The crackle pattern appears to have changed in 3.7.  I haven't found this
documented anywhere.  I ran the following code using different versions and
version settings.  I will post the output in p.b.i.

global_settings { assumed_gamma 1 }

light_source { <-1, 1, -1> * 1000, rgb 1.5 }

plane
{  -z, -5
   pigment { rgb 1 }
   normal { crackle 2 }
}

Incidentally, this test scene ran much faster in 3.7RC3 than in 3.6.2.

OS: Windows 7 Ultimate
Word: 32 bits
Cores: 1, to the best of my knowledge

Post a reply to this message

Am 06.10.2011 01:52, schrieb Cousin Ricky:
> The crackle pattern appears to have changed in 3.7.  I haven't found this
> documented anywhere.

In a sense yes - the crackle pattern has changed. The general properties 
of the pattern haven't though, so the description in the docs is still 
valid.

There are a few other random-determined patterns that have changed as 
well in a similar manner. I can't tell off the top of my head though 
which those are.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Am 06.10.2011 01:52, schrieb Cousin Ricky:
> > The crackle pattern appears to have changed in 3.7.  I haven't found this
> > documented anywhere.
>
> In a sense yes - the crackle pattern has changed. The general properties
> of the pattern haven't though, so the description in the docs is still
> valid.

But the change can still cause legacy problems.  A scene file can produce
significantly different results if the pattern is used on a macro scale: for
example, creating a landscape with a height field or an isosurface.  John
Guthkelch's Code Rot is noticeably different.  One may have found just the right
artistic composition in 3.6 by translating the crackle just so.  Such a
composition will be ruined in 3.7, and there doesn't appear to be a way to
update the scene file to obtain the previous result.

The revisions.txt documents changes in the hash, the cache, and memory
management, but there is nothing explicit about any change in the output.

> There are a few other random-determined patterns that have changed as
> well in a similar manner. I can't tell off the top of my head though
> which those are.

It would be good to document these.

Thanks for all the work you've done.  It is much appreciated.

Post a reply to this message

Cousin Ricky <rickysttATyahooDOTcom> wrote:
> But the change can still cause legacy problems.  A scene file can produce
> significantly different results if the pattern is used on a macro scale: for
> example, creating a landscape with a height field or an isosurface.  John
> Guthkelch's Code Rot is noticeably different.  One may have found just the right
> artistic composition in 3.6 by translating the crackle just so.  Such a
> composition will be ruined in 3.7, and there doesn't appear to be a way to
> update the scene file to obtain the previous result.

  Backwards compatibility cannot always be maintained.

  If the scene absolutely must be upgraded to 3.7, then the solution is
to render the heightfield elevations as an image in 3.6 and then create
the heightfield from that image in 3.7.

-- 
                                                          - Warp

Post a reply to this message