SideFX releases Houdini 14

[petz]

Is it recommended to rely on component orders in Houdini? Are they ensured to be in a consistent order/direction? I thought it was non deterministic. There is also the user factor.

 

I assumed these have to be solved in code to reorder points in a certain way for a particular tool.

 

This also prevents the tool to be broken by changes done by SESI I think.

 

 

well, i have to agree with magneto. your tool should take care of component ordering/reordering no matter how the selection was done. and since you´re coding it anyways, it won´t be much of a problem...

[Guest mantragora]

Yeah, but the version on Orbolt is almost codeless, because it uses nodes. And I would prefer to leave it like that.

It would be good to get some confirmation from SESI that this behavior will not come back (H13 like correct ordering from selection no matter which direction I select), and then I will play with reordering. I know that we will not get reversing till next mayor version of Houdini, but for already connected edges I don't see a reason why this shouldn't work right now.

EDIT: When I do the same selection holding SHIFT+A, it works both directions, so it looks that it sorts itself. I don't see a reason why this should't work like that if you select the same edges by only holding SHFIT.

Edited February 8, 2015 by mantragora

[rayman]

I hope it is not a problem to use this topic to confirm some issues I found in Houdini.

 

I'm trying to use new anchor_id attribute with bullet, but it constantly crashes.

I don't think it is graphics related issue, but the support is trying to convince me that the problem is my 750M card that is not supported with OSX.

Can somebody check attached file and confirm that it crashes houdini. I checked it on another machine with Linux - the same result. Unfortunately I'm not allowed to send any details about it.

All you have to do is to open file and press yellow button to activate wrangle node.

EDIT: Fixed

 

 

Another thing that may be related to my configuration is Cmd+F (find) command while using new web browser. It just does not work in most of the cases. It took me a while and a couple of restarts to get this screenshot:

I think some Button/RMB command may be nice alternative.

Edit: logged as #66799

 

 

Thanks!

BUG_DOP_Bullet_Anchor_Id.hiplc

Edited February 15, 2015 by rayman

[rayman]

Are you using any simulations caches?

 

Thanks!

Not really. I had the same issue today for a while, while playing with some alembic caches and vdb from polygons.

[Guest mantragora]

Can somebody check attached file and confirm that it crashes houdini. I checked it on another machine with Linux - the same result. Unfortunately I'm not allowed to send any details about it.

All you have to do is to open file and press yellow button to activate wrangle node.

 

It crashes on Windows too, but I also have unsupported graphic card.

[rayman]

It crashes on Windows too, but I also have unsupported graphic card.

Thank you! I did another test, this time with supported card - another crash..

[Guest mantragora]

I don't want to say it loud, but I think that SESI support became a little lazy recently. With selection, when I logged them, they tried to convince me that the reason why edge selection fail in H14 is because I have unsupported graphic card. And I also asked people to test it on their systems. Then finally mtucker came and said that they are indeed broken.

 

I understand that people that doesn't own full FX license and technical support are not priority. But If I send them video (two) with word instructions what to do, that shows the problem, and I get in response that it's working on their side, and then I hear that it isn't, that means that there is something wrong with communication between testers and service, or first line of people that tests problem is not experienced enough with Houdini and should watch more tutorials for it.

 

I hope that it's only temporary because of new release, and because they are really busy it's what's the reason of those communication problems. Because otherwise support was really good.

[Guest tar]

I don't want to say it loud, but I think that SESI support became a little lazy recently. With selection, when I logged them, they tried to convince me that the reason why edge selection fail in H14 is because I have unsupported graphic card. And I also asked people to test it on their systems. Then finally mtucker came and said that they are indeed broken.

 

 

Sounds like you want support to be run by the developers...  they are the only ones really qualified to deal with huge variety of issues that do arise.

 

Remember Support acts as a filter to send through real bugs so the devs can actually spend time developing new features as well as fix bugs. The least we can do as users is run the min specs ya?

[Guest mantragora]

Sounds like you want support to be run by the developers... 

 

Hmm... now that you mention it... maybe I should log RFE ?! It's a good idea 

 

EDIT: SESI, why Orbolt asset approve time is soooo long?!!!!

Edited February 10, 2015 by mantragora

[cwhite]

I'm trying to use new anchor_id attribute with bullet, but it constantly crashes.

I don't think it is graphics related issue, but the support is trying to convince me that the problem is my 750M card that is not supported with OSX.

Can somebody check attached file and confirm that it crashes houdini. I checked it on another machine with Linux - the same result. Unfortunately I'm not allowed to send any details about it.

All you have to do is to open file and press yellow button to activate wrangle node.

 

I've added support for using the 'anchor_id' attribute with packed objects in tomorrow's build (14.0.241). IMO, though, it's probably not too useful for Bullet - it was intended for things like wire simulations, where you might want to attach a constraint to a point on an object that has deforming geometry. For Bullet, you're (generally) dealing with rigid objects, and so it's much simpler to attach a constraint at a fixed offset from the center of mass (in particular, the solver doesn't need to unpack the packed primitive's geometry in order to figure out where to attach the constraint).

[rayman]

Thank you, Cameron!

Yes, probably not too useful for bullet, but it may help in some situations.

 

Speaking of center of mass I just found another issue with custom COM. Simple scene - one object at {0,0,0}. Recompute COM is off. After the first frame of simulation the geometry is moved a bit. It looks like COM for collision geometry is computed and copied to intrinsic pivot to compensate when point pivot/P is set, but in this case it is not because 'computecom' is zero. As a result the geometry is moved. It is related to collision geometry- if I change the type of Geometry Representation the offset is different.

BUG_DOP_Bullet_COM_Initial_Recompute.hiplc

Edited February 10, 2015 by rayman

[rayman]

I made a simple demonstration to show you why this is a problem when pieces are added to simulation over time:

 

 

Everything works as expected when computecom is set to 1.

The problem occurs only if it is 0.

Edited February 12, 2015 by rayman

[sebkaine]

I have a Bug in Houdini 14 with the point wrangle and this simple VEX expression :

@foo = 0;
addvariablename( "foo", "FOO" );

This works in H13 but not in H14.

 

 

addvariablename_bug.hipnc

[Mandrake0]

with the int value it works. in the bug message it says it can't find this function. 

http://www.sidefx.com/docs/houdini14.0/vex/functions/addvariablename

 

Message: 

 

[Guest mantragora]

Another bug. I think.

 

Steps.

 

1. Start Houdini.

2. Create Geo with any geometry. You can use even default Cube that comes with it.

3. Scatter some points on it. Enough so you could create at least two triangle polygons from them.

4. Group 3-4 points and send them to addSOP to create closed polygon.

5. Again group points that you used to create polygon, but make sure you do this in point selection mode.

6. Again call groupSOP, but this time switch to edge selection mode and select any edge of previously create polygon.

 

I get crash at this point. You have to start from points in viewport and all  polygons created have to be from those points. If you have already some polygons in the scene it doesn't crash.

 

 

[rayman]

Another crash using negative index inside "pcimportbyidxi" (if there are no points inside point cloud) (EDIT: Fixed ) :

int hnd = pcopen(1, "P", @P, .01, 1);
int farthest = pcimportbyidxi(hnd,"point.number",-1);

BUG_VEX_pcimportbyidxi_neg.hiplc

Edited February 18, 2015 by rayman

[Guest tar]

With all the recent post perhaps we should now change the thread title to 'SideFX releases Houdini 14 bugs'

[sebkaine]

with the int value it works. in the bug message it says it can't find this function. 

http://www.sidefx.com/docs/houdini14.0/vex/functions/addvariablename

 

Message: 

bugmessage.png

 

Thanks for your feedback Francis !

 

you are right, it is not a bug, but a modification of the function synthax,

 

In H13 :

@foo = 0;
addvariablename( "foo", "FOO" );

Become in H14 :

@foo = 0;
addvariablename( geoself(), "foo", "FOO" );

[Jason]

With all the recent post perhaps we should now change the thread title to 'SideFX releases Houdini 14 bugs'

 

That's a low blow, Marty.

 

I would say, however, that I think SESI should've had an open-beta period for a few weeks. before releasing. They added and changed so much I think it probably needed wider testing.

[Guest tar]

sigh- it's said in jest to the highlight the fact that everyone is posting bugs to this celebration of H14 release...