[tom]

Could you please try downloading the file again? (Flush your browser cache if needed.)

[Aniki]

ok, worked.

Interesting, as the caustics themselves are blurred due to the distances/size of the emitter. Is it possible to get nice and crisp caustics like they would be via sunlight too ?

So caustics are only correctly calculated by setting up a scene how exactly ? Only by emitters, not via hdri/ sun ?

Are there some general rules like emitter sizes/intensities to obey ?

I was so happy by the speed/ quality of the caustics the sun produced after 2.0. Only to recongnize that they are not calculated naturally behind transparencies. I hope this issue gets fixed soon.

Big thanks for the insight, Tom.

Aniki

[JDHill]

Only to recongnize that they are not calculated naturally behind transparencies.

It is precisely because they are calculated naturally that they take so long. If you could duplicate the scene in real life and then shoot only a small, finite number of photons at it, you should expect to find exposed on your film a similar distribution. As such:

So caustics are only correctly calculated by setting up a scene how exactly ? Only by emitters, not via hdri/ sun ?

...should read: "more quickly calculated by...", and the answer would be: the more parallel the light, the longer the calculation will take. It is a question of probability; parallel rays like those from the sun are by definition, all very alike, and reduce chances to a very low order.

[Aniki]

so there is no way of calculating them any faster ?

[Bubbaloo]

so there is no way of calculating them any faster ?

Only one way; faster computer, unless you don't care about accuracy, you can use an engine that interpolates the solution.

[Aniki]

nah, theres no single machine where SL 35 makes any reasonable sense

I'd rather wait for NL to optimize like they did with the sun caustics.

cheers

Aniki

[Lars Magnusson]

Hm, ok... I almost understand the technical descriptions.

Sooo, small tiny emitter or sun, fast caustic pattern, but if you look at the caustic pattern through glass/water it takes extremely long time to see any result.
(did a test with Aniki's scene but used a far-away tiny plane with strong emitter, wobblysphere caustic pattern looked good already at SL10, 1min, through AGS glass but was at that time black when looking through the "real" glass)

And when I made the emitter plane larger and closer (similar to Tom's scene, and my previous scenes) the blurry caustics seen through the glass/water clears faster.
(not as fast as tiny emitter and no glass in the way, but still rather fast)

I have no doubt that sooner or later NextLimit will come up with some magical (yes, consider math and optimizations to be a form of strange magic ) way to speed it up.

[Lars Magnusson]

Friendly Alien - WIP by LarsMagnusson, on Flickr

Slow progress this weekend.
Starting to get used to the hardsurface sculpting, so more detail done on armor.
Camera moved back to show more of it, and switched to the default grey clay material.

And the nose was altered.....yet again

[Bubbaloo]

For some reason, it's starting to look like Michael Jackson to me.

[David Solito]

For some reason, it's starting to look like Michael Jackson to me.

[Hervé]

ahahah good one Brian...

[Lars Magnusson]

He,he, considering the amount of nosejobs I've given him, it was probably inevitable that similarities would occur

Hm, I think I'll stick with this type of button nose... nose-less looked a little strange and was difficult to make "believable", and flatter nose would make it too much monkey like...the jaw and lips makes me think of monkeys and apes...even though that wasn't intentional.

And a little ZBrush tip:
When working with HD (highdensity) sculpting, DON'T lower the HD subdiv level before "picking up" with "A".
The higher HD levels seem to be lost then.
So a lot of work I did when resculpting skinstructure/pores yesterday was lost

And for those who are not familiar with HD sculpting in ZBrush:

My face object has ordinary subdiv levels up to about 1.7 Million polygons.
Then I created a couple of HD subdiv levels (up to about 30 Million polygons)

The HD detail isn't visible when I work (would be to straining for the graphics card), but by aiming with the brush at the nose and pressing "A" I get a circular workarea where the HD polygons are visible and sculptable.(this area size depends on how high I went with HD, 60 Million polys would give me a smaller workarea)
Then after sculpting pores and stuff, pressing "A" again gets me back to the ordinary 1.7 million model. (and HD data is stored in model)

The HD data is only visible when rendering with ZBrush,
BUT it's also taken into account when generating displacement/texture/normal/bump maps for Maxwell.

[Lars Magnusson]

Friendly Alien - WIP by LarsMagnusson, on Flickr

Weekend fun:
Subsurface scattering has been tweaked and decreased displacement height a bit.
Sculpted surface structure, started with texture and for specularity map I used a slightly modified displacementmap.
( Rendertime 2.5hours, SL 9.5 )

Specularity map and balance between texture and SSS needs more work.
And I have to think about how the armor should look...I like metal, but dirty worn plastic is probably better.