XYZ WPP pass with Vray

[Devhill]

Hi guys

I don't know if this is the right place to ask about vRay but because is directly connected to the new Deep compositing technique i think some compositor know about that.

How can obtain a WPP XYZ pass with vray and what parameters do i need (coordinate space, color space and so on...)
I have already tried google but i can not find good explanation about WPP. There is not many info.

Also can anyone explain better the difference between XYZ pass / UV pass (which is also used for vectors) and Extra texture i can use from VRay ?

You can also suggest some good forum about vray directly i will check there.
Thanks

[pingking]

for the WPP (world position pass) in Vray you must create a new pass with the an VraySamplerInfo and set "type" to "Point" and coord system to "world". ready to render. best is to use 32bit float images as render output

and be aware to not mess up deep compositing with deep image composting. the first goes for the stuff fusion does with WPP and the second is something diffrent (http://www.deepimg.com/)

an XYZ or WPP or P pass gives you the values of the position of that pixel in the world space. so if it has R = 20, G = 40, B = -25 than it means this pixel is in world space at X = 20, Y = 40 and Z = -25

if you use MAX be aware the max uses Z as up axis and not Y (as most other programs do), so you have to shuffle the Green and Blues channel

an UV pass can stand for alot of things. first it can be the UV texture coordinates, second it can be the motion vectors

[Roary]

I generally use extraTex with vrayDirt for AO passes, cant say I've had the pleasure of working with it in regards to deep images.

[Devhill]

great !

for the WPP (world position pass) in Vray you must create a new pass with the an VraySamplerInfo and set "type" to "Point" and coord system to "world". ready to render. best is to use 32bit float images as render outp

that's what exactly what i need! I've always searched info about sampleInfo but without any result even on vray online manual.

and be aware to not mess up deep compositing with deep image composting. the first goes for the stuff fusion does with WPP and the second is something diffrent (http://www.deepimg.com/)

very interesting i will check everything !

an XYZ or WPP or P pass gives you the values of the position of that pixel in the world space. so if it has R = 20, G = 40, B = -25 than it means this pixel is in world space at X = 20, Y = 40 and Z = -25

exactly, i know the concept but i missed the exact render element to use

if you use MAX be aware the max uses Z as up axis and not Y (as most other programs do), so you have to shuffle the Green and Blues channel

good !

an UV pass can stand for alot of things. first it can be the UV texture coordinates, second it can be the motion vectors

Ok I know about motion vectors which stores the vectors info in the red and green channels that i can use later in the comp.
But what about UV texture coordinates ? What render element do i need to re-texture in the comp and with which node usually ?
Speaking about Fusion do i have to use UV map or texture transform ?

pingking u rock !
thanks a lot

[pingking]

if you have laid out your geometry nicely to UVs (texture UVs) in your 3d app then the texture UV pass (or as it is called in renderman and nuke the ST map) will contain a image which shows a gradient with red and green over the texture UV coordiantes.

in fusion you can use the texture node to apply a new texture or this macro or the plugin by RE:vision

[Devhill]

thanks again pingking! simply great!

Because you talked about deep image concept i have some question now.
I prefer to write here so we can continue the discussion and explain the difference between deep compositing from eyeon and deep image compositing.
(if that's not the right place i can start a new thread)

If i understood well, the deep image concept stores information of the image about deph and opacity in each pixel and will be restored in the compositing software as a light projection maintaining the real and correct depth of the scene. So i can fly, with the camera, through real 3d-extruded objects that are represented by a cloud of points. And all the objects can have a real interaction without using holdout matte.
Is it correct ?

- How can these images from the 3d package can be generated ?
- Does this technique is experimental ?
- Does this new approach can be done only with Nuke ?
- Is this a new revolutionary way to composite or just an alternative to traditional compositing ?

Very interesting new argument.

Oh, by the way, what happens to Eyeon website ? http://www.eyeonline.com

[pingking]

the eyeon website has some technical troubles over the weekend. was mentioned on the fusion mailing list

you did understood the concept of the deep image compositing right.
it stores the color value and opacity of the the objects in diffrent depths, as you would cut slices in diffrent depths

fusion does something like this in the volume fog tool (since 6.2), but only for the fog volume

at the moment only nuke has some tools do work with deep images but works only with deep images from (pixar) renderman
deep images can be generated (afaik) by pixar renderman (didnt read about the other renderman renderer), a beta version of vray has deep image support and i believe arnold can render something like it. but there is no standard in how to store these data until the upcoming exr 2.0

and at the moment it will not be much fun to work with deep image files. in one of the nuke beta releases where a tutorial to the deep image tools which contained example files. one frame was about 300-1000MB!!!!! and this was just simple spheres and cylinders

the technic is production proofen by some big houses. it was used in avatar, rise of the planet of the apes, legend of the guardians and other

its an different way to solve problems, which cant be solved by traditional work. maybe we see it in the future more and more

[Devhill]

hi pingking yes i understood but i simply wrote a confused description. sorry.
the concept yes is like fusion volume fog but yeah... 1gb per frame is not very easy
thanks for the info!!

how can i stick directly cards/planes in merge3d with wpp pass?
thanks

[pingking]

over here i posted an very old example of how to use the fusion particles to show the wpp data in 3d space (its a pImageEmitter and a pCustom to drive the particles)

if you want to place a card on a specific position in the 3d space you could use the "pick Tranform" knob in the cards transform controls
(look in one view at your wpp pass and pick from there)

[Devhill]

good example even if still a bit complex for me.. pCustom is still an alien.
good for the placement of the cards in 3d space ! It works good and very simple workflow.
thankx again

[Devhill]

Hi again pingking, is there a way to obtain Ppass in TIFF 32 bit float instead of EXR?
If i have a clamped P pass how can i use it in Fusion?
Maybe some BC/CC node or Color Space/Color Depth connected after the Ppass can help me in any way ?
I noticed in the 3d histogram that, if the Ppass is correct with all values, the shape of the point cloud (colors) reflect the shape of the image. Right ?

[pingking]

how does a clamped Ppass looks like? and if it is really clamped then there is no chance to get the missing info back
if you mean a normalized ppass then i got now idea how to expand it to the orignal range. therefor you should now at least one position value of the real 3d scene and use that as mutliplier for the ppass
also, the ppas should not be clamped in the negative values

and you can save your WPP to any image format you like as long it supports 32bit floating point

[Devhill]

the clamped pass is this attached. I suppose all info are lost.

This velocity pass has been rendered with Vray. I know that for velocity the multiplier parameter should be set to 0.01 but has not been done.
There is a way to unclamp and extend the range from here ?
Does TIFF files can be saved as 32bpc ? (from fusion yes.. but 3dsmax of Vray doesn't have this option for render elements)
thanks !