wierd interpretations of screens and louvers

[chadsmith]

Does anyone know how to prevent the louvers in the final image from being translated into diagonal swooping psychedelic curves. Its as if the lines are too close together and so it interpretes horizontals biaxially. This is not a mesh construction issue just a final image issue.

[Ricardo Eloy]

Are you talking about the moiré effect? If so, there's no way you can avoid it with V iz alone (not that I know). As a matter of fact, one of the "great" features of finalRender Stage 0 was exactly this: it could avoid this weird effect.

BUT, if this is not what you're talking about, forget what I just said...

[]

Rick

[chadsmith]

YES THE "moiré" EFFECT!!! does anyone know how not to have it in the viz renderings short of buying additional plug-ins like final render?

[abicalho]

Just a heads up that Moiré is unnavoidable in the "pixel" world. Why? You're trying to make a matrix of pixels adjust to horizontal and vertical lines, and they won't necessarily scale correctly, especially when the lines tend to be the size of 1-2 pixels.

 

Wanna see a real world effect? Get a camera and go somewhere make shots of tiles. You'll see tons of moiré in the film.

 

What can you do to minimize this effect? You can Supersample the materials, so that the lines are less prone to cause Moiré effects.

 

Hope this helps,

 

Alexander

[quizzy]

Yep supersample helps if the moire occurs in a mapping, but when you have qeometry it won't work.. too bad...

[STRAT]

if it's the mesh then your pretty buggered, best render at a higher resolution.

 

if it's the material you got 3 choices:

 

1) try different renderer filters, or

2) turn on supersampling, or even better -

3) in the material editor, in the diffuse slot where the bitmap's filtering properties are, change the filtering from pyramidal to summed area (using the blur spinner to fine tune). this almost always solves the problem

[chadsmith]

THANKS EVERYONE!!!!

[nisus]

render at a large scale and rescale in ps

 

nisus

[ingo]

Just curious, why not simply use a higher aa setting ? :???:

[STRAT]

Originally posted by ingo:

Just curious, why not simply use a higher aa setting ? :???:

because if it's the material creating the offending effect then aa makes not a blind bit of difference

 

[ April 08, 2003, 01:36 AM: Message edited by: STRAT ]

[ingo]

Huh ? There is no difference between material or object, aa is mostly a post effect in 2D so it shoudn't make a difference, at least for that louvres.

[bigcahunak]

Maybe change the louvres map size (e.g. 512x512 instead of 256x256) or change the spacing in the map in PS.

[quizzy]

If I understand the question correctly Chad is doing those louvres using geometry not a mapping. So what nisus pointed out could work, just render at a large scale and rescale in PS.. Anti-aliasing could also work..

[quizzy]

Well I just did some testing with different anti-aliasing filters, and this is what I found. All fitering values are set to default values.

 

area

 

blackman

 

blend

 

catmull-rom

 

cook-variable

 

cubic

 

mitchell-netravalli

 

plate-match

 

continued in the next post due to the limit of 8 piccy's in one post...

[quizzy]

here we go: PART 2

 

quadratic

 

sharp-quadratic

 

soften

 

video

 

How is this for a long nisus-style post!! :angerazz:

 

And strat, these filterings also work on mappings in MAX5, don't know if it filters out the moire effect, but you can always go with supersampling and summed area mapping as you pointed out above...

[nisus]

you faked!

 

nisus-long is full with words that actually mean something... this is what I call a quizzy-fae-long-post ;-p

 

nisus

[STRAT]

i call it spam

[ingo]

...but Quizzy's nisus-long post are easier to read :ebiggrin:

[quizzy]

and people can understand them too....

[nisus]

How is this for a long nisus-style post!!

You cheat ;-p

[STRAT]

i use catmull-rom as per standard anyway - seems to give the crispest results.

 

have you got any render times on those filters to compair?

[nisus]

Good point strat. Quizzy, can you provide us some specs?

 

 

nisus

[quizzy]

well guys.... I didn't write down the render times, and this is not a good scene to test the different AA settings on.... But yes, it does affect the render times, and sometimes in a big way...

Maybe if I find the time I'll do a larger scene with more polygons and do a comparison of the different AA... that should be in the next 50 years or so....

[ihabkal]

Maybe change the louvres map size (e.g. 512x512 instead of 256x256) or change the spacing in the map in PS.

I guess then the solution is custom made for the custom louvers in each rendering.

[ihabkal]

someone should do the same tests with mental ray's anti aliasing opptions...

can the one who posted all these image tests do it?

thanks,

Ihab