PDA

View Full Version : occlusion pass mixed with color and fog layers?



prometheus
12-21-2014, 08:31 PM
should post images..and will do that later I think, just a quick question

I am using image processing/image filter composite buffer, and just include the ambient occlusion and use range in the occlusion settings.
I also got realistic fog on, in after effects I apply the occlusion layer over the color/fog layer, I use soft light since I think that looks best, in this tutorial he is using multiply, and that is what I see others use to, but for me it doesnīt look good, comes out to dark..so soft light works nice., however, the main issue is the occlusion layer overriding the fog in the color layer...so my fog is wiped out by the occlusion layer.

this is mentioned in this tutorial, but the advice to include "distance fog in the AO"...I dont see such setting, is he just mentioning it the wrong way here?

Quote..."04 Environment treatment

Next, we need to make sure that for the AO pass we set the environment backdrop to pure white. This will assist us when we add the AO pass as a ‘multiply’ layer in our compositing package, as it will only take the dark parts of the layer and apply that over the colour image. It is also important to note that if you are working with a level of distance fog in your scene to include the distance fog on the AO, otherwise you will find when compositing your image that the AO is running over your fog, which won’t look right. Just be sure to set your fog colour to pure white for the AO pass."



hereīs the link to the tutorial....thereīs a difference though, since I use the composit buffer, not per shader basis.

http://www.3dartistonline.com/news/2013/09/how-to-create-a-quick-ambient-occlusion-pass/

Michael

Sensei
12-22-2014, 04:37 AM
I am using image processing/image filter composite buffer,

But in tutorial she is not using composite buffer AO, but node as far as I can see..

Tutorials should be followed literally. At least 1st time while learning technique. Set up node, environment bg color to white, then try playing with fog settings.

prometheus
12-22-2014, 06:47 AM
But in tutorial she is not using composite buffer AO, but node as far as I can see..

Tutorials should be followed literally. At least 1st time while learning technique. Set up node, environment bg color to white, then try playing with fog settings.

Ye...absolutly, I knew that, I was just being obnoxius..thinking it might work anyway, tried buffer compositing since I just didīt want to add shader node to the objects, but it might be wise to follow it properly.
Should perhaps try dp pixel filter occlusion too, and see how that behaves with fog.

Michael

Sensei
12-22-2014, 02:33 PM
Ye...absolutly, I knew that, I was just being obnoxius..thinking it might work anyway, tried buffer compositing since I just didīt want to add shader node to the objects, but it might be wise to follow it properly.

AO built-in is accidentally calculated during GI calculation. Because rays are send in the all directions, they're just added together and AO can be calculated and stored without any additional cost. It's limited to GI settings.

But node AO is completely different story. When it's plugged to Diffuse Shading it can be modified at later staged by post-processing effects. Like fog.



Should perhaps try dp pixel filter occlusion too, and see how that behaves with fog.


I doubt it will work with standard fog, because pixel filters and image filters are applied at the end of rendering.

But if you would spread occlusion node using Global Materials (http://globalmaterials.trueart.eu), it should be exactly like in manual spreading them to each object/surface, so like in tutorial.

prometheus
12-22-2014, 02:47 PM
AO built-in is accidentally calculated during GI calculation. Because rays are send in the all directions, they're just added together and AO can be calculated and stored without any additional cost. It's limited to GI settings.

But node AO is completely different story. When it's plugged to Diffuse Shading it can be modified at later staged by post-processing effects. Like fog.



I doubt it will work with standard fog, because pixel filters and image filters are applied at the end of rendering.

But if you would spread occlusion node using Global Materials (http://globalmaterials.trueart.eu), it should be exactly like in manual spreading them to each object/surface, so like in tutorial.

Thanks for the insight on it..much appreciated and yes..the built in GI/radiosity occlusion is somewhat not appropiate I think for animations, also have to deal with radiosity flickering if not wrestling with the animated cache or non interpolated radiosity.

I donīt think I will go and buy global materials for this task though, I could set up the node for one material copy it and then select all surfaces and paste only the node in the surface editor..I think that should work decently..but havenīt tried it.

Right now I am trying out the dome lights and see how that will work out for animations, first test is not good enough with light samples and aa, so I will have to increase those and see how flicker and noise free I can get it and to what rendertime cost.... compared to AO, a downside maybe if you want full control in compositing with mixing occlusion and color, when domelights is final non seperation color render.

Michael

gerardstrada
12-29-2014, 04:15 PM
Just in case, fog component is stored in shaded refraction buffer. If you don't have anything to refract, fog will be the only component you'll get there (even if RT refractions are OFF). This can be added to an AO pass or even better to an illumination pass or shaded diffuse buffer.

But fog effect is not usually baked in other passes but rather added later. A custom-made fog pass can be got with DP FNE (http://dpont.pagesperso-orange.fr/plugins/nodes/DP_Filter.html) at the same time with your regular final color through a Depth buffer, a RGB Depth pass (http://www.lightwiki.com/wiki/Multipass_Rendering_with_Filter_Node_Editors#RGB_D epth_Pass) (pick your colors there) or with a slight variation of a FoCuS pass (http://www.lightwiki.com/wiki/Multipass_Rendering_with_Filter_Node_Editors#Focus _Pass) (use a "distance to camera" gradient instead).

If you are rendering with GI, you might want to try with an illumination pass (http://www.lightwiki.com/wiki/Multipass_Rendering_with_Filter_Node_Editors#Illum ination_Pass) (or Radiosity buffer) for getting a Radiance occlusion pass (http://www.lightwiki.com/wiki/Multipass_Rendering_with_Filter_Node_Editors#Radia nce_Occlusion_.28RO.29_Pass) instead of an AO pass. It's more consistent with actual lighting (quality and direction) and it really has no additional cost if GI is used. Native AO buffer does actually has an additional cost for us in the sense that LW calculates AA twice when we enable the AO buffer. This doesn't happen with Radiosity buffer though.

You might want to try also with an AO pass (http://www.lightwiki.com/wiki/Multipass_Rendering_with_Filter_Node_Editors#Ambie nt_Occlusion_.28AO.29_Pass) in DP PFNE. SG_AmbOcc and DP Amb Occlusion offer far better results than native AO nodes. Consider that for illumination purposes, AO pass multiply ambient lighting.

Merry Christmas (delayed) and Happy New Year!



Gerardo