Is there a Houdini-like 2d app similar to Photoshop?

[magneto]

Thanks old school. I thought Mark Elendt only worked on Mantra. I didn't know he was involved in other areas. That's great to know.

I wasn't aware PRISMS lacked networks that Houdini have. Digital Assets are also very interesting to me because at first I thought they were designed from day 1. Later on I read that it was introduced much later. Still it doesn't feel like an afterthought feature.

Will Houdini move to heterogeneous networks in the future? I never thought of this but how would that work? I mean you have Add VOP, Add SOP, Add COP, all of which have different parameters and inputs. Also there is the difference between VOPs and other contexts. I should check out TouchDesigner I guess. I always thought that was a watered down very small subset of Houdini, kind of like Winamp Visualizers.

One thing that's confusing to me is POPs though. Some people suggest not using them and doing it in SOPs where I can see the advantages. I read that POPs are not as up to date or complete and will be obsolete in the future, etc. I am not experienced in particles yet but is the POP context an afterthought that will likely to be replaced in the future?

To me particles are also points and could be worked within SOPs. Then you don't have access to the same local variables that POPs provide. But I feel like these could have been added into SOPs if POPs didn't exist in the first place.

This division between POPs and SOPs is interesting, but I think it also creates some confusions as some people here strongly feel about trying to keep particles in SOPs exclusively.

It would be cool if we had more info discussing this

[anim]

...

One thing that's confusing to me is POPs though. Some people suggest not using them and doing it in SOPs where I can see the advantages. I read that POPs are not as up to date or complete and will be obsolete in the future, etc. I am not experienced in particles yet but is the POP context an afterthought that will likely to be replaced in the future?

...

as you could have already noticed there are some POP like operators in DOPs in H12.5 (mostly done using SOP Solver DOP or Geometry VOP DOP)

there is:

POP Source DOP

POP Replicate DOP

and as well some other that are hidden by default:

POP Axis Force DOP

POP Color DOP

POP Drag DOP

POP Fan Cone DOP

POP Flock DOP

POP Force DOP

POP Integrator DOP

POP Location DOP

POP Split DOP

you can unhide them by editing OPcustomize

so this may be some indication that POPs may get replaced/merged with DOPs in the near future

[dennis.albus]

We still haven't got to the "heterogeneous"network bit (just too much too soon at the time) although Greg at Derivative has implemented heterogeneous networks with Touch Designer where you can pace different node types in the same network and wire sensible things together. Hopefully one day...

Wow, that sounds really tempting. Is this something that's planned for the future of Houdini? If so I would like to put all my votes for RFEs on that one

-dennis

[old school]

Wow, that sounds really tempting. Is this something that's planned for the future of Houdini? If so I would like to put all my votes for RFEs on that one

No. Not a priority.

Since we allowed different network types embedded in other networks way back in H5 I believe, you can create pretty compact workflows and with digital assets, you can wrap any network type up and include it as a specific operator. For example take a DOP network and wrap it up in to a SOP with four inputs each dedicated to say sources, colliders, forces and sinks and there you go.

If you do want to have a whirl, go to the Derivative web site and give it a go.

[old school]

Thanks old school. I thought Mark Elendt only worked on Mantra. I didn't know he was involved in other areas. That's great to know.

Mark is the key architect of all things Geometry and that means all things Houdini.

Mantra is as much about procedural geometry as it is about raytracing/micropolygon rendering. Many current render engines are not efficient at refining geometry. Mantra is.

I wasn't aware PRISMS lacked networks that Houdini have. Digital Assets are also very interesting to me because at first I thought they were designed from day 1. Later on I read that it was introduced much later. Still it doesn't feel like an afterthought feature.

Prisms did have networks but represented them as a stack. Each SOP would have inputs, one, two, three, etc. But the inputs were actual menu lists of all other SOPs in the tree and from all other objects so in essence it was a network but represented as a stack. You can see this in Max, Maya and C4D stacks how they do it as well.

There was a converter that we supplied that converted Action .mot files in to .hip files as all the SOPs were ported over with the same interfaces and many with the same code. I still have quite a few of these files that I converted way back. Actually some of the help cards for the Point SOP are from these converted Action files from PRISMS.

Networks were far easier to work with than the stacks, trust me on that. You can create fare more complicated networks and actually manage them.

Will Houdini move to heterogeneous networks in the future? I never thought of this but how would that work? I mean you have Add VOP, Add SOP, Add COP, all of which have different parameters and inputs. Also there is the difference between VOPs and other contexts. I should check out TouchDesigner I guess. I always thought that was a watered down very small subset of Houdini, kind of like Winamp Visualizers.

No. Read my post above.

Yes check out touch designer. Especially if you are interested in real-time graphics.

One thing that's confusing to me is POPs though. Some people suggest not using them and doing it in SOPs where I can see the advantages. I read that POPs are not as up to date or complete and will be obsolete in the future, etc. I am not experienced in particles yet but is the POP context an afterthought that will likely to be replaced in the future?

To me particles are also points and could be worked within SOPs. Then you don't have access to the same local variables that POPs provide. But I feel like these could have been added into SOPs if POPs didn't exist in the first place.

POPs are fine, especially when working on tricky behaviour stuff. They aren't going to be deprecated any time soon.

H12.5 has some interesting ideas in this area with the POPs in DOPs being utilized by the FLIP solver.

This division between POPs and SOPs is interesting, but I think it also creates some confusions as some people here strongly feel about trying to keep particles in SOPs exclusively.

It would be cool if we had more info discussing this

This would be for performance reasons.

POPs is single threaded so there is the inherent limitation. VOPs are threaded so as more and more POP operations being collapsed in to POP assets, it only makes sense to see this evolve in to SOPs explicitly in DOPs and get threading essentially for free by exploiting VOPs and SOPs.

Due to the inability to write expressions on DOPs, we added the Snippet VOP which allows you to wrangle vex code at the top level in DOPs/SOPs/COPs/etc which is far more efficient than writing expressions in SOPs or POPs which may not be possible to thread. Plus expressions in parameters on DOPs is not supported. Wrangle it with vex!

Have a look at the POP Integrator DOP. I always wanted access to the POP integrator, but no. Now we do, at least in an early prototype unsupported form.

POPs are a rules-based network type to perform time based simulations. They work on a particle primitive type which back in the day was mandated as Houdini at one time didn't allow arbitrary attributes (we're talking H1 to H3) and to make things as efficient as possible, had very unique attributes such as the one-off age and life 2-float "life" attribute which doesn't really fit in today's Houdini world. Remember in 1998 SGI's and WindowsNT had limited memory compared to today so you had to pack a lot in to a small footprint to make it usable for Feature Film requirements .

SOPs are about evaluating geometry at the current frame although this is getting blurred more and more. With DOPs, Houdini supports infinite timelines and DOPs is becoming the simulation go-to network type.

SOPs can be made to be recursive working on the previous time step's data when put inside a SOP Solver DOP (of which the Solver SOP is a wrapper of). That is how you use SOPs to do particle and geometry simulations. So to be fair, when you compare SOPs to POPs, you should say SOPs_in_DOPs vs POPs.

[magneto]

Thanks alot old school. Your post really clarified a lot of things. I am sure people would agree with me that you might have joined late to the party but you are still the face of SESI

Btw I can't find the post but I remember reading not long ago in the forums how you wanted a particular feature or something you have implemented (?) in Houdini but couldn't when answering a question. Based on your position, I would have thought what you say/wanted would be the law of the land? I understand companies generally don't function like this, but given SESI is a small tight-knit company, I thought it might be different.