenting until the shape came out right,

I find this to be the case also.

> Another difficulty is the base of the column which consists of a flat CSG of
> torus and cylinder. I have not (yet) been able to approximate that correctly
> with the macro.

Don't forget, you can relocate the object to be eroded before calling the macro.
If you move it upwards slightly it may improve things (the underside won't look
nice, but obviously that's not visible!). Another thing to try is the inner
shape - a wide rounded cylinder could well work better than the default
squashed sphere.

Looking good!

Bill

Post a reply to this message

On Tue, 10 Jun 2008 13:44:56 +0200, "Thomas de Groot"
<t.d### [at] internlDOTnet> wrote:

>About your second comment: It might depend on the type of space involved. 
>The present setting would be rather claustrophobic, depending on the size 
>relationship with human beings. 

And the local gravity ;)

>I am not too concerned about this now, 

I didn't think so :)

-- 

Regards
     Stephen

Post a reply to this message

"Bill Pragnell" <bil### [at] hotmailcom> schreef in bericht 
news:web.484e748d7b589d73731f01d10@news.povray.org...
> Not to blow my own trumpet too blatantly, but I think this looks much 
> better!
> Although that's partly because you've also got the lichen texturing just 
> right
> - it looks really realistic. How do the render times compare?

I am very satisfied with your macros, Bill. It does an excellent job where 
my previous experiments with isosurfaces grinded to a halt because of 
excessive render time. This scene parses and renders in about 20 minutes, of 
which about 12 or 13 is parsing. But of course, the mesh objects are already 
done. The parsing of the column segment takes about 3 to 4 minutes with a 
resolution of 300.

>
> "Thomas de Groot" <t.d### [at] internlDOTnet> wrote:
>> took some experimenting until the shape came out right,
>
> I find this to be the case also.

Size and position of the inner shape are crucial in relation to displacement 
depth. And the resolution of course :-)

>
>> Another difficulty is the base of the column which consists of a flat CSG 
>> of
>> torus and cylinder. I have not (yet) been able to approximate that 
>> correctly
>> with the macro.
>
> Don't forget, you can relocate the object to be eroded before calling the 
> macro.
> If you move it upwards slightly it may improve things (the underside won't 
> look
> nice, but obviously that's not visible!). Another thing to try is the 
> inner
> shape - a wide rounded cylinder could well work better than the default
> squashed sphere.
>
Yes, I am aware of the relocation posibillities and have used those 
extensively (min_extent and max_extent are pretty helpful here) and have 
used also a cylinder as inner shape which works better for these column 
segments.
My problem for the time being is with the column base which consists of a 
complex CSG of toruses and cylinders. I probably have to think about a 
complex inner shape too, especially as the object is very flat.

Thomas

Post a reply to this message

"Thomas de Groot" <t.d### [at] internlDOTnet> wrote:
> I am very satisfied with your macros, Bill. It does an excellent job where
> my previous experiments with isosurfaces grinded to a halt because of
> excessive render time. This scene parses and renders in about 20 minutes, of
> which about 12 or 13 is parsing. But of course, the mesh objects are already
> done. The parsing of the column segment takes about 3 to 4 minutes with a
> resolution of 300.

I'm very glad to hear it. Render speed was my primary motivation.

> Yes, I am aware of the relocation posibillities and have used those
> extensively (min_extent and max_extent are pretty helpful here) and have
> used also a cylinder as inner shape which works better for these column
> segments.

Good, I was just checking!

> My problem for the time being is with the column base which consists of a
> complex CSG of toruses and cylinders. I probably have to think about a
> complex inner shape too, especially as the object is very flat.

I can't quite make out the cross-section in your image. You could try merging a
vertically-stretched torus with a flat cylinder, if that makes sense. I guess
you've probably thought of all this already... good luck with it, and I'd be
interested in your eventual solution!

Bill

Post a reply to this message

Bill Pragnell napsal(a):
> "Thomas de Groot" <t.d### [at] internlDOTnet> wrote:
>> I am very satisfied with your macros, Bill. It does an excellent job where
>> my previous experiments with isosurfaces grinded to a halt because of
>> excessive render time. This scene parses and renders in about 20 minutes, of
>> which about 12 or 13 is parsing. But of course, the mesh objects are already
>> done. The parsing of the column segment takes about 3 to 4 minutes with a
>> resolution of 300.
> 
> I'm very glad to hear it. Render speed was my primary motivation.
> 
>> Yes, I am aware of the relocation posibillities and have used those
>> extensively (min_extent and max_extent are pretty helpful here) and have
>> used also a cylinder as inner shape which works better for these column
>> segments.
> 
> Good, I was just checking!
> 
>> My problem for the time being is with the column base which consists of a
>> complex CSG of toruses and cylinders. I probably have to think about a
>> complex inner shape too, especially as the object is very flat.
> 
> I can't quite make out the cross-section in your image. You could try merging a
> vertically-stretched torus with a flat cylinder, if that makes sense. I guess
> you've probably thought of all this already... good luck with it, and I'd be
> interested in your eventual solution!
> 
> Bill
> 
> 
You may try a vertically squashed superellipsoid (rounded cylinder)

-- 
the ultimate time-killer:
+a0.0 +am2 +r9

Johnny D

Post a reply to this message

"Bill Pragnell" <bil### [at] hotmailcom> schreef in bericht 
news:web.484ea28a7b589d73731f01d10@news.povray.org...
>
> I can't quite make out the cross-section in your image. You could try 
> merging a
> vertically-stretched torus with a flat cylinder, if that makes sense. I 
> guess
> you've probably thought of all this already... good luck with it, and I'd 
> be
> interested in your eventual solution!
>

I am sorry for the very bad composition of this image, but it is just a test 
to show what I am talking about. The red object is the original column base, 
the green object is obtained with a cylinder as inner shape. Not good yet, 
as I more or less expected. I tried with a scaled down copy as inner shape, 
but that was not satisfying either. Next, I want to try with an adapted copy 
of the original.

If you want to experiment, I attached the original inc file to this message.

Thomas

Post a reply to this message

Attachments:
Download 'BP_meshdemo.jpg' (29 KB) Download 'IonianColumnBase.inc.txt' (2 KB)

Preview of image 'BP_meshdemo.jpg'

"Thomas de Groot" <t.d### [at] internlDOTnet> wrote:
> If you want to experiment, I attached the original inc file to this message.

It is not surprising that you're having trouble; there are too many gaps in the
base object. Look at the cross-section in the image below!

You need to close these gaps, especially the thin gap across the centre and the
concave underside. Unless the square peg-hole has some special purpose I'd plug
this too. And don't forget to make sure it's a merge, not a union, or the
internal surfaces get hit first!

A default flattened sphere as inner object should work OK once you've filled the
gaps.

Hope this helps!

Bill

Post a reply to this message

Attachments:
Download 'coltest.jpg' (30 KB)

Preview of image 'coltest.jpg'