VRayDistanceTex + distance to VRayDisplacementMod object not working?

[Art48]

Is this a limitation of the VRayDistanceTex?

I'm trying to get water in deep puddles to kind of "crawl up" the edges a bit via a bump map.
What I thought is just plug in a VRayDistanceTex into the bump slot of the water object (via VRayColor2Bump) which calculates distance from the displaced surface.
What I noticed is that generally it works but only to the extend of giving me the distance between the water and the undisplaced object. Is this a limitation or do I have an error in my thought process?

Plane A at z = 0 (ground)
Plane B at z = -1 (water)

Plane A gets displaced (also in negative range) then the Plane B penetrates at some areas and forms the puddles.

With the VRayDistanceTex I always get a solid color, no edges. I have red as near color and green as far color. When I change the distance, I always get either full red, full green or a full mixture with no edges of the both which leads me to the impression it calculates distance on the undisplaced surface.

If so could we get that changed? I guess it would be useful for some situations.

Is there a workaround to get this effect?

[^Lele^]

try setting the displaced object to static geo.
It looks like you may be running into a wrong operations order.

[Art48]

Just tried it, same effect .
I'll set up a simpler scene and try it there really quick, in case I forgot something.

[Art48]

Yup, definitely doesn't work - no matter if static geometry or not.

Scene file (3ds Max 2014)
https://www.dropbox.com/s/2cp8ptjl0d...ceTex.rar?dl=0
(this thime, displacement only done in positive Z, so the lowest point on the displaced object is actually where the plane is at)

Render:

[benb32]

Couldn't you try using vrayEdgeTex with rounded corners?

[Art48]

I'll try. And you got me thinking about using ambient occlusion, that might work, too. But getting it to work with the VRayDistanceTex would be the best solution .

I'll try and report back.

[Art48]

I ditched the water aspect I was working on there, therefore nothing to report back yet. What would be good to know is if this is an issue being worked on by CG?

[^Lele^]

I'll try. And you got me thinking about using ambient occlusion, that might work, too. But getting it to work with the VRayDistanceTex would be the best solution .

I'll try and report back.

Either would need to have the mesh as displaced in ram before the render process starts. So static geo would still be the way to go i guess.

[Art48]

Yeah I see that - but since that doesn't work either I'm gonna assume it's a bug and therefore needs to be adressed.

[^Lele^]

Yeah I see that - but since that doesn't work either I'm gonna assume it's a bug and therefore needs to be adressed.

bah, i had missed your earlier replies. ignore me.

[^Lele^]

So, confirmed, it doesn't work, but i feel at least with the static geo option it should.
I have absolutely zero idea about potential performance issues, having to trace all those implicit triangles, but that's luckily for us all out of my hands.