Mike_A

habernir
anyone know what area is copernicus feature related to?

COPs - I believe.

Awesome, Super Jazzed if Copernicus is the long rumoured COP's re-write. I was hoping we'd see a glimpse in 20.0.. long time coming.

seelanv
Currently the Unity plugin does not directly support Houdini heightfields as Unity terrain. As a workaround, use a convert_heightfield, which will then allow the plugin to bring it in as a regular mesh.

There is development currently being done to support heightfields directly. As for ETA, the best I could promise is sometime this year.

Thanks thats great to hear.

I still have a question though, I guess I asked 3 above.

Can the Unity HEngine plugin work with Native Unity terrains? What I would like to do is be able to just drop the terrain prefab into my HDA within unity and work with it. I have not had luck so far.

Has the status of this changed at all? ( can Hengine work with Unity terrains now - 16.5 323 ?)

I made a simple tool to blow away any geo that is under a terrain, but sadly in Unity when I drop the Terrain into the specified slot in my HDA slot nothing happens - i would guess it is not recognizing it as geo. Is this user error on my part? should this work?

My hack at the moment is to embed an exported OBJ version of the terrain into the HDA, but that is not ideal as we have many levels and it is not dynamic if the terrain is edited.

I am going to try adding a convert_heightfeild node into the terrain input node chain just to see if that fixes things.

Also is there an ETA on when the Unity plugin will support the new 16/16.5 houdini terrains the like the Unreal ones does?

Thanks.

I have come up with a solution. Migrating to using “copy to points” instead of using the “old” copy stamp sop and stamp functions. I am basically doing a hybrid. I am using the copy stamp to set and manipulate the $ID values. ID was always stable. I have not sorted out how to replicate this with the “copy to points” because I think you need the upstream functionality of stamping inputs. But I have migrated all the rotation and scaling to the “copy to points” sop. And because data only flows down stream with this, as long as I set the rotations and scale attribs before I disrupt things with my local override, everything is stable. Which is how I though the copy sop would work - but makes sense now that it cannot. I also seem to have gained about 3x interactivity performance boost this way, with the copy stamp only worrying about copying the right mesh to the right ID, and the copy to points doing the rotations and scaling. I will try compiling the copy to point sops soon.

In the mean time, my tacky solution is to create point groups based on each $ID and use a copy to points to copy each mesh type to the appropriate point group. Its not pretty, doesn't scale up and down with object count but it improves performance by about 3x (not-compiled yet) vs doing the ID, scale, and rotations all on the old copy sop.

I was looking at this video earlier. So I have things kinda working.. i have random scale, rotation.. and even object selection.

What I don't have is any control like I had before. The Stamp function sends an attribute up stream. I am scratching my head about how to put the $ID value into the switch node.. basically copy mesh #5 when there is a point attrib of ID#5.

Like I do with stamp(“../copy4”, idval, 0) on my “old” copy stamp setup.

Thanks for your help once again Tomas. I was looking at this earlier. So I have things kinda working.. i have random scale, rotation.. and even object selection.

What I don't have is any control. The Stamp function sends an attribute up stream. I am scratching my head about how to put the $ID value into the switch node.. basically copy mesh #5 when the point has attrib is ID#5.

Like I do with stamp(“../copy4”, idval, 0) on my “old” copy stamp setup.

So I am looking at trying to move from using the “old” copy sop with stamps functions to the new copy to points sop. I have hit a bit of a snag. How do I us a “switch” node with this to randomize the mesh copied? I am also using a probability matrix along with $ID to control the frequency that a certain mesh might get used.

Looking at the doc's I don't see $ID, are any type of equivalent attribute the copy to points recognizes? any ideas?

Now I am trying to sort out if i can somehow dictate the $ID with the old copy sop, and then use the new copy to points to do the scaling and orienting. Maybe some sort of transfer attributes hack .

So, I am copy stamping onto about 10,000 points, and I am randomizing copied objects scale and rotation with your classic rand($PT). The problem I am having is I have made a system to do local overrides, so within a subset group of the 10,000 points I can change specifically what objects are copied, clumping amount, delete a percentage etc. Its all stable until I translate the override group selection, that causes the point order to change and all of the 10,000 scales and rotations are updated (re-stamped).

As far as I know $PT is a protected variable. I have not had any luck trying to manipulate this. Is there some way to capture the $PT of the whole group before I edit things and then paste it back later down the chain so only the override group changes. I assume ideally the override group would be appended to the end, so points 0-8999 are the same, and only the 1000 points within the override group change. So the override groups is always ptnum's 9000-9999 for example.

Any Ideas? Also is there a way to use something besides $PT.. in a copy stamp expression.. or point expression and get similar results where all points (or even just most) have a unique value? I am currently using $ID to dictate what is copied to each point.

So far the only solution I can think of is to break the 10,000 points into 4-9 regions, and have a separate copy stamp for each. The idea being to limit what the override affects and causes to be re calculated. Not a great solution.

Hi there, I have a simple curve decorator knock off, where I am copying a simple HEngine asset (glorified cube with some controls) onto my filled curve decorator hda knock off.

I am seeing that after I have more than about 90 copies of the cube in the curve the display switches to cards. I assume this is a performance optimization on HE's part, not unity's. I am just curious what the trigger is? Its no where close to the 32K points per object unity supports in Unity 5, and I think that limit skyrockets in Unity 2017.x. So just curious what triggers it so I can potentially replace it with a more representative stand-in or have the cards not come on till more draw etc.

Thanks.

Update. Did a test with an intermediate stand-in (unity can't render my usual wire sculpture type preview ).. and got to about 39K verts before things disappeared. Performance seems ok.. its still pretty interactive both adjusting the curve, or editing the object count when things turn to cards or disappear.

So this seems like a dumb, stupid simple question but I have wasted enough time trying various things, I need to ask.

I often add string(detail)attributes into my networks basically as labels to build file names etc down the line in rops.

I got the bright idea that I could use these detail attributes to tell me in my OTL interface what the asset is I am looking at, like what i see in the spreadsheet. So instead of the slider just saying 22, having an accompanying string field tell me 22 is Dinosaur_xyz. My question is how?

I added a string parameter to my interface, but seem to only be able to get the literal expression of the attribute ie. details(“../switch1”, “name”) … not the actual string value I see in the geometry spreadsheet of.. “Dinosaur_b12”. Do I need to use a callback script?

I have tried a few ideas.. an attribute vop, where I bind the detail attribute, but exporting the string seems to get me no closer.. not sure how to actually populate a text field. Same with and attribute wrangle.. can make a chs(“name”); field but can't seem to populate it with result of the dynamic expression result. Seems like it should be 2 lines of vex, but it escapes me.

Thanks, Ken

Got it working with these settings on OSX. Huzzah!

*bump*, any news here? Last I tried I was unable to get this to work on OSX. I might try replacing the $temp paths with a real path since “output” seems to be the error.. perhaps a permissions issue?

kevinthebright
These game tools are great. I get an error message with IMB though:

Anyone else getting this on MacOSX?

I see the same issue. The stand alone works fine btw.

The HDA seems to error on line 32 of the python node in the HDA. Basically the last line.. geo.loadFromFile(out_path)

How do I get debug/error feedback in H16 when using python in nodes on OSX? Is launching Houdini via the terminal the only way? I am not sure I know how off hand.

When things are successful, I can see print statements in the “python shell” window.. but nothing if say Houdini cannot locate a module.

It seems Houdini no longer ships with a console.app in the utilities folder.

Initially I would get a pop-up window error over the UI, but sadly I clicked the button at some point to send errors to console.. which does not seem to exist on osx. Any idea how to get that pop-up error back? It was better than nothing.

Any tips or words of advice? Should I just do everything in sublime and then bring just finished code over. I worry that Houdini will introduce new errors.. like locating non-standard modules that sublime will be able to locate.

Thanks, Ken

Awesome, works perfect. Thank you. I tried removing the @primnum earlier, since I knew detail should not need this 3rd argument, but I guess the backticks were generating an error so I kept it in. Thanks again.

Ok, a lightbulb went off during this, thank you for your help! I thought I was smart for a few seconds but then realized I am not

So I have used a connectivity and a partition node to give all the mesh pieces a $class prim attribute. I then used a attribpromote with the “mode” function to create a detail attribute with the most prevalent piece - most polys. And called it “main_class”.

I plugged the attribpromote into the 2nd input of the group expression node.

Then tried @class==`detail(1, ‘main_class’, @primnum)` and it worked.. same as typing @class==79.

Unfortunately when I scrub the timeline and get a new mesh with a new “main_class” value.. of 171.. the expression fails. Any ideas?

Thanks for the tip on the “Nth”, prim/point.. that is much simpler and about as fast as typing N like before. Ironically I got the tip to use nprims() from a video on using the new group sops.. So again, perhaps some examples in the docs would clear up confusion and best practise.

Sadly I am still not having any luck with your group by area instruction. Even when the measure sop is directly above the group expression node in input 0.

Pasting - prim(0, ‘area’, @primnum) > 0.5 or @area > 0.5 into the Vexpression field just makes me a group with no members.

Any chance you could post an example file on a stretched out box or something

Duh.. sorry I realize now that 0.5.. is huge.. it needs to be more like >0.0005 to get something meaningful

Thanks

So how do I group by primitive area in H16? I assume I use the group expression node.. but I'm not having much luck trying to do “prim(”../measure1“,$PR,”area“,0)>0.5 .

It also seemed less than intuitive to type in the nprims or npoints expression in the group range sop to get the “Nth” prim.. or to get just the first and last points.

The H16 group docs really need some useful examples attached, even though its basic stuff that ppl have done for years, its not obvious what the new syntax is, or sometimes even which of the 5 group nodes to use. I have seen ppl suggest using a wrangle instead of the group expression node entirely as its a bit more forgiving.

So, I have a few issues. I am trying to export the result of a decently large SOP network as an FBX file instead of .OBJ. OBJ has no material (mtl) support out of Houdini AFAIK it would seem so I am wanting to try to use FBX.

My first issue is FBX 2016, which I believe is the default file type in houdini 15.5.(523) I have nothing that can currently open that new a version of FBX.

Is there a way to set FBX 2014 as the default? I have set the file>export>filmbox to be 2014.. I have also set filmboxfbx ropnet node to use FBX 2014 by default. However when I use my sop - Rop output driver.. and make the extension .fbx the file I get is UN-importable by any DCC I have, so I assume it is fbx 2016. The file size looks good though.. a few K different from the OBJ version.

This must be user error on my part.. but when I use the explicit file>export>Filmbox fbx dialogue box.. and set to 2014, in ascii.. I get a tiny 17k file that is nothing but a few named nulls and no geo. I get the same thing using the FBX ropnet. I can open them but they are empty. The output driver looks to make a proper file I cannot open

Is there a way to force the SOP output driver to use FBX 2014?

Is there something special I need to do in the file>export>Filmbox fbx dialogue box to actually get me geo out of my network and use the renderflag i have set. It seems I can only pick at the OBJ level of the network.

Thanks.