Bifrost scattering positions

[InvisioMedia]

Hi,

  is it possible to scatter pre-defined models to pre-defined positions in Bifrost from json data?  It is important for me to be able to scatter items with ID's with specific matrices, i.e. object_A needs to go on specific points, etc.

  At the moment I am using Ian Waters' workflow in MASH. 

http://ianwaters.co.uk/wp/mash/setting-mash-point-positions-with-json/

It also has the advantage that you can edit the items after the fact as well, but I am thinking Bifrost could be a more powerful solution for this?  

Thanks for your help!

[syracoj]

not sure. 

you could wrap up in a compound a system that scatters points (on a read-in alembic) and distributes objects (also alembics) based off of ID's using instances. If you export that compound it will be exported as a json file. Perhaps this will offer a recipe you could use to edit the json to extend the graph. I'll have to try it myself to look at the data you get in the json. It may not work for what you need.

another thing that could provide some insight is when you create a graph if you select the nodes in the graph and do cntrl + c to copy you can past in a text editor and it will give you a type of markup of that graph. Perhaps this could be useful.

[InvisioMedia]

Hi Syracoj, thanks for your answer and help.

  The json data needs to be used as an input rather than an output.  The json files are stored on the network and contain scattering positions of various scenes and instance systems.

  I think it could be a great idea to add a node that can read external custom data like this.  The problem with alembic objects in maya is that it's not super easy to read custom attributes coming from (or to) the outside world (other dcc's).  Therefore, it can be hard to store things like which id goes on which point in an imported alembic file - and it will not be easily editable.

  Having a node that can store arbitrary data (lists, arrays, multiple vector information, etc) would be very powerful, and you could then easily build it with python like you do with MASH.

[syracoj]

AFAIK we don't currently have any input capabilities like that yet.

[syracoj]

could you upload a sample of the json data with position data so I could take a look and try something?

[InvisioMedia]

To bring it to the most simple form, it would be like this - an asset and an xform

[
	{
		"asset": "BoxA",
		"xform": [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 1.0]
	},
	{
		"asset": "BoxA",
		"xform": [0.0, 5.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 1.0]
	},
	{
		"asset": "SphereA",
		"xform": [5.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 1.0]
	},
	{
		"asset": "SphereA",
		"xform": [5.0, 5.0, 0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 1.0]
	}
]

However, for the sake of making it as powerful as possible, the best solution I think (for the future) should be a Bifrost node that can hold custom data, and accessible with python and other bifrost nodes.  So for example, it could store anything that would be accessible in the graph in the form of attribute / property, just like an excel sheet would do.

Thank you!!  : )

[InvisioMedia]

A good idea in this case would be to replace SphereA

"asset": "SphereA"

 by a full maya dag path, but really the idea is just to feed predefined data in the bifrost graph that is not a maya object in the scene, but more abstract information (with python for example).  

[syracoj]

I will forward this forum discussion to design for further analysis and consideration.

[InvisioMedia]

Thank you syracoj 🙂  For me, at the moment, I have a working workflow with MASH so no need to troubleshoot to much for me...  But what I am talking about is to open a door to something equivalent to the houdini point attributes in the bifrost graph - which is an effort in that sense from the beginning I think.

The core idea is to be able to inject, modify, read, shuffle information (think of an excel spreadsheet that can roll on faces, points, vertices, objects, transforms, etc) in the Bifrost graph in a custom "data node".  You may think that it's dangerous to inject data in an abstract form in the graph, but if the data format is not supported or recognized, you just disable the evaluation of the node and put an error on the graph.  These things are super powerful for TD's, vfx power users, motion graphics, programmers, etc.

In that sense, the bifrost graph would not need to be able to specifically read json data!!  Because in python you would decode it and store the filtered data in the mentionned data node, to be used (... modified ... manipulated) in any safe way in the bifrost graph.

I am sorry in advance, maybe some concepts I say are just not relevant for a Bifrost compound - I haven't used it much I am just throwing ideas.  Thanks for listening syracoj!!

[syracoj]

we do have schemas that are json based. I wonder if it would be possible to build a custom schema that could handle this for you.

to give you a peak of the schemas look in the bifrost install directory under:

C:\Program Files\Autodesk\Bifrost\Maya20##\bifrostVersion\bifrost\resources\bif\schema

In this ^ dir there are json files that describe certain schemas (point.json for example)

in bifrost there is a node called create_object_from_schema where one can input a schema (on disk) into the nodes input path and construct a bifrost object from it.

If this doesn't work, perhaps future work for the functionality you need could be done here around schemas. Or perhaps design will come up with a better solution (as they often do).

just something I wanted to run by you anyways to let you know it's there.

[InvisioMedia]

Oh..!  Yes exactly, very interesting!  It's not complicated for me to change the json data encoded form, I will look into this.  Thanks again!