depth of field subdivs

[peterguthrie]

apologies, just did a search and couldnt find what i was looking for (probably as the search function doesnt pick up DOF)

was there a maxscript way to access dof subdivs added or not?

[joconnell]

There is but it won't do anything. I think I checked that back in the initial transition from 2.4 to 3.x

[peterguthrie]

okay dokay

[vlado]

You can search for "DOF*" instead.

The DOF subdivs don't affect the image anymore, only the AA samples.

Best regards,
Vlado

[mattclayton]

Am i right in thinking that increasing your min shading rate can effect DOF as well? As in you will need to increase your AA max values as well to compensate for more samples being used for GI etc?

[Macker]

Am i right in thinking that increasing your min shading rate can effect DOF as well? As in you will need to increase your AA max values as well to compensate for more samples being used for GI etc?

This was my understanding of it too.

[Companioncube]

Am i right in thinking that increasing your min shading rate can effect DOF as well? As in you will need to increase your AA max values as well to compensate for more samples being used for GI etc?

there are a couple of good images in the online docs showing min shading affecting the DOF http://docs.chaosgroup.com/pages/vie...pageId=7897184 down in the "Example: Min Shading Rate" section with the hand rail images

[mattclayton]

there are a couple of good images in the online docs showing min shading affecting the DOF http://docs.chaosgroup.com/pages/vie...pageId=7897184 down in the "Example: Min Shading Rate" section with the hand rail images

Ahh yes, i knew i'd seen it somewhere! Cheers

[joconnell]

Just having a think about that comparison and what the underlying cause is for the difference.

The key thing here is that the render is progressive and limited by time and not by noise threshold or max samples.

One render has a shading rate of 1 and the other is 16. For every aa ray we get 1 secondary sample in the first example and 16 per aa ray in the second render. Lets say that we're letting both renders run for 60 seconds each. Lets also say that taking a sample whether it's shading or anti aliasing / camera takes 1 second. For the first render we take one aa sample and that has a single shading sample piggybacking on it, it's two seconds for the pair of them. In the second example we get one aa sample and 16 secondary samples, that's 17 seconds in total. If we were to let the renders run for 60 seconds and assuming that we're using the default progressive settings of min 1 max 100 for aa, the first settings are taking 2 seconds total for a loop of aa and msr shading - it can get up to 30 samples of aa in its 60 seconds. In the second example, it takes 17 seconds for a loop of 1 aa and 16 shading samples so it'll get through 3 full loops, 1 extra aa and 8 extra shading samples, so only 4 aa samples in total.

The big thing here is that DOF and motion blur are both AA problems and not shading ones so in the above examples having higher shading is slowing down each loop of sampling and not letting vray fit in more of the aa samples it needs. Again like a lot of my lower aa / more shading nonsense, it's because we're guiding the renderer towards the source of the noise and using the appropriate type of sampling to clean things up.

If you were to let the above renders run to a noise threshold rather than just for a specific time, the AA based one would hit your noise threshold quicker since it won't have the same overhead of shading rate holding it back - your only difficulty is if you've got some sharp areas of the image to get the right balance of AA for your dof and enough shading samples for the sharp areas.

[mattclayton]

I see, so it's not technically affecting the sampling of the DOF in this scenario, rather the time limitations are not allowing it to reach the same level of sampling in the specified time, makes more sense. Thanks for the clarification John