[polynurb]

thanks jason for your reminder..

.. to chuck something into the pot..

basically i am having a few days off work, and time to do some testing with 3dcoat & maxwell 2.6

all work is done on a 12" core duo tablet pc... it's the machine i took with.. got it for the sake of painting textures in PS, but i was amazed how well 3dcoat (and sculptris too) are running on it.. so right now i see about how far this little cpu can take things. anyway resolutions are somewhat reduced..

in rhino i started to model a coffe cup made of glass, one that has a double shell for insulation.
initially i was thinking of a series of glass tests, redoing the Liquid in Glass and other object in object scenarios with ~2.6...







but i went another way and discarded the testig series, in favor of filling some coffe into the cup and trying to sculpt some crema on top. as pictured in image1 i do not use the legendary LIG setup, but a microgap variant because i wanted to investigate behaviour of 2.6 in this respect... in the end for animated liquid this is the only way.

all the materials are quickly made starting from the rhino wizard:
chrome, lowgrade glass, and a modified orange juice for the coffe.

concerning the sculpting part, i haven't gotten much into the voxels of 3dcoat yet, until now i used it mainly to paint on existing meshes or to create their uvs.

..for the start a flat rounded voxel plane:



..adding "bubbles" with the sphere tool...



and cutting them away with the inverted tool option:


using cup geometry to cut the sculpted surface. this seems very strong in 3dc. i have had so many issues with mesh (and nurbs) booleans in different software packages, but voxels seem to solve better, and i could cut the high res part without problems.



then exporting from 3dc as obj into rhino, adjusting some scale and smoothing the whole mesh some more, it looked like this in the rhino viewport :
(for anyone concerned about the shading, rhino 5 can with a little tweak, display HW shadows on an intel integrated !)



after placing the whole setup looked like this..
noticeable i still scaled the "crema" into the glass a little bit... in order to see what it looks like



the light setup is as simple as it gets,.. skydome lighting to help the SSS in the coffee and to prevent sharp caustics...
the crema material is derived from the frosted glass preset.. although quite strongly modified,the idea was to avoid using another sss material, but still get acceptable translucent effects... and a frosted glass surface reminded me a bit of tiny foam.

here are two layers differing only in color and bump/roughness maps:
planar and spherical mappings are used on the object, all uvs setup in rhino.



now the first try showed the brew looking like this..:

i know... not exactly the kind of stuff you'd want to buy at the coffeeshop..
i guess i missed the scale .. and the threads in the middle make it look a bit like milk gone bad

(sl 13.6/2hrs)


never the less, i let it sit over night for further investigation..

(sl 16/~9hrs)


(sl 16/~9hrs)



..now the "old milk" look gave me the idea to solve the problem by adding some proper milk foam...
so i went back to 3dc to add some more smooth geometry to combine with the existing..


i gave the whole thing a twirl motion and combined both sculpts in rhino...
looking like this:





the milk foam is the crema shader with higher transmisson and white colors
the whole study was more to see how these shaders would behave in complex overlapping scenarios.. with sss/and dielectrics ....than to create the perfect barista macchiato..
so albeit still not super munchy.. this is about as far as i will bring it for now...


(sl 18/ and some PS)

[Polyxo]

Hi Polynurb,
that's a great challenge you chose and especially interesting as you use a technical program as the Render-Host.
I think these are very interesting approaches - but they at this point still fail. But hey - Coffee with Crema in a Glass - that's
nothing to do easily, at least in a physical modelling manner.

The first things which come to my mind: The liquid is totally off. With real Glass it really looks as if there was actually no Glass...
The liquid visually expands to the outer Surface of the Cup. I think the proper strategy is still to let the Container and the Liquid truly intersect.

Concerning Crema: I would guess that one came closer by using a good photo of Crema and trying to create a Displacement-Map from it.
This map might get used directly or could get loaded into 3DCoat to actually deform the Geometry and to do some additional tweaking.
Then you already have a corresponding Diffuse Map which you could use to paint the mesh in 3DC.
The foam structure would still remain difficult to simulate, it surely needed SSS - I leave it to HalfLife or other Nerds to get specific with that.

[Half Life]

Apparently I've been elected king of Nerds

I also think this is an amazing challenge -- and I'm very excited to see it's progress.

I might have 3 separate parts to the milk sculpt with 3 different material strategies:

1) Bubbly Foam (more opaque as you have now)
2) Where the foam melts into the surface of the coffee (a masked gradient from foam to coffee colors/materials)
3) the swirling tendrils sinking down inside the coffee itself (sss for sure)

I'm very impressed with what you are doing with 3DCoat here -- I've often wondered if there wasn't some way to use the voxels as voxels in Maxwell since everything gets voxelized anyway... maybe something for the Wish List .

Best,
Jason.

[Jay325]

Your brains scare me

[polynurb]

thanks for the input guys!

i share most of your critique ...

concerning the glass/liquid, here are some example shots of similar configuration:





although there are 2 layers of glass behind each other, the refraction is low... imo the liquid is just not extending far enough..there should be something like an almost smooth ellipse.. but somehow the whole refracted part is stretched/shifted downwards..... the other problems shows where the crema sticks inside the glass:





i do like the effect it creates, looking like it is sticking to the glass.. but refraction is wrong and extending towards the actual edge of the glass.. i will see about how a complete intersection would look like, but i doubt it will be the correct way either.

for the crema i agree that displacement would be the way to go, as it might be possible to actually render every little bubble for real... but remember i was doing this on a core duo.. so not option at this time.

i have pixplant/shadermap.. so i might check out how this goes along.. although there is always the limitation of undercuts..
.. i mean perfection is an endless road.. vector displacement.. even more possibilities


in the meantime i rendered out another view, with adapted milk shader. i had some issues in one of the roughness channels.. working with maps here i always found very tricky.. just 5% contrast/brightness +/- can make a huge difference.
so, the grey stripes are almost gone, and i didn't have to remove them in PS like in the last image.

over all i am more happy with this image.. shot with f4.0.. so more dof and drama.. starting to look tasty..?!

[polynurb]

another test with high res, same scene/mats, + a bit PS