Hello,

In Softimage I could drive a rig with a clip from the mixer and later apply/bake the result onto the affected controls and remove the clip's influence if needed.

I finally figured out how to store a clip in houdini but copying the overriding keys from the CHOPS network onto the parameters of the rig it is driving is not as obvious as I hoped.

Is there any way to “plot” or “bake” a clip coming from a “math” node (I use it to add animation from the source channel operator to the destination channel operator)in the CHOPS network (that forms when you right click > motion FX > create clip)?

mwessels
Hello,

In Softimage I could drive a rig with a clip from the mixer and later apply/bake the result onto the affected controls and remove the clip's influence if needed.

I finally figured out how to store a clip in houdini but copying the overriding keys from the CHOPS network onto the parameters of the rig it is driving is not as obvious as I hoped.

Is there any way to “plot” or “bake” a clip coming from a “math” node (I use it to add animation from the source channel operator to the destination channel operator)in the CHOPS network (that forms when you right click > motion FX > create clip)?

As i already told you you can create a pose from animated clips as well(instead of create clip, choose create pose; this will make an ANIMATED pose). From there on you can apply the pose to any identical rig. The method i showed you was in response to your question on how to TRANSFER the motion from one rig to another. Once the motion is transferred you don't need to ‘bake’ anything. Houdini is procedural, you need to adapt to this new type of thinking. You can modify everything in the source clip and it will transfer along the network. You can filter, or in any other way alter the initial clip and it will automatically transfer to the target. The best part is that you do this with nodes, but your original clip is untouched so you can alter it, or revert to it as you wish. A lot of this topics where covered in the animation section in conversations between me and Jordi(thanks Jordi) so you might want to check that topic here on Si forum.

Constantin X

mwessels
Hello,

In Softimage I could drive a rig with a clip from the mixer and later apply/bake the result onto the affected controls and remove the clip's influence if needed.

I finally figured out how to store a clip in houdini but copying the overriding keys from the CHOPS network onto the parameters of the rig it is driving is not as obvious as I hoped.

Is there any way to “plot” or “bake” a clip coming from a “math” node (I use it to add animation from the source channel operator to the destination channel operator)in the CHOPS network (that forms when you right click > motion FX > create clip)?

As i already told you you can create a pose from animated clips as well(instead of create clip, choose create pose; this will make an ANIMATED pose). From there on you can apply the pose to any identical rig. The method i showed you was in response to your question on how to TRANSFER the motion from one rig to another. Once the motion is transferred you don't need to ‘bake’ anything. Houdini is procedural, you need to adapt to this new type of thinking. You can modify everything in the source clip and it will transfer along the network. You can filter, or in any other way alter the initial clip and it will automatically transfer to the target. The best part is that you do this with nodes, but your original clip is untouched so you can alter it, or revert to it as you wish. A lot of this topics where covered in the animation section in conversations between me and Jordi(thanks Jordi) so you might want to check that topic here on Si forum.

I understood your post, and I'm starting to understand the Houdini thinking. But in my experience character animators like to stick with familiar things and simple quick answers since characters animation in essence is difficult enough even before you picked your software of choice. I'm exploring how much of XSI's handy animation workflows can be replicated and if not is the Houdini approach compatible with people that aren't necessarily technically driven.

I'll look for the Animation post between you and Jordi, thanks. For interest sake - you might not need to “bake” out the motion FX result … but can you?

Cheers.

mwessels

Constantin X

mwessels
Hello,


I understood your post, and I'm starting to understand the Houdini thinking. But in my experience character animators like to stick with familiar things and simple quick answers since characters animation in essence is difficult enough even before you picked your software of choice. I'm exploring how much of XSI's handy animation workflows can be replicated and if not is the Houdini approach compatible with people that aren't necessarily technically driven.

I'll look for the Animation post between you and Jordi, thanks. For interest sake - you might not need to “bake” out the motion FX result … but can you?

Cheers.

I just responded to your questions and tried to help you get started(i was in the same position as you few few months ago). It's up to you to see if Houdini suits your needs or not. To me so far is perfect, although i still learn. I will include it in my workflow as my main app in 1-3 months and then will move to 90% Houdini and 10% Cinema 4d or Modo. I don't know if you can bake animations into Houdini, i just don't need to do it. I construct library of poses(both static and animated) and just re-apply them as needed. When you work with NETWORKS you don't need to freeze, delete history and bake. It's a different type of workflow…

Constantin X

mwessels

Constantin X

mwessels
Hello,


I understood your post, and I'm starting to understand the Houdini thinking. But in my experience character animators like to stick with familiar things and simple quick answers since characters animation in essence is difficult enough even before you picked your software of choice. I'm exploring how much of XSI's handy animation workflows can be replicated and if not is the Houdini approach compatible with people that aren't necessarily technically driven.

I'll look for the Animation post between you and Jordi, thanks. For interest sake - you might not need to “bake” out the motion FX result … but can you?

Cheers.

I just responded to your questions and tried to help you get started(i was in the same position as you few few months ago). It's up to you to see if Houdini suits your needs or not. To me so far is perfect, although i still learn. I will include it in my workflow as my main app in 1-3 months and then will move to 90% Houdini and 10% Cinema 4d or Modo. I don't know if you can bake animations into Houdini, i just don't need to do it. I construct library of poses(both static and animated) and just re-apply them as needed. When you work with NETWORKS you don't need to freeze, delete history and bake. It's a different type of workflow…

Thanks for taking the time to help, I really appreciate it. I will try work with it as you suggest without breaking my head over baking it - the non-destructive network approach.

Man, I can only imagine what awesome rigs you could build in Houdini… and changing a rig is also a piece of cake since at it's core a rig will always be modular in Houdini.

mwessels

Constantin X

mwessels

Constantin X

mwessels
Hello,

Thanks for taking the time to help, I really appreciate it. I will try work with it as you suggest without breaking my head over baking it - the non-destructive network approach.

Man, I can only imagine what awesome rigs you could build in Houdini… and changing a rig is also a piece of cake since at it's core a rig will always be modular in Houdini.

Glad to help you, i received help when i needed too. Well, the killer thing is that you can change the geometry and weight paint even AFTER the animation is completed. Because they are just nodes on the network. It's mind boggling what you can achieve. I recently had a project, and after client approved modeling stage i went into animation. AFTER i finished the animation too, client wanted some ‘small’ changes to modeling. This is the ‘bang your head to the desk’ moment that made me decide the switch to Houdini. Because if i had the animation and modeling in H i could just modify the input geo and the animation with be kept….
Cheers,
C

http://forums.odforce.net/topic/20483-baking-expressions-to-keys-for-export/?p=122328 [forums.odforce.net]

sorian
http://forums.odforce.net/topic/20483-baking-expressions-to-keys-for-export/?p=122328 [forums.odforce.net]

Handy link, but the one step I'm missing in Houdini is the “adjust parms if needed then -> Copy To Export Destination to bake the keyframes”

The only place I can find “Copy to Export” is in a channel operator and that is usually further up in a network chain and not the last node that handles the export. You don't happen to know where that “Copy To Export Destination” is hiding? The search continues.

The “fetch” and “object” operator this link shows seem pretty handy. I find that you can also get a good transfer results by, instead of “making a clip” you then simply use a fetch node that pipes into an export node - problem still is I can't “copy to export” except from a channel operator which only has an output.

Thanks Sorian.

mwessels
problem still is I can't “copy to export” except from a channel operator which only has an output.

No, i didn't meant that button on the channel CHOP.
look at the picture.
honestly, SESI should implement a bigger and more visible button for such a high usage function for experienced animators.
your final output will be an optimised fcurve. if you want full frame keyframing, i think the only way currently is scripting via a python code.

it would be nice if SESI made the ‘Fit Panel’ button bigger, then do add an option there for full frame keyframing.
look at the attached file for more info.

sorian

mwessels
problem still is I can't “copy to export” except from a channel operator which only has an output.

No, i didn't meant that botton on the channel CHOP.
look at the picture.
honestly, SESI should implement a bigger and more visible botton for such a high usage function for experienced animators.
your final output will be an optimised fcurve. if you want full frame keyframing, i think the only way currently is scripting via a python code.

it would be nice if SESI made the ‘Fit Panel’ button bigger, then do add an option there for full frame keyframing.
look at the attached file for more info.

O my word haha, that is an incredibly subtle button! Amen to all the request to make it more prominent. I'm so used to hovering with my mouse over a button and getting a tooltip, but this “Fit Panel” button gave me no tooltip so I moved on, so close… The rename node is great, another thing I would not have guessed to use.

Dude, I really appreciate your effort, sticky note annotations in the network view, the works - awesome. High fives and thank you.

So I've been trying to tackle this now and unfortunately I have to say that it's nowhere near usable enough for character animation.

Firstly a comment towards what Constantin said, that you don't need to bake animation. This is simply not true. The main reason for needing to save animation clips, is to apply them to another shot with the same character later on. This is however not a one to one thing and always requires a lot of tweaks on top to make a shot work, hence you're only using the saved animation as a starting point to save time. For example you save clips of character running, walking, jumping etc. Then you need to quickly put them on a character in a new shot, bake it (ideally preserving keyframes as in the original clip) and continue working from there using standard animation techniques. Animation is mostly a destructive and artistic process and it needs to stay like that to preserve some character and differences between shots. You don't want a bunch of shots with exactly the same run, but rather do small tweaks on top.

Now to the actual workflow here. Yes saving a clip is fairly straightforward, even though I didn't find a quick way to only save keyframed channels for instance. It's either parameter by parameter, or all of them at once if you change the channels string to something like /obj/Rig01/*.

Applying them to another character is still quite easy with the use of rename chop for instance. But actually baking it is a mess. Using the ‘fit panel’ unfortunately doesn't preserve the animation exactly as it was saved, because it resamples it and then recreates the keys I suppose. The point it that if I save a subtle animation with ease ins and ease outs of a characters face. When I apply it back using this workflow delivers very different results that the original, hence not usable in many cases.

This is exactly one of those areas where there's very little benefit in trying to keep everything opened and procedural. What animators need is, save animation from a rig to disk as whatever file format. Open another scene 2 weeks later, load it from disk onto a character, ideally combine a few of them on a time line and continue animating from there.

We're currently evaluating Houdini for usage on a full cg tv show with lot's of character animation, but it looks like for it to be of any use for fast animation turnarounds I'd have to bite the bullet and write fully custom animation saving and loading tool, with simple to use Pyqt UI.

I'll very much appreciate any insight into things I've mentioned. Even though I have the feeling that there isn't a lot of masochist trying to actually animate characters in Hou.

mkolar
This is exactly one of those areas where there's very little benefit in trying to keep everything opened and procedural. What animators need is, save animation from a rig to disk as whatever file format. Open another scene 2 weeks later, load it from disk onto a character, ideally combine a few of them on a time line and continue animating from there.

Have you tried the Export Displayed Channels menu item? The hotkey is > (greater than) and the import hotkey is < (less than). Importing will only load the saved channels onto the currently scoped ones that have the same name (as I recall) though.

There's the new pose library in H14, probably need the latest daily build to do a good test.

I have indeed. It works most of the times if the animation is being loaded to the exact same character. It seems to me that it's not applying it back based on parameter names, but rather their order in the scoped channels, hence it breaks really easily. If one animator uses ‘create nested channel groups’, but another doesn't, they will most likely have slightly different order in scoped channels, rendering this tool useless. If it went purely by name. I could save anim or pose for the whole character, then scope only arm and loading animation back would apply it only to arm (because it would only find those parameters for matching by name.)

Now this would be actually useful and if otworked it would be certainly a step in the right direction.

In terms of pose library. I know it's been rewritten in h14, but I can't find any information at all about it in docs. I can't find how to access it (it's not on the character shelf), hence can't try using it . Again any tips on where to find it would be great.

mkolar
So I've been trying to tackle this now and unfortunately I have to say that it's nowhere near usable enough for character animation.

Firstly a comment towards what Constantin said, that you don't need to bake animation. This is simply not true. The main reason for needing to save animation clips, is to apply them to another shot with the same character later on. This is however not a one to one thing and always requires a lot of tweaks on top to make a shot work, hence you're only using the saved animation as a starting point to save time. For example you save clips of character running, walking, jumping etc. Then you need to quickly put them on a character in a new shot, bake it (ideally preserving keyframes as in the original clip) and continue working from there using standard animation techniques. Animation is mostly a destructive and artistic process and it needs to stay like that to preserve some character and differences between shots. You don't want a bunch of shots with exactly the same run, but rather do small tweaks on top.

Now to the actual workflow here. Yes saving a clip is fairly straightforward, even though I didn't find a quick way to only save keyframed channels for instance. It's either parameter by parameter, or all of them at once if you change the channels string to something like /obj/Rig01/*.

Applying them to another character is still quite easy with the use of rename chop for instance. But actually baking it is a mess. Using the ‘fit panel’ unfortunately doesn't preserve the animation exactly as it was saved, because it resamples it and then recreates the keys I suppose. The point it that if I save a subtle animation with ease ins and ease outs of a characters face. When I apply it back using this workflow delivers very different results that the original, hence not usable in many cases.
This is exactly one of those areas where there's very little benefit in trying to keep everything opened and procedural. What animators need is, save animation from a rig to disk as whatever file format. Open another scene 2 weeks later, load it from disk onto a character, ideally combine a few of them on a time line and continue animating from there.
We're currently evaluating Houdini for usage on a full cg tv show with lot's of character animation, but it looks like for it to be of any use for fast animation turnarounds I'd have to bite the bullet and write fully custom animation saving and loading tool, with simple to use Pyqt UI.
I'll very much appreciate any insight into things I've mentioned. Even though I have the feeling that there isn't a lot of masochist trying to actually animate characters in Hou.

This where already debated, over and over again in the animation improvements dialog, between me and Jordi Bares and few others. As a side note, i have an animation studio with my own optical MOCAP system and i done hundreds of animations(literally) so far. So far my tool was Softimage, because Maya is subpar in this domain. So i am not sure i need advises on how to do animations or how/what tools i need in my workflow, generally speaking. I am not sure what are you trying to achieve, or what are you trying to do, but i have a library of clip poses and i do transfer them between characters, and no i don't need to bake anything for this. I can then alter them inside CHOP network as i find suited. Jordi also does a lot of MOCAP in his studio and is very pleased with tools inside Houdini(you can use MIDI devices and you can alter however you want). So instead of starting a war, and stating that others don't know what they are doing, maybe you will want to study the issue a bit more. I was in the same boat and all i was asking was for, guess what, DESTRUCTIVE WORKFLOWS. I finally came to the conclusion that is not needed, so i am now happy with Houdini. What i am trying to say to you and others, is that you can save library of clips without baking anything. You transfer them to your new character, and then you alter as you see fit. All is procedural done with the help of CHOPS. Best part is that you have your original animation(be it MOCAP or keyframes) untouched, and all changes are done trough the usage of nodes. What you fail to understand is that the clips in XSI are just a way to store things. You can always go inside a clip and alter curves and keyframes after you done the clip. So it's a convoluted system, and the Houdini way is almost always much better. And yea, i filled a request for changing the icons in Motion Viewer but so far no luck… The new animation layer system is a God send and i was asking for this along other SI users, so thank you SESI(a very BIG thank you that is). There is one topic i started in the SI forum section where i asked how to enable Pose Library in Houdini. There are some steps involved. Check that section first, check the Pose Library function and if that does not help you either….. This is my last post here, i don't want to start a war or something. So peace and all the best.
C

Man you seem to have taken that very personally. I apologize for that as it honestly wasn't anywhere near my intentions. I also can't find where I said that someone doesn't know what they are doing .

I merely said that ‘not needing to bake animation’ (or whatever we call) is not a universal truth. In a traditional animation workflow (which we want to keep because of many reasons. A lack of technically savvy animators being one of them) it sometimes really is desirable or necessary to use ‘destructive workflow’.

Now to your points. Ultimately this is what I'd like to achieve (even 10x less polished version would do.
http://www.studiolibrary.com/ [studiolibrary.com], because such animation workflow is fast and comfortable for animators. They are the ones who need to use, not TDs.

I've seen the topic about Pose library. Unfortunately, this doesn't work on any machine with H14 I've encountered so far. That's why I was asking about the rewritten pose library, because I can't access it. Secondly if something is so well hidden that it doesn't even have a simple way to get to it and is not mentioned in documentation. I'm afraid i have to assume it's not production ready.

So please instead of getting touchy, read the post with calm head again and see if you have actual answers to our problems (saving clip/poses for only scoped parms, importing .chan files scrambles channels …). I'm sure it would be helpful to others too. If you don't never mind. This is a discussion, not a war as you called it.

It's a shame one needs to jump so many hoops for smooth animation workflow, because our animators actually really like houdini for animating apart from these issues. (and slow vieport )

mkolar
(and slow vieport )

Which part of the viewport are they finding slow in H14? Bones display have been optimised extensively now.

mkolar
Man you seem to have taken that very personally. I apologize for that as it honestly wasn't anywhere near my intentions. I also can't find where I said that someone doesn't know what they are doing .

I merely said that ‘not needing to bake animation’ (or whatever we call) is not a universal truth. In a traditional animation workflow (which we want to keep because of many reasons. A lack of technically savvy animators being one of them) it sometimes really is desirable or necessary to use ‘destructive workflow’.

Now to your points. Ultimately this is what I'd like to achieve (even 10x less polished version would do.
http://www.studiolibrary.com/ [studiolibrary.com], because such animation workflow is fast and comfortable for animators. They are the ones who need to use, not TDs.

I've seen the topic about Pose library. Unfortunately, this doesn't work on any machine with H14 I've encountered so far. That's why I was asking about the rewritten pose library, because I can't access it. Secondly if something is so well hidden that it doesn't even have a simple way to get to it and is not mentioned in documentation. I'm afraid i have to assume it's not production ready.

So please instead of getting touchy, read the post with calm head again and see if you have actual answers to our problems (saving clip/poses for only scoped parms, importing .chan files scrambles channels …). I'm sure it would be helpful to others too. If you don't never mind. This is a discussion, not a war as you called it.

It's a shame one needs to jump so many hoops for smooth animation workflow, because our animators actually really like houdini for animating apart from these issues. (and slow vieport )

“ don't want to start a war or something. So peace and all the best”. So you might want to re-read my posts before stating that i want a war(where did you come up with this one???). Again you are offending me by saying that i need a ‘cool head’ . I am not sure what you're trying to achieve here: get help or offend people who are not agree with you? Since you mention Studio Library i do have few comments:
- we are on XSI forum talking about equivalents of XSI tools into Houdini and how to replicate XSI workflows into Houdini so i am not sure why we are talking Maya tools???;
- Studio library is a custom tool witch does not do what you state; it mixes blend shapes with poses for a stated character; so no blendshapes transfer between characters(not sure about poses though);
- you do have the option to build custom tools for Houdini so feel free to do so if you find they are missing;
- for the milionth time, Houdini has built in tools called Pose Library and CHOP motion channels who does exactly what you asked for, including saving poses(fixed or animated) for scoped channels only and transfer them between characters as already mentioned by me and Edward; i already posted a sample where i transfer motion between two rigs with a simple Math node inside CHOP network.
I am with you on one point though: why the heck is the Pose Library so hard to access and not ready available on the shelf tools? I aksed SESI folks to include it on shelf tools out of the box, but they skipped for this version. maybe if you file more requests(bugs and RFE's) stating clear: WE WANT POSE LIBRARY ON THE SHELF TOOLS EASY TO ACCESS OUT OF THE BOX will solve the issue? So please, find the time to fill RFE and bugs requests also, as i already done it…
Please do your best to avoid me, i don't have time for disputes(sorry, but you are offending in your posts towards me).
Thanks,
C

Bones seem absolutely fine, the improvements are very clear in that area. It's mostly textures and deformers when having multiple characters in the scene. Unfortunately the visuals we're working on now are very dependent on textures so animators need to see them (flat characters paper cutout style but skinned to bones, with some blendshapes). Faces and some details are done as textures with transparency on small cards for instance.

It might very well be that my rigs are not fully optimised (almost certainly ), but still when I try similar scenarios in other max or maya I have to say that's is incomparable in terms of viewport speed. As soon as I put 3-4 characters in, it goes down to 12-13 fps on 2GB gtx760. Maya flies with 10-15 characters with textures. I'm planning on doing some testing to narrow it down, but currently we I can't afford it as deadlines are pushing.

MartybNz

mkolar
(and slow vieport )

Which part of the viewport are they finding slow in H14? Bones display have been optimised extensively now.

Houdini is slow on gaming cards. I am on Quadro and it is much faster then Cinema 4D. So maybe that is the culprit…..

mkolar
I've seen the topic about Pose library. Unfortunately, this doesn't work on any machine with H14 I've encountered so far.

Sorry, I was mistaken. There's a new “character picker” in H14, not the pose library.