> On 10/26/2013 08:10 PM, Alain wrote:
>>
>> I have'nt tryed the third version, but I got very good renders, much
>> faster after some little changes to your code.
>>
>> First, max_gradient 160 if huge overkill.
>> Using max_gradient 1.5 resulted in a nice render without any message
>> about max_gradient been to small or to large.
>> Next, I used:
>> #declare MinFact = 0.6;
>> evaluate 1.5*MinFact, sqrt(1/MinFact), 0.7
>> and got a slightly faster render.
>>
>> Next, I reduced the contained_by object as follow:
>> box {<-0.34,0.02,-0.34><0.34,0.908,0.34> }
>>
>> It's now much tighter and large enough to contain the whole isosurface.
>> This got me another performance increase.
>>
>> Alain
>
> Yes, good suggestions. I didn't tighten up my scribble and render set up
> before posting and I've never gotten my head to understand the evaluate
> option of isosurfaces.

Relatively simple:
You first do a render using the default max_gratient value and get the 
reported max_gradient in the message.
Next, you just plug that value in the first parameter. The two others 
are normaly correct using sqrt(1/MinFact) and 0.7 with MinFact set to 
the proposed value of 0.6.

This information on how to use the feature was prety obscure in the 
documentation for version 3.6, it now much clearer in version 3.7.

The main advantage is that it allow faster renderings.
You don't need to understand it to use it.

>
> Your post reminded me of one note I forgot to mention. On reading the
> image file, it is possible to play with the gamma option - even outside
> the range of common gamma values. You can get interesting scalloping
> effects. Especially true where the image is used for flat surface
> displacements.
>
> Bill P.

Isosurface tip: Always make the containing shape as tight as possible.



Alain

Post a reply to this message