Re: Arnold Volumes Object ID type of AOV?

[Jason S]

  
  

This may be of help if Deep Exr's are an option:

http://jokermartini.com/deep-compositing-101/
http://jokermartini.com/deep-compositing-102/
(shown using Vray for 3dsM but should apply to anything)

 

They Mention that deep can be slow, and it really can be much
slower, but the benefits can still make it worth it.

Quote:
There are many benefits to compositing with
Deep Image EXR’s such
as no edging errors, 3D depth data and having more information
to manipulate the digital images. The main drawback is that the
image files can become very large and slow to composite. Below
is an image showcasing the errors produced when compositing
‘Non-Deep Image EXR’
compared to ‘Deep Image EXR’. If you view the image below full
screen, you’ll see the edges which overlap the middle teapot
have a slight glow around their edges. Comparing that to the
image on the right, you’ll see there is no glowing edges, making
for a correct result after color correction. I drastically
changed the color of the middle teapot to blue. You can see
below the left image is full of errors while the right image
contains no errors.


  
  
  
  On 08/31/15 22:13, Jason S wrote:

 Are
  you exclusively using pure Red Green Blue (and Alpha) as your tile
  colors?
  
  (or are you using deep images?)
  
  Because otherwise you cant use object ID colors to isolate
  elements of variable opacity midpoints (inherent to volumes) to
  get something like this (below) once isolated.
  
     
  
  For the same reason (non-deep) arbitrarily colored objectID images
  are typically rendered at 2X+ size, to extract grayscale tiles
  (white (1)  OR black (0) ) 
  and then resize down to have antialising midpoints around edges of
  isolated elements.
  
  Or perhaps not as convenient as for all other elements with ID's,
  but how about just setting-up a pass for your volumes with all
  objects black (with black alpha)?
  
  
  On 08/30/15 6:18, Perry Harovas
wrote:
  
  

  Hi all, 
  
  
  Sorry to ask this here.
  Have asked in other places with no
real response.
  
  
  Does anyone know if it is possible to
have Arnold output
  what is essentially an object ID pass
(different colors for each object)
  but for volumes within Arnold?
  
  
  I have some VDB smoke, then some
other smoke in a different VDB.
  They both use different
Volume_Collector shaders with different settings.
  
  
  Is there a way to spit out a pass
that shows me the volume, but lit flat
  so I can use it as an object ID mask
to tweak inside of Nuke?
  
  
  I can get all the geometry in the
scene to output a Object ID AOV
  except the volumes. They are rendered
lit with shadows and lighting colors, etc.
  
  
  
  
  Without having to render a totally
different pass, is there a way to output an AOV
  that can isolate the volumes, each
one differently with a different color for each volume, so I
can tweak them independent of each other?
  
  
  I can use the Volume Opacity AOV, but
it outputs ALL the volumes as one color (white, with
transparency) in the resulting AOV. This would be perfect,
if each volume had a different color, not all the same white
color.
  
  
  Hope this is clear.
  
  
  Thanks for any help!
  
  
  Perry
  
  

  
  
  


  

Redshift maya artists?

[Eugene Flormata]

this'll probably sound blasphemous but any generalist maya 2016 redshift
artists around? Vancouver based, But remote might be alright, I'll have to
get back on that.

Archvis rendering(redshift maya) / composite(after effects) work with an
extremely tight deadline coming up. Thanks and sorry! Links for demo reel
will be sent to my supervisors

A bit OT - Kraken issues

[Oscar Juarez]

Hello Eric and Phil,

I don't know if this list is the correct place to send issues and questions
about Kraken, if not, where would be a good place? On the github page for
the project?

Right now I started testing it and everything works as expected, opening
the editor, building guides and building rigs.

The rigs work, although in Softimage much slower than in Maya at least with
the Fabrice rig, if I save the scene and close it and re-open, the rig
stops working, still interaction is like before, but the deformers are not
moving anymore, I guess it is still evaluating but something gets
disconnected.

In maya this is not the case, I can save and re-open and everything still
works. I checked and the spliceOps in the cmpOut nulls and deformers are
still there, but somehow they are not working.

Does this ring any bell? I'm using Fabric 1.15.2 and the latest Kraken.

Thanks in advance
Oscar Juarez

Re: A bit OT - Kraken issues

[Eric Thivierge]

Hey Oscar,

I'll answer on this list but you can email Phil and I directly in the
future. We'll probably setup a google group for questions / issues.

Regarding the performance, it's a known issue and we're actively working to
find out how to speed it up.

Can you try using Fabric 1.15.3 and getting back to me off list on whether
the save / load issues are still there?

Thanks,
Eric T.


Eric Thivierge
http://www.ethivierge.com

On Mon, Aug 31, 2015 at 12:52 PM, Oscar Juarez 
wrote:

> Hello Eric and Phil,
>
> I don't know if this list is the correct place to send issues and
> questions about Kraken, if not, where would be a good place? On the github
> page for the project?
>
> Right now I started testing it and everything works as expected, opening
> the editor, building guides and building rigs.
>
> The rigs work, although in Softimage much slower than in Maya at least
> with the Fabrice rig, if I save the scene and close it and re-open, the rig
> stops working, still interaction is like before, but the deformers are not
> moving anymore, I guess it is still evaluating but something gets
> disconnected.
>
> In maya this is not the case, I can save and re-open and everything still
> works. I checked and the spliceOps in the cmpOut nulls and deformers are
> still there, but somehow they are not working.
>
> Does this ring any bell? I'm using Fabric 1.15.2 and the latest Kraken.
>
> Thanks in advance
> Oscar Juarez
>

Re: A bit OT - Kraken issues

[Eric Thivierge]

I've just opened the kraken mailing list to the public:
https://groups.google.com/forum/#!forum/kraken-project


Eric Thivierge
http://www.ethivierge.com

On Mon, Aug 31, 2015 at 1:25 PM, Eric Thivierge 
wrote:

> Hey Oscar,
>
> I'll answer on this list but you can email Phil and I directly in the
> future. We'll probably setup a google group for questions / issues.
>
> Regarding the performance, it's a known issue and we're actively working
> to find out how to speed it up.
>
> Can you try using Fabric 1.15.3 and getting back to me off list on whether
> the save / load issues are still there?
>
> Thanks,
> Eric T.
>
> 
> Eric Thivierge
> http://www.ethivierge.com
>
> On Mon, Aug 31, 2015 at 12:52 PM, Oscar Juarez 
> wrote:
>
>> Hello Eric and Phil,
>>
>> I don't know if this list is the correct place to send issues and
>> questions about Kraken, if not, where would be a good place? On the github
>> page for the project?
>>
>> Right now I started testing it and everything works as expected, opening
>> the editor, building guides and building rigs.
>>
>> The rigs work, although in Softimage much slower than in Maya at least
>> with the Fabrice rig, if I save the scene and close it and re-open, the rig
>> stops working, still interaction is like before, but the deformers are not
>> moving anymore, I guess it is still evaluating but something gets
>> disconnected.
>>
>> In maya this is not the case, I can save and re-open and everything still
>> works. I checked and the spliceOps in the cmpOut nulls and deformers are
>> still there, but somehow they are not working.
>>
>> Does this ring any bell? I'm using Fabric 1.15.2 and the latest Kraken.
>>
>> Thanks in advance
>> Oscar Juarez
>>
>
>

Re: Arnold Volumes Object ID type of AOV?

[Jason S]

  
  
Are you exclusively using pure Red Green Blue (and Alpha) as your
tile colors?

(or are you using deep images?)

Because otherwise you cant use object ID colors to isolate elements
of variable opacity midpoints (inherent to volumes) to get something
like this (below) once isolated.

   

For the same reason (non-deep) arbitrarily colored objectID images
are typically rendered at 2X+ size, to extract grayscale tiles
(white (1)  OR black (0) ) 
and then resize down to have antialising midpoints around edges of
isolated elements.

Or perhaps not as convenient as for all other elements with ID's,
but how about just setting-up a pass for your volumes with all
objects black (with black alpha)?


On 08/30/15 6:18, Perry Harovas wrote:


  
Hi all,



Sorry to ask this here.
Have asked in other places with no real
  response.


Does anyone know if it is possible to
  have Arnold output
what is essentially an object ID pass
  (different colors for each object)
but for volumes within Arnold?


I have some VDB smoke, then some other
  smoke in a different VDB.
They both use different
  Volume_Collector shaders with different settings.


Is there a way to spit out a pass that
  shows me the volume, but lit flat
so I can use it as an object ID mask to
  tweak inside of Nuke?


I can get all the geometry in the scene
  to output a Object ID AOV
except the volumes. They are rendered
  lit with shadows and lighting colors, etc.




Without having to render a totally
  different pass, is there a way to output an AOV
that can isolate the volumes, each one
  differently with a different color for each volume, so I can
  tweak them independent of each other?


I can use the Volume Opacity AOV, but
  it outputs ALL the volumes as one color (white, with
  transparency) in the resulting AOV. This would be perfect, if
  each volume had a different color, not all the same white
  color.


Hope this is clear.


Thanks for any help!


Perry


  



  

Re: Cone generator

[Olivier Jeannel]

Absolutely interested !!
I'm not good at cloning or repeating.
Stil bangging my head at generating arrays.

I get the correct number of point per (original) id (ID 0 have 0 point, ID
1 have 1 point, ID 2 have 2 points, ...etc... )
But I'm having difficulties in sorting the points per clusters, so that :
Cluster 0 has point ID 0
Cluster 1 has points IDs 1,2
Cluster 2 has points IDs 3,4,5
Cluster 3 has points IDs 6,7,8,9
Etc...

Ultimatly the clusters should be organised like this
Cluster 0 has point ID 0
Cluster 1 has points IDs 0,1
Cluster 2 has points IDs 0,1,2
Cluster 3 has points IDs 0,1,2,3


On Mon, Aug 31, 2015 at 1:15 PM, Cesar Saez  wrote:

> Hi Olivier,
>
> What about clonning the point using its id to drive the number of clones (
> (id+1) * 10 )?
> I don't have softimage installed at home anymore, but I can give it a try
> tomorrow during lunchtime at work if you are still interested.
>
> Regards,
> Cesar
>