[feynman]

I see. So there would be no time penalty for using GLASS glass only (instead of the AGS glass around GLASS glass trick)?

On another note - won't the bigger mock-sun radius make shadow even less sharp? In some setups, I need to achieve subtle but recognisable foliage shadows on walls and floor, such as in this example



that I was hoping to achieve by putting a clipmap of a tree on a plane just outside the windows, lit by the mock-sun.

[Fernando Tella]

...won't the bigger mock-sun radius make shadow even less sharp? In some setups, I need to achieve subtle but recognisable foliage shadows on walls and floor, such as in this example

Yes, it would.

[feynman]

Thanks; so that means for fairly crisp foliage shadows on interior walls and floors, one must revert to the physical sky plus sun method that, in general, renders much slower, right? So far, I have mock-sun = unrecognisable blurry shadows = no fireflies vs. physical sky plus sun = crisp shadows = many fireflies (I tried with a room - two walls, ceiling, floor - with no objects inside for speed).

[Mihai]

Regarding using a fake sun, check this post by Thomas:

http://www.maxwellrender.com/forum/view ... 52#p377752

So there you have your solution for sharp 'sun like' shadows without using the Sun from Physical Sky. You don't need to set it to millions of watts either. Just use only emitters with this one being stronger than the others and you'll get the fastest renders. Remember, it's always when you mix emitter types especially in interiors that things really slow down (noise from Phy Sky, and Sun will be most visible while light from emitters will look clean much faster).

For these glass top tables, I wouldn't consider it a sin to render in two passes. One with the glass top hidden from camera, the other with only the glass top and set it to hidden from GI. Combine renders in PS, using Multiply for example.



The advantages are that you get two very fast and very accurate renders, plus of course you can get colored caustics. This example rendered the caustics "pass" in 6 minutes to SL 13 on my i7 2600K, and the glass top only in 2 min. 10 seconds to combine the renders in PS.

It's really unfortunate that this caustics seen through glass thing gets in the way of appreciating how fast and accurate caustics really are in Maxwell. And I don't consider a really bad workflow to think that in some cases you CAN get them even seen through glass if you render in two passes and think a bit on what should be in each pass. In your case it's really simple.

[feynman]

Thanks, I might consider the two-pass solution next time I have to do glass tables. Now I have to do a fartoomanyzillion of luminaires (on/off)...

Regarding the mock-sun, what was suggested in the thread you linked works very well (90mm Rhino 6-polygon-disc as "sun" at 10m distance). However, after reading this http://www.maxwellrender.com/forum/view ... er+balance thread - what does "balancing" emitters mean in concrete terms? In my 1st attempt, I have a strong mock-sun and an HDR image to provide overall ambience (the room has only two walls, a floor and a ceiling); how would one "balance" emitters here to achieve the least noisy render? This "emitter balancing" aspect is mentioned in other threads but not explained in concrete terms...

1280x853px, 15min, SL 13.32, mock-sun emitter, HDR image based lighting

[numerobis]

mock-sun, HDR image

did you try with emitter + emitter?
i think hdri + emitter is also slower

[feynman]

Good idea, I try emitter (mock-sun) + emitter (softbox inside) next and see how that very same thing renders in 15min. I'm fairly chuffed that it's already better looking (noisewise) than with that physical sky.

[choo-chee]

I think all the above are good ideas but I chose Maxwell to avoid the need for such tricks ....
You can achieve good renders with many engines - the main goal is to work less and get quality results.
If the main thing is to be physically correct and unbiased, so improving render speed is the main goal.

[feynman]

Numerobis; that was a good hint: Ditching the image based lighting for a second emitter (besides the mock-sun emitter) makes for a huge difference. Same duration, but considerably less noise @ same SL; less noise even in the clearcoated polished brass.

1280x853px, 15min, SL 13.12, mock-sun emitter, emitter (plane with softbox HDR image)

[feynman]

The "mock-sun plus emitters" technique really speeds up rendering and feels much more controllable, just like in a photographic studio. Hardly any noise now, apart from JPG compression.

2400x1600px, 5hours, SL15, mock-sun emitter, soft-box emitters (planes with HDR images)

[Fernando Tella]

Looks good. No environment of any kind, right? No dome or anything.

[feynman]

No. Only a six-polygon mock-sun disk emitter and two two-polygon softbox emitters. With either Sky or Image Based on, the render time was too high and far more noise was present. Also, the mock-sun disk allows for easy fine-tuning of shadows and their softness/hardness. What also helps is the AGS glass around GLASS glass trick.

[choo-chee]

that's looking very good and clean. can I ask you to give some more info about your mock-sun and softbox emitters? how are they setup ?

[feynman]

Best is if you look through the set-up yourself. You can download the pack & go scene here https://www.dropbox.com/sh/srtcj73tpetv ... k5-Aa?dl=0

[choo-chee]

thank YOU
checking it now...