I haven't dived into PDG at all, I only know what they're generally used for. I'm wondering if they're a good tool to use for generating thousands of variations of a model or more "traditional" approaches should be considered.

The basic idea, for anyone unaware about the current NFT craze, is something like - combine and shuffle 'n' attributes (not Houdini attr, but color/ size/ etc) of 'm' elements (blue, red/ small, big/ etc), but the combinatorics doesn't really matter.

citizen
I'm wondering if they're a good tool to use for generating thousands of variations of a model or more "traditional" approaches should be considered.

yes, pdg/TOPs is ideal for that

tamte

citizen
I'm wondering if they're a good tool to use for generating thousands of variations of a model or more "traditional" approaches should be considered.

yes, pdg/TOPs is ideal for that

Sorry for super late quote but i have a question. NFT collection need rarity data (ie. level 6 Red, level 1 gold etc.), can we add these data to still render or animation seq?

Hello Guys can anyone shed any light to adding rarity data in PDGs similar to blender

I wonder if we can use PDG with opendata, but not sure where to start with this.

since PDG allows implementing custom nodes using Python you can use it for anything you can use Python for
if however opendata has some sort of web API, you can also use URL Request TOP directly to get all sorts of data

I didn't know PDG can be used for URL request, that's pretty advanced, does each task wait for the request to complete before moving to the next TOP node? Or do you need a wait node to wait for all the data to be retrieved?

My original idea wasn't that complicated, I was thinking of using opendata like if there was a kaggle dataset on furniture with specs to feed the specs into a PDG setup and create all the furniture, lol. But your URL Request sounds more interesting. Because then we can feed in some stats from the Twitter API and I don't know, draw a smiley face for every time someone tweeted about Houdini or something. I feel like there's so much potential, I just need an idea.

bubblepins
does each task wait for the request to complete before moving to the next TOP node? Or do you need a wait node to wait for all the data to be retrieved?

each workitem awaits for it's correponding request to be returned and then it will be marked complete and dependent workitems can run and process the data (you also have time out and retry options)
TOPs don't execute node by node, but have workitem dependencies so one request can be fully processed until the end of TOP chain while another can still be waiting to be returned, etc (unless of course you deliberately make it wait at some point of the TOP chain for all upstream workitems to complete before continue processing )

but yes if you have datasets in the form of csv or json for example, you can read them using corresponding TOP nodes and proces further like you mentioned, generating furniture or anything really

I'll check it out, I have been wanting to do a tutorial on PDG's, maybe I can find some interesting API's online to feed into Houdini for some fun.