[itsallgoode9]

Regarding Maxwell animation render and noise, what is the best way to render to get rid of tha"sparklely" noise in the highlights? I have only rendered one animation with Maxwell and it worked fine for the most part, although the areas with highlights were a bitch to get rid of noise. I sent it to The Ranch but budgets did not allow time to get every frame as high SL as it should be. So I ended up with the sparkly noise thing. Should I be rendering with passes so I can blur the reflection layer enough to get rid of that? Or would that take just as much time as rendering in single pass but to a higher level?

[Aniki]

can you show an example of what you mean exactly ?

Did you use sunlight/physical sky on this ?

cheers

Aniki

[itsallgoode9]

http://vimeo.com/15910826
http://vimeo.com/15910780


there are links to the two animations. I was using planes as emitter with HDRI emitter applied o them. The issue of the noise just has to do with the noise in an image being random and kinda moving trhought the animation, creating a kindof sparkly effect in the highlights. I was able to get it toned down a bit by some blurring and noise reduction in after effects. The truth is, if I had let these render to a shading level where the noise was reduced enough, it would have doubled or tripled the cost of the animation which is why i'm looking for better and more efficient methods of getting a clean render.

[Aniki]

Hm, I see.

right now, it looks to me like the highlights are overbright anyways, for what the reason is hard to tell. ( vimeo compression, AE noise reduction applied already etc..)

I'd say use a well balanced light, with rather even lighting situations, not to saturated and contrast driven as you'd want it to be looking finally in the end.
That could help reducing the noise in general. Try to avoid overly bright highlights and too dark, detail-loosing shadow areas.

This way you could colorcorrect the results ( AE even reads MXIs now to the desired look afterwards, yet avoiding getting noise issues due to edgy/harsh lighting situations for maxwell render to calculate..

just a suggestios.

also I'd try to render the whole animation with either the same cpu ID, or depending on how many machines you use in total, with concurrent cpu IDs for every machine and mximerge in the end.

thats all I can say for now.

cheers

Aniki

[itsallgoode9]

Thanks for your response, although most of the suggestions won't work for me as they are about changing the lighting look to account for render issues. When it comes to th animation renders where I'm interested in getting rid of noise they are for marketing/advertisemng materials, so there isn't room for compromise in the way the lighting is. What the client requests, the client wants.

With the CPU ID thing I do have a couple questions, as this is something I haven't fully looked into yet. When using a renderfarm, such as the ranch, is there a way to utilize this? Also, at home I have a 6 core CPU, is the CPU idea only useful when it's different physical processors or or can/should it be used in multi cote CPUs as well n


Thanks fo your help

[Aniki]

You got me wrong there, Justin:)

Its not about changing the lighting look, its about rendering everything lighted more even, like maybe utilize a soft ambient in addition to the rest. Afterwards you colorcorrect the result to the clients wishes. So the client never gets to see the in between steps:)

In general, when you export an animation sequence to mxs, the cpu ID should stay fixed. Its just if you, say split the whole clip into several parts, and the cpu ID varies, it could result in inconsistencies..

Besides that, I think the mxi merging really could be an option here. Like rendering the same image 10 times to a lower Sl( this time with a different cpu ID iff course) and merge the results, so you end up with the desired final SL.

These are all general ideas, that might help in your special case, but not necessarily do.

The problem you got can be the result of many things, eg shader setup, too harsh bump, antialiasing in high contrast areas, etc..

You could run more maxwell instances on your 6 core machine and apply different cpu IDs to each mxs, too.

Maybe you can post 2 or 3 sequential screenshots in full quality, so others can have a closer look too;)

Cheers

Aniki

[3dtrialpractice]

ya post hires still as i dont see much sparkly noise in the vimeo compres vids...

also.. regarding highlight intensity on an object.. i used to be annoyed how the highlight would be "noisy" . i figured it need to render to higher SL.. but then i started messing around with the "k" setting on the Nd,K parameter.. by default it its.. 0.. but even setting it to 1-10 greatly influences the look/strength of the highlight (spec highlight reflection of an emitter).. so you may just wanna try bump up your "K" value to 1,10,50,100 and you may get cleaner better highlights...

WARNING: after you bump up your K value.. depending on your materiel setup.. setting the "K" too high may desaturated/whiteout the look of your material.

-Luke

[itsallgoode9]

Thanks for the suggestions guys, I'll try both of those out in future animations.

As far as sparkly noise showing up in the animation, I was able to pull most of it out in After Effects, although if I was doing something that had more hard edges, such as graphics or something, it would be a bit more iffy if I could get the sparkling in the highlights to disappear.

I'll try those out though...thanks guys!

[Bubbaloo]

Is your material using unrealistic roughness settings like 0? You should keep it at 2 or 3 at the least, and maybe duplicate the BSDF with a higher roughness of 10 so that it will blend the two roughness settings and it won't be so perfectly uniform.