Told of at the beta test group, beta 6 release messages.

This image shows (Windows) official 3.7 rendering at top, beta 6 bottom. Both
are from the 64 bit program, and I haven't tried anything else.

Default 'tolerance' was used for both. Vastly improves if that value is changed
to 0.3, although I don't believe it gets better upwards of that.

Post a reply to this message

Attachments:
Download '37_beta6_comparison-sphere_sweep.jpg' (31 KB)

Preview of image '37_beta6_comparison-sphere_sweep.jpg'

"omniverse" <omn### [at] charternet> wrote:
> Told of at the beta test group, beta 6 release messages.
>
> This image shows (Windows) official 3.7 rendering at top, beta 6 bottom. Both
> are from the 64 bit program, and I haven't tried anything else.
>
> Default 'tolerance' was used for both. Vastly improves if that value is changed
> to 0.3, although I don't believe it gets better upwards of that.

And here's a look at the b_spline and linear sphere_sweep, which also show
artifacts but something surprising to me about that.

The linear has only minor trouble; and the b_spline, while almost as bad as
cubic, will clean up entirely if tolerance 0.1 is used.

However, nothing rids the linear type of the smaller artifacts seen on the right
end of this sphere_sweep. I tried small tolerance and large tolerance and it's
appearance stays the same.

The camera location is a little to the left (or -x) of there along the z axis,
so isn't in direct alignment. look_at is a little more toward left too.

Sorry about image size, had to be this large to see all clearly.

Post a reply to this message

Attachments:
Download '37beta6-sphere_sweep-b_spline-linear.jpg' (57 KB)

Preview of image '37beta6-sphere_sweep-b_spline-linear.jpg'

Am 15.05.2017 um 07:25 schrieb omniverse:
> Told of at the beta test group, beta 6 release messages.
> 
> This image shows (Windows) official 3.7 rendering at top, beta 6 bottom. Both
> are from the 64 bit program, and I haven't tried anything else.

This is a known regression, and will possibly /not/ be fixed in the
final 3.7.1.

The background story is that these artifacts had been around in 3.6
already. We did make an attempt to fix them in 3.7.0, but we found out
just recently that in doing so we had broken other things.

So we removed the fix in beta 6, and are now back at square 1.


The sphere sweep updates in beta 7 are unrelated to these artifacts.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Am 15.05.2017 um 07:25 schrieb omniverse:
> > Told of at the beta test group, beta 6 release messages.
> >
> > This image shows (Windows) official 3.7 rendering at top, beta 6 bottom. Both
> > are from the 64 bit program, and I haven't tried anything else.
>
> This is a known regression, and will possibly /not/ be fixed in the
> final 3.7.1.
>
> The background story is that these artifacts had been around in 3.6
> already. We did make an attempt to fix them in 3.7.0, but we found out
> just recently that in doing so we had broken other things.
>
> So we removed the fix in beta 6, and are now back at square 1.
>
>
> The sphere sweep updates in beta 7 are unrelated to these artifacts.

Wups. Thanks for saying that, because these were actually done in beta 7!

Too bad there's so much trouble with sphere_sweep. It always was one of those
problematic things, I know. Just so useful an object. I was thinking it was
finally tamed.

Post a reply to this message

"omniverse" <omn### [at] charternet> wrote:

> [in 3.7.1 beta 7]
> The linear has only minor trouble; and the b_spline, while almost as bad as
> cubic, will clean up entirely if tolerance 0.1 is used.
>
> However, nothing rids the linear type of the smaller artifacts seen on the right
> end of this sphere_sweep. I tried small tolerance and large tolerance and it's
> appearance stays the same.
>

Those linear_spline sphere_sweep artifacts are a suprise (to me.) I never
noticed them before (I always thought linear_splines were well-behaved.) Does
moving the camera around change their location? If you can run the older v3.62,
do they show up there as well?

Post a reply to this message