Hi, I've been making a sim where on object collides with a ground plane and retains it's color (been following a tutorial where they use fluid sim points to point deform the original mesh). When I replace the ground plane with a static object everything goes haywire. I guess the point deform node is calculating the static object (which is acting as a ground plane), whereas I didn;t have this problem with the ground plane. Any ideas how I can resolve this (and why it would be ok with a ground plane and not a static object?)

The point deform can make crazy results like that when the point count is changing each frame. I'd just check that is consistent.

Snackovich
The point deform can make crazy results like that when the point count is changing each frame. I'd just check that is consistent.

Geometry spreadsheet shows some values changing - P(x) and P(y) are changing each frame, but i don't know what to do about it - how would I make them consistent? - I unchecked reseeding, but it hasn't done anything. Also tried all the options in Point Deform, but no luck (this is the first sim i've done).

P can be changing as that's just a position of the points
but the number of points has to be the same (which without reseeding and sourcing most probably is)

but in your case since you are adding static object in the sim, that object is most probably imported from the sim too therefore the geo will differ from the input geo which is just your source and confuse pointdeform
to avoid that just specify on your Dopnet1/Object Merge Tab Flipobject1 instead of * to import just that
or import it using Dop Import or Dop Import Fields SOP

on the side note, isolated points especially from flip sim are not good for point deforming
you will be much better off by advecting your geo using vel field from FLIP
also I'm not sure what you are trying to achieve by point deforming, there may be simpler solution to your actual goal

Thank you very much, changing the merge object to specific flipobject1 fixed that issue, although the geometry is a bit messed up now. I've found I have setup up the scene from scratch each time because Houdini is saving/caching each iteration that I do and tended to distort the meshes. Will rebuild it now, so fingers crossed it will work.

What i'm trying to do is create melted looking geometry/objects. Currently I'm bringing objects I have made, and that have material i.d's, into Houdini where I make a flip sim and then save out the interesting, partially melted looking shapes as individual static meshes - which I can then texture. I've struggled to find ways of preserving the shape of an object and the materials i.d's (when i say Material i.d's i mean the different colored sections of the object that can be used to fill with textures - https://imgur.com/a/pGziBFA). The tutorial i've been working from is this https://vimeo.com/173335049 [vimeo.com] . If you know of a better method i'd love to know.

If you want to preserve certain shapes you could adjust the viscosity in those areas before you sim. Have a look at the attached hip, you should be able to switch out different objects at the top. There's a switch at the bottom so you can see the difference when using the point deformer or tamte's advection idea, which is smoother.

This amazingly helpful, thank you Snackovich. Such a clean and light solution compared with mine. Will have to study the network, but for now it's giving me exactly what I need.

Hey man, just to add thanks again. Just seen that the nodes have little descriptions on them as to what they do - amazing.

Hi snackovitch and tamte, i've tried to recreate the file in my commercial version, but am not having luck. Any chance either of you could take a look?

Just had a look, your dop io node is looking for a dop object called ‘deforming’, but if you dive inside the dop network, your flip object is called ‘flipobject1’. If you change the names to match I think that should work again.

Thanks, but it's still not working.

Hey snappy, a good way to debug a scene is to start at the top, turning on the nodes one at a time, and just work your way down. When you find a node that's erroring or not doing what you expect, see what's up by middle mouse clicking or clicking the ‘i’ button. As you already have a reference scene, it should be straight forward to see what parameters are different. If you've updated the scene with some new geometry, see what's different between that new geometry and the geometry that was working…

there is a couple of things going on in your scene

1. as already mentioned the name in both dopio nodes should be “flipobject1” or just rename your flip object to “deforming”

2. biggest problem is that your object is transformed at the object level, try to avoid this at all costs for objects containing dopnets unless you fully understand what it means for your sim, in this case both of your DOP objects are double transformed
to solve this you can do 2 things
- remove transforms from the geo1
and either transform your rock and source at sop level before bringing to dopnet
or bring them in from different objects whose can be scaled
OR
- on your flipobject1 and geo1 DOP nodes uncheck Use Object Transform to avoid double transform that is currently happening, which will allow you to keep transforms on geo1 object, however I'd still not recommend this

3. your timeshift1 pose and source of the sim don't align cause you have another transform8 node in your source branch
so connect timeshift1 directly to transform8

4. if you still plan on comparing both methods, turn off Reseeding on the flip solver as Point Deform will not work with Reseeding, advection method will be fine

Hi, really good advice from both of you.

Snackovitch, I was really scratching my head because I went through each note very carefully, in the end I think it's probably my lack of familiarity with the software that I was unable to progress.

Tamte, thanks so much for going through my messy hip file, i'll go through all of your points when i've some time, but it looks very comprehensive. Brilliant stuff.