[tom]

Hi Dennis,

The reflections on the water are not ment to be caustics, they are specular reflections. I see there are glasses in the windows and lights are coming from interiors. For this reason it will take more samples to see their reflections on the water. At least I can say the sampling level you've reached (11.47) is not enough for such complexity. Also know that enabling/disabling caustic layers won't affect render speed as in biased renderers. So I suggest you not to turn them off anyway. Your scene seems like it will surely need caustic layers to be fully turned on and have to be sampled at least 18 levels.

Best regards,
Tom

[VisualImpact]

At about sl 16 is when these reflections from the lights show up.
They will appear in time. Great looking render by the way.
What size are you renderering though, that seems to be a long time for sl 11.
Image below was about 28 hrs, sl 18, 800x600.

[tom]

Reflections appear by the calculation of bounces and unbiased way makes this process end forever.
Depending on your scene complexity, possible reflections may take longer to become visible.
This example simply demonstrates the increase in visibility of reflections among samples.

[tom]

Dennis,

- Yes, you can resume previously saved MXI files and keep rendering more.
Unfortunately, it's not possible to change parameters while resuming.

- It's true that with more light interactions (complexity) it will take longer to render.
The best way to reduce render times is to keep emitters at low poly and
using diffuse shader instead of plastic when not necessary.
Also do not forget to keep the base plane as small as you can (closer to
bounding box of the main object and scene total)

I advise you to test render your scene step by step, like only building (I mean
just diffuse walls) with no details, then add lights, add other objects etc...

Best regards,
Tom

[tom]

Thank you Dennis, I really enjoy this scene and wish the best for you.

[tom]

u.biq,

The word "forever" is a theorical approach, practically what you see after reasonable samples should be enough. Actually in real life with human visual system and light speed, we don't see the final thing. It's just the matter of samples and time. I agree it's not good to wait very long time to see such a solution and as you know Maxwell is still under development, so you'll always have better versions in time.

Best regards,
Tom

[tom]

u.biq,

I agree with you, too. However it's impossible to judge the situation with just looking at the result for me. I suggest you to try hiding everything other than room and mirror, then render and see what happens. When I have problems I quicky focus where the problem exists and simplify the scene temporarily for testing purposes. Otherwise, it will be very hard to render the scene many times.

Best regards,
Tom

[Intuition]

I wish Maxwell rendered as fast as your sample animation.

F-Prime style.

I eventually figured out Maxwell sees these reflections better as the sampling goes up. For a while I just thought Lightwave was not sending the right surface data.