I understand the reason for dirt, but why is this not an integral part of the render ? Surely it means that until you composite a vray dirt pass into your render your image wont show the more detailed areas correctly and wont show occusions correctly…isnt that what vray and all its settings is about.
Does this mean that when I render a scene in vray, it isnt shaded correctly ?
I understand the reason for dirt
No you don’t 
Dirt it is an ambient occlusion pass - fake, GI illusion. It is one bounce QMC (DMC, brute force - whatever ). It is used mainly for a hardcore compositing needs. When you don’t need or can’t afford pure GI (dirt and ambient occlusion - it is not a GI, as GI uses more than one bounce).
Cheers 
Ok, but i find without a dirt pass the extra fine details are lost. So how do I manage to get fine details from a single render.
Also, Ive noticed for instance when rendering out a specific scene that has a computer keyboard the shadows are hardly noticeable at a distance…until i comp the dirt pass it doesnt look right.
Maybe your settings are extremely low, as low irrmap tends to blur everything, as it is an interpolation method. The more samples it has the more details you will see. Try to use brute force on primary bounce and LC on second. You will notice all the details scene got. Or use medium irrmap + de (detail enhancement - but you must play with radius of de affected zone).
AO pass is good, when you use it to reveal details, when you primary goal is rendertimes.
I need to experiment with other methods more by the looks of it.
Thanks for the advice.
If you’re curious, the difference between AO and GI is that GI is traced from the lights while AO is traced from the objects.
In essence, GI has the lights fire off photons and calculates how the bounce around the scene. It then stores that information and uses it when shading the pixels in the render.
Ambient Occlusion on the other hand shades the pixels by shooting out samples in a cone from the normal of the surface. It then traces the samples until they hit an object or the “max distance” is reached. if the sample hits an object the sample returns black and if it don’t it returns white. Then all the samples are put together and the resulting color is used to shade the final pixel. For instance, if you shoot out 4 samples and 3 of them hit an object, you would get a dark grey.
So as you can see AO uses a much more efficient calculation than GI, but is “fake” in that it doesn’t take into account the lights in the scene at all but only decides the shading based on the distance between the objects. You get fine details more easily but it isn’t as accurate as real GI.
AO is still very useful though but you should render it out as a separate pass to have more control over it. Even though you can render it out as a part of the final image if you want to. Just use the dirt map in any of the map slots of your material or use it as a mix map. Nice way to add rust to objects although if you’re gonna show that much detail you might as well go all the way and add the rust manually…
Okay that’s enough blabbering for today.
Class dismissed.
Shimakaze is slightly off the point here; “GI” as a term does not imply tracing from either the lights or the camera - it simply means the effect of bounced light, regardless of how that is calculated (do not confuse this with the specific usage related to other renderers). Except for the photon tracing methods, V-Ray does not trace from the lights - all tracing is done from the camera and objects forward (for brute force, irradiance map, light cache methods).
The dirt map is useful if you need to represent worn out objects, or other shading effects, but it is not a good idea to use it everywhere in your scene, as this typically creates an unreallistic look to the images. Check out the RGB values in the photo below - there is no noticeable darkening near the corners of the white walls (in fact, in some cases, there may be brightening there because of the bounced light).
If you want to bright out the geometric details in the image, you can use brute force GI, or detail enhancement or modify specific materials to use the dirt map.
For VFX purposes, the dirt can indeed be useful as a quick and dirty replacement for GI, since the render times need to be as small as possible; however there is a trend to move away from this and use actual GI instead.
Best regards,
Vlado
“Sometimes reality is just not good eneugh”…
VrayDirt. I think the name is pretty self explanatory. I’ve found that the detail people want to add with dirt and occlusion passes to their renderings as “GI detail” is actually just dirt in real life too. For example, the outside of a building gets dirty quite fast in an urban environment and what you may see as a shadow in the corners, is just some dirt. 
Best regards,
A.
Indeed, and in this case, VRayDirt is very useful; for newly designed buildings and interiors however, you seldom want them to look “dirty” That’s why I don’t understand the urge to use an AO pass on everything…
Best regards,
Vlado
a handy thing about the vraydirt is using it as follows:
for example if you are doing a interior there are sometimes problems with suff beeing on a floor, the seems to float and not connect/sit on the floor.
one option would be moving the object just a little into Z so that you have some shadow there (using high IR)
a other option is to use a vraydirt on the floor. this causes a little bit of “dirt” at the vertext close to the floor. therefor a somewhat same effect…
I like to use vraydirt on just about everything, even new buildings. Why? Because it can add ‘depth’ to areas that would otherwise look flat. Just because you use vraydirt doesn’t mean you have to slap it on with the default settings at 100% opacity. Subtlety is the name of the game here.
I’d like to know more about exactly how VRay calculates the GI. Mind sharing?
If it’s traced from the camera, how do you get the effect of first bounce objects that aren’t visible or are out of frame?
I almost never use dirt but that’s because I prefer painting in things like that in post. I usually make stills though.
In the past 6-9 months, I dont think ive made a single (non-glass/liquid) material that doesnt have a vray dirt in the diffuse. Much prefer the control you get over doing on a per-material basis, because then you can use a different bitmap in the dirty area or make some areas more subtle/spread out than others.
Thanks for the info…I guess I used vray dirt as a complete pass rather than on single objects. I will try that…
Would you mind elaborating on your process of adding dirt in the diffuse - I’m a little confused (although that’s fairly standard).
Thanks
It’s literally putting a copy of the vray dirt map in the diffuse slot. The major difference between this and not using it is that the vray dirt map has controls to let you tweak the distance and speed of falloff between surfaces where as it’s quite difficult to control GI in the same way - you do get an option of using GI bounce mutipliers but it’s a single simple control as opposed to the finesse of vray dirt. Also rather than using black and white or simple colour swatches, cubicle seems to be using a normal version of his map in the clean slot and a darkened version in the dirt slow so not only does he have control of how fast or slow these light and shadow areas work together but also has a levels control to tweak how bright or dark certain areas of the same material are.
For some this might be an overcomplicated method of doing things but on the other hand having such fine control is great if there’s one specific thing you want to tweak in an image - instead of having to changes lights and materials and re-render you only have to tweak a single material setting with more predictable results.
Put the diffuse bitmap in the occluded/unoccluded color slot at the bottom of the vraydirt map. Make the occluded version a darker version or just keep it black depending on your needs.
Ahhhhhhhhhh…I see.
Thank you.
