|
|
On 7/24/2018 6:15 PM, Kenneth wrote:
> Mike Horvath <mik### [at] gmailcom> wrote:
>> It's all working now, thanks to you all!
>>
>> I've attached the latest test code and render.
>>
>
> That's some really nice math work and problem-solving. I played around with the
> bowl thickness and object scaling in your code, just to test the 'uniform wall
> thickness' idea-- which works well!
>
Thanks!
> You might try scaling the final bounded_by object to more closely fit the
> resulting isosurface (for the sole purpose of making the scene render a bit
> faster.) Even though the initial two isosurfaces' contained_by spheres can't be
> scaled (that is, *non-uniformly* scaled), the final bounded_by sphere can. For
> my render, I used <2.0,0.8,0.7> in the two initial functions...
> function {f_spheroid(x,y,z,2.0,0.8,0.7)}
> and
> function {f_spheroid_normalized(x,y,z,2.0,0.8,0.7)+surf_thick}
>
> .... with surf_thick = 0.4, and bounded_by{sphere{0,1 scale <2.0, 0.8, 0.7>}}
>
> This works well; I even tested the resulting (hidden) 'bounding box' shape with
> min_extent/max_extent, to make sure of the close fit.
>
> Of course. maybe a scaled BOX for the bounded_by shape could be an even closer
> fit (to coincide with the differenced PLANE object in your code.)
>
> I did notice some artifacts, in the self-shadowing area of the bowl. From
> testing various things, it seems to be solely due to the isosurface 'accuracy'
> value. I changed that from 0.001 to 0.000001, which appears to eliminate the
> Moire patterns there. (I actually don't know how *small* a value that accuracy
> can be, before it has no further effect.)
>
> I was going to attach an image here of my tests, but I can't (using the web
> interface, anyway.) I'll post it to the images section instead.
>
>
>
I'm not too worried about the artifacts, since 99% of the shape is
hidden behind another texture. Only the edges are important to me. (See
globe render in p.b.i.)
Mike
Post a reply to this message
|
|