Indigo Renderer Forum

Ok, this is a test for rendering realistic water.
Depending on the results, this should serve as a guide for achieving realistic renders of glasses containing liquids in future versions of indigo.
Below is a diagram with some associated requirements for the model.



Requirements:

a) need sufficient polygonisation of the curve here because sharp edges cause shading normal artifacts. For example 10-20 points may be needed on piecewise curve.

b)distance should be > 0.2mm

c) distance should be > 0.2mm.

Note that the distance between the glass and water surfaces in the walls of the glass needs only to be greater than 0.2mm, it should probably be about 1mm, depending on the width of the glass

d) distance between glass and z=0 plane should be in range [0.2mm, 1mm]

e) meniscus should be modelled

Addtionally:

f) all geometric (given by tri winding order) and smoothed normals should be as labelled.

g) total number of tris is expected to be in range 10000-60000

h) surfaces should be two-manifold and closed (not self-intersecting).

i) water and glass should be assigned different materials.

j) part of the surface of the water component should lie inside the glass as shown.

k) The model should be created in units of meters.

CAN I TRY?

posso provare?

Yes, you can try modelling it if u want... in fact that would be awesome. Kosmo is already having a go tho...
Perhaps you could do the wine glass version?

No rush.. the code to render using this technique hasn't been written yet

ok nick here is my wireframe....i try to keep your spec..
the triangle counter indicates 13.000...later i will post and full wireframe..

I don't get the distance requirements...
In blender everything is relative, so how do I know if they are under or above 0.2mm?
Is that an absolute value, or are you saying that there should be 0.2mm distance in relation to a glass of 10cm (for example)

I added a condition k): units should be in meters.
the 0.2mm condition is there to keep the model robust against the small 'epsilon' distance I add to ray origins to avoid self intersections.
(which is currently 0.1mm IIRC)

kosmo: looks good.
Feel free to alter the cross section of the glass from my diagram to make it more realistic if needed.
Note that the distance between the glass and water surfaces in the walls of the glass needs only to be greater than 0.2mm, it should probably be about 1mm, depending on the width of the glass

"Note that the distance between the glass and water surfaces in the walls of the glass needs only to be greater than 0.2mm, it should probably be about 1mm, depending on the width of the glass"

aahhh..thats ok...i will fix...and i make some test with real materials...stay tuned..

Kosmokrator wrote:.and i make some test with real materials...stay tuned..

Ok, but bear in mind that this technique will look rubbish in the current version of indigo. Well the glass by itself may look alright

I've done some glasses some time ago, and they look quiet good i think
I had the liquid completly inside the glass (no overlaping between glass and liquid), is there a physical error in my render so?



greets zuegs

Yeah that looks pretty good Zuegs.
Yes, there is some error because the water-glass interface is not yet handled entirely correctly. That's what i'm working on in these tests.

ok Ono i make a quik render of 35minutes...here is the results...
lighting is exr map
glass ior 1.5
water ior 1.33
plane difuse mat..

what u think?

Done 79310000.00000 mutations (258.17057 mutations per pixel)
mutations_per_sec: 37105.61932 (26.95010 micro-secs / mutation)

I'll try the new method, just to see how it works on current indigos.

This is the current method (the posted maxwell one):


I increased the contrast, lost a little detail, but I quite liked it .

I'll knock up another version later.

Look forward to this update, Nick!

Waouh, the glass looks really nice here.

here are my test:

The glass:

The gap is exactly 0,2 mm betwenn the 2 material



The first probe, but the model is in millimeter and has about 120000 triangles



The second test, now the model is in meters and has about 60000 triangles


I found the difference big in the material, (it should be wine) what do you think about this?
The material I used are the same on both pictures.

Could someone give me some gould value for rendering wine?

Another question:
When I look at a glass of wine I see a grey shadow but red caustics, is this
possible to render with indigo?

i have a problem with my glass...i dont know if its exporter bug or my modeling bug or indigo bug...the left glass is green and the right red!the red glass is empty the green is filled with water!tha water rendered green...why?ono?some one explanation?

It's phisically correct. Remember this shot?
http://afecelis.gdlib.net/Indigo/sala03.jpg
in this discussion:
http://www.indigorenderer.com/joomla/fo ... .php?t=585

In the previous shot, I changed the color of the glass and it imediately tinted the whole interior. Same is happening in your scene. The glass is transmitting its color to the water.