[bathsheba]

This is about 3 hours. The remaining noise is clearing out slowly, but the stripes indicated by the arrows are not going away. I ran a similar render last night for 12 hours, and the stripes are still there.

How can I get rid of them? Or is there an aliasing bug?

-Sheba

[oscarMaxwell]

Hi Sheba,

Please, can you send this scene to maxwelltech@nextlimit.com so that we can see what is happening?

Thanks.

Best regards.

[Thomas An.]

I may be wrong, but it looks like the natural highlight from your light setup. Judging from the other rectangular reflections from your emitter, it seems like the logical highlight location and shape on those sharp edges. Possible?

Yes these look like highlights to me ... in which case they are correct.
It would be alarming if they were missing.

[zoppo]

i don't think it's about the highlight - it's about the "scratchy" appearance of this highlight. look a few cms below the lowest arrow - there is a perfectly sharp highlight.

[Thomas An.]

i don't think it's about the highlight - it's about the "scratchy" appearance of this highlight. look a few cms below the lowest arrow - there is a perfectly sharp highlight.

Well, if that was the intention then her wording is not accurate:

...but the stripes indicated by the arrows are not going away

She wants the stripes to dissapear.

[rivoli]

She wants the stripes to dissapear.

actually i think those reflections are severely aliased, i think sheba was expecting maxwell to take care of it after reaching a good sampling level. it looks like maxwell has some antialiasing issue with very strong and very sharp speculars.

[rivoli]

increase the resolution, it will be a non issue.

yes, this usullay helps.

edit:
i'm having some serious spelling problem, what the hell means usullay?

[DELETED]

DELETED

[Thomas An.]

Stripe is "a long narrow band" ... the only long narrow bands I see are the highlights themselves.

But you are probably right ... I suspect (and hope) she means the aliasing... but in that case the word "pixelation" (or "jaggedness") would be more fitting.

[bathsheba]

I may be wrong, but it looks like the natural highlight from your light setup. Judging from the other rectangular reflections from your emitter, it seems like the logical highlight location and shape on those sharp edges. Possible?

I agree the highlight is in the right place, but the way it's anti-aliased makes it look jaggy, like there are one-pixel blue stripes across it. Maybe it doesn't look like that on every monitor, but I see this on two LCD displays and one CRT.

It's an especially annoying type of defect because shrinking the image doesn't make it go away.

-Sheba

[bathsheba]

Hi Sheba,
Please, can you send this scene to maxwelltech@nextlimit.com so that we can see what is happening?

Thanks for the reply! I sent the mxs file, let me know if that's not the right thing?

-Sheba

[bathsheba]

actually i think those reflections are severely aliased, i think sheba was expecting maxwell to take care of it after reaching a good sampling level. it looks like maxwell has some antialiasing issue with very strong and very sharp speculars.

That's what I meant, I'm sorry for the ambiguity. I've seen other posts that mention this issue in passing, but since this image seemed to have a particularly bad case of the jaggies, I thought I'd hang it up and give it its own thread.

I'm mostly happy with what Maxwell has done here. The shadow sucks, the background bitmap is pixilated, and the square emitter reflections are clunky, but those are my problems not Maxwell's. Otherwise, the steppy highlights seem to me like the worst feature of a juicy render.

-Sheba

[bathsheba]

I do love these mathematical models though... thank heaven for rapid prototyping, eh?

Yep. But working with Maxwell is depressing me: when this render is finished everyone will want to know why they can't buy the pretty blue plastic piece in the photo.

PS, it's not really a mathematical model, I just drew it.

-Sheba

[Hervé]

It was me that had problem with AA... and so it was not fixed.... ahhhh...

look here.... when a plastic surface needs long edges highlights... they take the stairs... I'd say... take the elevator... !!

here it is another case...

http://www.maxwellrender.com/forum/viewtopic.php?t=3421

you'll se in that thread lots of people posted ...

well just so you know the solution is to render it much bigger... and it's gone...

Frances also had that problem some time ago...

[Mihai]

I think this problem is present in other renderers as well, it's just that they have different sampling techniques which perhaps Maxwell can't use, or hasn't implemented yet.

The problem I think is not really thin sharp speculars, but thin lines (in general) that are almost horizontal or vertical:



So one possible "fix" for now would be to slightly rotate the camera in either direction.

Bathsheba, could you please try and see if that takes care of those lines in your particular example? Just rotate the camera say 10-15°. The canvas can easily be straightened later in PS.