[Tim Ellis]

Condensed thread, all the working results in one post:-


Test:-


Comparison:-


Conclusions:-

Using a camera 'window' with 'Hidden to GI' enabled, with strip all caustics from a scene.
Using a 50% Glass & AGS material, works fine for correct shadows and some refraction. (Check shadow lines running through glass and edge caustics, for refractiveness.)
Using a reduced reflective AGS & glass material, no reflective layer for AGS portion, works well but with no reflections.
More Glass % allows better refractions at the cost of incorrect shadows, in a Glass & AGS mixed mxm. Self shadows viewed through self are incorrect for increased glass %'s.

Tests runnning Glass mxm 'Hidden to GI' object, solves faster than AGS & Glass mix mxm, with 'Visible to GI' enabled.

I think user discretion and application, denotes method to be used, for optimum results.

If you want correct shadows, use 50% AGS & Glass mxm.
If you want correct sunlight, but no shadows added by the glass, use 'Hidden to GI' & Glass mxm.

AGS vs Glass vs Hidden to GI vs 50% AGS & Glass:-





Swimming pool application, displaced geometry for water layer:-



50% Glass & AGS mxm:- http://mxmgallery.maxwellrender.com/sea ... arch=ellis
________________________________________________________________________________________
Original first post & tests:-
In case, like me, you didn't know that sunlight through glass now works. (To a certain degree.)

You need to tick the 'Hidden from GI' box, in the object parameters window, for the glass object.

I'd not seen this mentioned anywhere and only learnt about it yesterday, so please forgive me if you already know.

First tests work, sunlight can be seen through glass, from an exterior camera into an interior scene.

Shadows are correct and not produced, or added to by the glass object as before. Caustics are not working!



(See my post below for detailed explanation of this image.)

Maxwell Render 1.5.1

Tim.

[giacob]

in a sense this is logic... if glass dont look trasparent because it take too long time for their caustics to solve if u exclude them from GI u solve the problem.....
... in effect i noticed that Nl introduced this chance to give up total GI by excluding object from GI or secondary ray but didnt thought of this possible use

[PA3K]

This is great idea, i have to try it. And what about to make sandwich glass object for it? I mean glass(GI excluded) - AGS - glass(GI excluded). Or window box 100% size with glass(GI excluded) and inside 99,9% scaled box with AGS applied?
...i have to try it

Thanks for this idea...

Patrik

[PA3K]

I don`t want to start separate thread for this,so i will explain here what i have discovered after trying some combinations.
I thought it will be the best idea to combine plate of regular glass (not visible to camera) with plate of regular glass (hidden from GI) - both placed on the same position. I checked view from outside and it was the same result like single plate of regular glass (known problem). Then i deleted one plate, so there was just one-not visible from camera. I created another plate of glass and placed it partly in front of camera. I was surprised that plate not visible to camera became visible just in that part of view through the glass in front of camera as you can see on picture below. So i think if visibility to camera will work fine there will be possibile to make sun caustic through glass. Is it a known bug, that i didn`t know or something else? Sorry it is just very simple scene

Patrik

http://img175.imageshack.us/img175/9268 ... ityfx0.jpg

[Tim Ellis]

Cheers for testing PA3K.

My tests:-


Test with Hidden to GI glass, some with AGS face or box surrounding the glass boxes.

Each object is glass with 'Hidden to GI' ticked. AGS face and box objects, are a tiny (>0.1mm) gap from the glass surface.

AGS face fr means single polygon of AGS in front of the glass object etc.

Each box is 20mm thick.

Caustics only present with glass unhidden to GI, except for ags reflections.

So an AGS box surrounding a hidden to gi glass object, produces sunlight through glass, with correct shadow.

Glass box on it's own with hidden to gi on, doesn't produce any shadow.

Testing now with spherical objects for better caustic representations.

Tim.

[Tim Ellis]

Sphere test.

Glass boxes inbetween shadow poles and glass spheres, are GI enabled and then GI not enabled, repeated 2x for each sphere.

Using an AGS 'jacket' for a non GI glass object produces slight shadows for the object, but at the cost of high reflectivity. Reducing the ND vale for the reflective AGS layer will reduce this effect.

Activating dispersion for the glass mxm, makes no difference to shadows.

One more to come using a coating on the glass mxm.

Tim.

[tom]

Very interesting Tim. Watching your progress here.

[Tim Ellis]

Cheers Tom.

Last one. Coatings on glass.




And a key to what you're looking at:-



So conclusions:-

Sunlight through 'Hidden from GI' glass works.

Sunlight through 'Hidden from GI' glass with AGS jacket, works and produces slight shadows.

Sunlight through 'Hidden from GI' glass produces no caustics, either reflected or refracted. NOTE:- Caustics are still produced if the camera is not looking through glass.

Coatings on glass mxm for 'Hidden from GI' or dispersion, make no difference to shadows.

Using an AGS 'jacket' for a 'Hidden from GI' object creates missing shadows for glass, but at the cost of increased reflectivity.
Reducing ND value in AGS reflective layer and/or using 30' recalced smoothing angle (inside MXST) reduces reflectivity for AGS.

Agreed not great for the best unbiased renderer, but still an improvement over before.

Might be possible to decrease transparency or increase tint for AGS, to increase AGS 'jacket' shadow levels.

Tim.

[Bubbaloo]

Thanks for all the work. This is good to know. And maybe it can be fixed?? Ahem.

[Tim Ellis]

No worries, glad to be able to work this out.

Just wanted to say thanks to Inaki (aka 'Toxie') who showed me the Hidden to GI method and set these tests off. Cheers buddy.



Tim.

[Inaki]

That is looking amazing and the way to short out this huge problem from archi viz is being close to get solve!!!

This is the proof of the hard testing work done, keepin on it, all we will find the way soon!!!

Thanks mate!!!

[giacob]

i wonder.. this "new type of glass hidden to GI " gives correct refraction? it seems to me it doesnt .. if this is true ( but i could be wrong) it doesnt solve the major problem with AGS : the fact that it doesnt have refraction....
in case refraction works ( in effect they should) why not use a glass hidden to GI but visible to the camera in conjunction with a glass visible to GI but hidden to the camera ? thats also for water... perhaps it gives the right risult.( we could eventually obtain water with caustic and correct refraction at the same time, which could not be obtained with ags + hidden to camera water) .. must try.. or have u already tried Tim

[wagurto]

Tim nice work? are you planning to share your glass material on the MXM library?

[sunmade]

that would be very cool tim.

[PA3K]

Great work Tim,

i want just ask again someone from NL or A-team. Is it bug if i place glass in front of camera and every objects not visible to camera became visible through this glass?

Patrik