glass with liquid and ice cube - to improve

Started by LayC42, February 26, 2017, 12:48:30 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

LayC42

Hello people with an eye for good renders.

Currently I'm working on an example for a glas - liquid - ice cube - material study. And I need some c&c to find the right material and values.
The scene was modelled in creo parametric with surfs to split them along the borders of the different materials. I used a description from Richard Funnel to find the right material types and IOR values. The normals should be in the right direction.

There are some different unfinished things to improve. I attached the scene as a bip (KeyShot 6.3), 'cause I didn't use any texture and linked the default environment (Dosch Apartment).

The caustics aren't perfect 'cause of the bloom radius. And the right bottom edge of the glass is blown out to the background. But this is the raw render (no post).

So thank you all for the helpful advices.


cheers
Marco

Will Gibbons

Nice. Lots of the results here will depend entirely on your lighting setup. Check how photographers shoot this type of object and emulate their lighting. I think you'll be surprised of the results.

mattjgerard

Yes, good advice. Also remember that no surface is perfect, especially ice cubes, and the caustics you get through imperfect surfaces is what makes the render more believable. Some noise displacement on the ice cube will do wonders for your light transmission.

LayC42

#3
Hello Will, hello Matt.

Thanks for the comments.

I've changed the environment to a simple studio hdr ("Light Tent White Open 1 2k") and added two procedural textures (scratches and cellular) to the bump channel of the ice materials. And also changed the camera angle a bit.

Still no post work and I changed the gamma to 2.2.

Now I let render two version with the difference of polygon and NURBS mode.


mattjgerard

Cube is looking much much better. One logistical note, unless that is a very viscous dense fluid, the cube would be floating a lot lower in the liquid. Almost completely submerged. that will help with the reality of it. What will also help is if you can place it in an environment. On a table or something, to give it a purpose, a place, and some reflections that are real-life. Not sure what your end purpose of it is, but I am a big fan of placing objects into environments to cast the illusion of reality.

LayC42

Quote from: mattjgerard on February 27, 2017, 01:05:22 PM
Cube is looking much much better. One logistical note, unless that is a very viscous dense fluid, the cube would be floating a lot lower in the liquid. Almost completely submerged. that will help with the reality of it. What will also help is if you can place it in an environment. On a table or something, to give it a purpose, a place, and some reflections that are real-life. Not sure what your end purpose of it is, but I am a big fan of placing objects into environments to cast the illusion of reality.

Hello Matt.

Thx for the comment. I thought about the position of the ice cube. But I wanted its top above the liquid surface. I know what you mean that an ice cube is much heavier and sinks more to the sub surface. I'll have to give this a try tomorrow.
This model / scene is a preparation as a training example. It's intended as a material study.

I load up some pics with post work in PS.


Every c&c is welcome!


cheers
Marco

mattjgerard

Looking great, i've done these studies before where I just keep trying to improve and make it look better incrementally. I learned a ton from doing those.

I should do more!

The only other thing that is catching my eye right now is the wide tonal range of the gradient in the whisky. it goes from a very light tan to a dark brown, and while I'm not a whisky expert in any sense, my eyeball doesn't like that vast of a difference. At its lightest, it is almost clear, and that isn't computing with what my eye is expecting looking at a tumbler of whisky. Maybe try to pull the color difference a little closer so the gradient isn't so drastic. It might be your lighting setup that is causing that, and like I said before, I have a hard time pulling reality out of images on white. So, take that with a grain or 2 of salt as to whether that comment merits any validity or not.

cool to watch the progress!

matt

Will Gibbons

Hey, definitely some progress. You shouldn't see much of a difference with NURBS mode vs Non-NURBS mode. As long as your tessellation is high enough to not see faceted surfaces, no real need for NURBS... it'll just take longer to render. As for changing the Gamma, also, I'd avoid that and just work within the HDRI editor. I did a whiskey glass a while ago here: https://www.keyshot.com/forum/index.php?topic=12600.0#lastPost and that bokeh background you see is an emissive material pushing colored light through the transparent materials. I found that to help. Also, dropping the HDRI brightness and lighting with a physical light can really help. Good luck!

LayC42

Hello Matt and hello Will.

Thanks for the comment. I was surprised to see a difference between poly and NURBS mode. I have to rework the CAD data as well, 'cause the normals directions. Then I can change the position of the ice cube.
And if all is prepped up, the scene can be build from scratch.

If I found the time to fix this scene,  I'll post the result.

This example was an unlucky one.


Enjoy the weekend!


Marco

Will Gibbons

It's all about learning! Which, it looks like you stumbled upon some good insights with this one. Also, Normals can be flipped in KeyShot... not sure if you've found that in the geometry editor. Just thought I'd mention.

INNEO_MWo

yeah Will you're right. But I had no luck with it. But the problem occurs by direct import the creo file and not, when I use the plugin (latest version) from creo to KeyShot.


I have to look into the file an do some tests.


Here's the KeyShot 5 example without Photoshop correction. I did some improvements in creo as well in KeyShot. (and I used KS5, 'cause of the customer)

mattjgerard

whisky color is looking much more believable. Nice work.

Will Gibbons

Pretty clean! Only think I'd mention is the 'perspective' seems a bit low. Could use a wider lens 80+ in my opinion. Also, I think it cold be improved if it was rendered with something other than a pure white background (even if it was like, 95% gray), but I'm assuming the client demanded a white background. I only say that because caustics might look better on a non-white background.

LayC42

Thank you Will and Matt for your suggestions. That helped me a lot.

I think that I haven't the time to recreate a newer version of this. But I will keep your ideas in mind!

Cheers
Marco