There aren't any purpose-built musle tools right now in Houdini. Are there plans? We haven't seriously approached this part of character rigging yet.
That being said, there have been some fantastic muscle rigs built in the past in both Prisms and Houdini. The tools are there but it is currently up to you to implement your own system. As a brief start, you can parent metaballs to bones then scale with boneangle() using bulge or magnet SOPs. It can get much more complicated and predictable depending on your approach.
I would imagine the biggest advantage to a product like ACT is that they have musle rigs pre-built for you already. This is where the time is spent: Placing muscles inside your character and fine-tuining their effect on the skin.
Now for some history. :wink:
There was Digital Domain and their “HARD” project driven in part by Caleb Howard and Academy Award Winner Kevin Mack way back in 1994 using PRISMS!
David Oliver also pioneered a muscle system with help from some R&D for the Disney Dinosaur project. From what I understand, David was able to acomplish with Prisms what took quite some time of hard development with what would become maya and they still didn't capture all of what was in the original Prisms solution.
Most recently, Primal Pictures developed their own muscle system tools for Houdini that are probably more accurate than those in ACT. Very clever those Primal guys. Many more muscle projects inbetween of varying complexity.
The bulge SOP actually came out of C.O.R.E.'s custom tool for doing muscle simulations.
Found 7 posts.
Search results Show results as topic list.
Houdini Lounge » Character tools
- old school
- 9 posts
- Offline
Technical Discussion » Randomizing points randomly using POintSOP
- old school
- 9 posts
- Offline
Ed's multiply is fine but I have always used random offsets like:
$TX + (rand($F + 1.01)*1000)
$TZ + (rand($F + 10.01)*1000)
Since rand goes from 0 to 1, you will certainly see a skew to the upper right looking down the z axis.
To fix that, use the fit function as follows:
$TX + fit(rand($F + 1.01), 0, 1, -1, 1)*1000
$TZ + fit(rand($F + 10.01), 0, 1, -1, 1)*1000
The fit works like this:
fit (float num, float oldmin, float oldmax, float newmin, float newmax)
Return a number between newmin and newmax that is relative
to num in the range between oldmin and oldmax. If the value is
outside the old range, it will be clamped to the new range.
eg. fit(3,1,4,5,20)=15
see also: fit01 fit11 fit10
$TX + (rand($F + 1.01)*1000)
$TZ + (rand($F + 10.01)*1000)
Since rand goes from 0 to 1, you will certainly see a skew to the upper right looking down the z axis.
To fix that, use the fit function as follows:
$TX + fit(rand($F + 1.01), 0, 1, -1, 1)*1000
$TZ + fit(rand($F + 10.01), 0, 1, -1, 1)*1000
The fit works like this:
fit (float num, float oldmin, float oldmax, float newmin, float newmax)
Return a number between newmin and newmax that is relative
to num in the range between oldmin and oldmax. If the value is
outside the old range, it will be clamped to the new range.
eg. fit(3,1,4,5,20)=15
see also: fit01 fit11 fit10
Technical Discussion » light rays
- old school
- 9 posts
- Offline
Put an Atmosphere (Fog) Object in your scene.
In SHOPs, place a Lit Fog SHOP. Assign the Lit Fog to your Atmosphere Object.
In the SHOPs parameters, first set the Fog Density to a value of 0.2 to 0.4 to see the effect. The default value of 0.05 is a bit subtle for a first test.
Set the Max Steps to 100 to test out the effect when rendering. When you get the right look, up the steps until you get what you want. 100 or less may do and renders much faster. A bit of blur in the compositor hides a lot.
The Step Size can be left at default. This is a ray marching shader in that the ray marches along at discrete steps sampling all the lights every time, summing up the light values as it marches along. Decreasing the Step Size will slow down your render whereas increasing the Step makes the render go faster.
All your lights will now render as volumetric lights. If you set your lights to cast shadows, say fastShadow in the Shadow Shader field, then you will get the nice streaks in the lightbeams.
In SHOPs, place a Lit Fog SHOP. Assign the Lit Fog to your Atmosphere Object.
In the SHOPs parameters, first set the Fog Density to a value of 0.2 to 0.4 to see the effect. The default value of 0.05 is a bit subtle for a first test.
Set the Max Steps to 100 to test out the effect when rendering. When you get the right look, up the steps until you get what you want. 100 or less may do and renders much faster. A bit of blur in the compositor hides a lot.
The Step Size can be left at default. This is a ray marching shader in that the ray marches along at discrete steps sampling all the lights every time, summing up the light values as it marches along. Decreasing the Step Size will slow down your render whereas increasing the Step makes the render go faster.
All your lights will now render as volumetric lights. If you set your lights to cast shadows, say fastShadow in the Shadow Shader field, then you will get the nice streaks in the lightbeams.
Technical Discussion » Group Sop "Transfer Selection to Pattern" gone.
- old school
- 9 posts
- Offline
This is long gone from Houdini now. The change happened in H5.0.
It is replaced with the new modelling workflow. You need to use the “Transfer Selection” red arrow in the viewport pane's left hand side, mid way down.
Press it, select your geometry, then tap the RMB and you now have set the selection for the Group SOP.
As for the docs, I see your point. :?
I will submit the bug
It is replaced with the new modelling workflow. You need to use the “Transfer Selection” red arrow in the viewport pane's left hand side, mid way down.
Press it, select your geometry, then tap the RMB and you now have set the selection for the Group SOP.
As for the docs, I see your point. :?
I will submit the bug
Houdini Lounge » Character Animation tools in houdini
- old school
- 9 posts
- Offline
I guess not, (new feature for them), we are not talking about hobbyists or singal artists, but about major studios who spends thousands on training, a guess the word NEW wouldn't fit !
Bear in mind that there are a few distinctions to be made when it comes to character work. The more insignificant the character, the more likely it will go to Maya 100% as the rig has to be developed to a point where the character moves in a believable fashion.
For full-blown feature animation, it is a whole different story. The character rig is the key and tools the software gives you to help out in this area are of primal importance. This is where MEL integration in to Maya is brilliant. It has taken Houdini several years of constant improvement to reach this stage, probably because Side Effects realised this and focused on improving the UI to the state where they could add those tools to rig a character without a mel-like programming language.
Pixar uses proprietary tools for animating their features. So does PDI. These are the two big successfull Animation shops, right? Why don't they use Maya for animating charcters? Training, costs, and a proven workflow with existing tools. In-house tools are already developed and Maya presents nothing new to them, so why switch? That is currently Houdini's problem. The classic chicken vs. egg problem. ILM has their own tools as well but for animation, they work with both Maya and Soft, but just for pushing bones around and generating animation channels to a great extent.
If a character animator for feature film is presented with a well implemented rig and workflow and they can meet their production quota's, then they will be happy. It is meeting the production quota's that concerns experienced animators the most so when presented with a new package to animate in, they have severe reservations and deservedly so. It is up to the riggers to settle the animators down by listening to their needs and providing the production tools to deliver the shots.
I am convinced that you can rig successfully in Softimage, XSI, Maya and now Houdini as well. All the packages have the proper hooks to do this now.
so, my point is that Maya has the same tools found at power animator, but they are more advanced.
No, you are wrong. The biggest complaint with Maya1 was that it did not have the full complement of PA tools. As a matter of fact, it was sorely lacking on the NURBs side and took release 2.5 to finally get the basic subset of tools in place. They promised interactive viewport rendering IPR but users had to wait over a year to get that in place. As for particles, they still don't duplicate Dynamation's power and flexibility without extensive MEL programming. The particles are easier to use though. Many more shortcomings in maya 1.
The point is that Maya was not nearly as complete as you claimit was when released.
Why Maya clobbered Soft is simple. Speed. Sure wasn't cost. Maya truly was 3x faster and more than Soft was at the time. It also had mel so you could customize the tool to suit many styles of character rigging. It also didn't happen that Soft had a new owner every year, or so it seemed.
Utilizing the OpenGL talent at SGI, Alias programmers were able to take advantage of efficiencies in OpenGL that fit with their character workflow. Houdini didn't implement OpenGL improvements untill it's second and third major releases. Remember that first siggraph with the robot kicking the can? The playback was phenomenal for the time, and it kicked the can to boot! The large studios could measure the savings in time. A no brainer to switch with no matching offering from Soft. Even given this, there were still Soft die-hards who wouldn't switch for various reasons that baffled me at the time but time has given me new insight: Very few of us are capable of real change once a certain comfort level is attained. Protecting that comfort level becomes a primary daily activity unless pushed to the brink. Oh well, to each his own.
Don't forget that I posted here to check the possibility of transiting to Houdini, am not trying to prove anything here, not the most powerful or least or anything, I do believe that the artist is the one key not the software.
Exactly. Experience is the key. Experience gives you the opportunity to choose different methods with experience to back up each decision. If the animator is worrying about posing a character and pushing him through the shot rather than worrying about how the character is behaving, the problem generally is with the animator.
If your choices are truly being hampered by the software, then seek alternatives, like Houdini.
I have witnessed many animators transition from other packages to Houdini for series production work. It is interesting to see who succeeds and who doesn't. Out of all the animators (around 100 now), only four or five bailed claming that Houdini was too much, and that would have been with Houdini v3! They all seemed to get frustrated and were extremely resistant to change in any way, shape or form. The current success rate is much much better. Only two in the last year and that is with outdated rigs in one specific case.
The point is: what you get from your software? How easy? And how fast?
You have to add in the change factor; the point of view of transitioning packages. Houdini is now in the same field as maya for animation (pros and cons to both) but doesn't present a radical improvement (3x or more) in speed and efficiency. You will continue to use your existing tool then. If you wanted to tap in to Houdini's procedural power, then the transition is worth it as you don't need to master mel to do the advanced stuff like muscles and advanced tertiary animation effects.
Well .. MR isn't that bad, you probably didn't even try to render with it, now it's fully integrated with Maya 5.0 providing direct access from within Maya, just as XSI.
MR has the same problem with PRMan that houdini has to crack in to the Maya market for Feature Film. It is already established and MR doesn't provide that 3x or more increase in efficiency (speed) or cost-effectiveness. As a matter of fact, with GI in renderman, MR has slipped back a bit in that regard. Yes, MR has implemented better motion blur now but in order to get that magic 3x multiplier, they will have to be on par and then cost practically nothing to use. This is fine for Mantra as Side Effects gives mantra for free but they sell Houdini. Mental Images has MR as their only base for income so therein lies their dilemma.
The proof is in the adoption of the RAT tools. At $10,000 US a pop, it is certainly not a cheep option. Many shops use Houdini as the pipe to generate the RIB to forgo this expense and just pass maya data through the non-graphic houdini called hscript to output rib. Varies by team and by facility. It seems that every TD has his own way of doing things that works for them. Is it the most flexible way? Doesn't matter to them if it gets the job done. Yes there are other solutions to output RIB from Houdini but they too cost money and rib output is not brilliant.
Smaller shops using Maya and not satisified with it's renderer now have two options: PRMAN or MR. Good for them that this choice now exists.
Feature film is an entirely different beast. Proven production shaders for PRMan can now contain GI calls to heighten the realism, and the render times. No new programming language to learn for those shader writers. Don't have to rewrite 100's of shaders to support a new paradigm as MR presents the shader writer. Cost of implementing MR is then too prohibitive for many teams at the SFX shops.
Pixar filled that hole. So for studios using Maya, PRMan has become the renderer of choice, and Pixar is selling plenty of licenses"What hole? I thought that PRMan was always the renderer of choice. :wink:
Broadcast, yes RAT makes sense at that cost. Feature film. Sorry. The hardened shader writers for Features generally use vi-vim or emacs as their shader tool of choice. FYI vim and emacs are raw programming shell text editors. They see those GUI tools as an impedement, not a productivity boost. You either know how to write shaders, or you use RAT. Who do you think will develop the better, more efficient custom shaders?
One note, I consider the VEX Builder in Houdini to be extremely close to writing shaders. An amazing tool that generates excellent shader code to learn and presents the best stepping-stone to get in to writing programmable shaders if that is one of your goals. The VEX language is so close to PRMan's shader language that it is frustrating when you need to use both tools daily. Remembering all the subtle differences, that is.
if possible provide me with the most complex and amaizing animations made with Houdini, the ones hard to be done in other packages (weather fully CGI or combined with a film) ..
Now we are running in the bullshit part of this industry. This will quickly become a pissing contest that I don't really care for.
Let's look at it from a SFX team point of view. A team has to put out a bunch of shots along side other teams. Wether that team is technically (users) or politically (TD lead) motivated to use a particular package, so be it. Some teams do characters while others wrap the SFX around those characters, others just do clean SFX shots. One team animates a character, perhaps doing the skinning or the muscles, textures, etc. or not is an issue.
It is better to look at the failures as an indication of true limitations. We all learn from those! Nothing like a kick in the teeth to make you think twice next time, if you don't have anyone or anything else to blame that is.
These teams also go in over their heads occasionally and try to do some of the SFX for that character or shot, say feathers on a hero bird. They program the snot out of the package trying to lay feathers on a bird but proceed to spend 6 months and countles amounts of $$$ in the process. Finally a Houdini team, already ahead of it's schedule volunteers to tackle that part, and succeeds. They even get some partial animation!
Example 1: The phoenix in HP2
Now we have a serpant that has to have particles run down it's surface in a believable way accessing painted texture maps like displacements, force and drag. Get some animation too!
Example 2: Serpant in HP2
It just isn't cost effective to use Houdini talent to animate given the lack of users out there.
There are countless other examples where Houdini was used to play a vital supporting role in characters where if Houdini were not in the equation, a sub-standard solution would have to had been developed with costly R&D talent. R&D just aren't as creative as artists as most of their time is devoted to providing a solution, a single solution and little is left for the creative side of the craft and they are left somewhat frustrated. Pray to GOD that the solution works. Hope that paper we are copying works under all circumstances solution. R&D just doesn't have the cycles to execute alternate creative options. Houdini technical artists do. The experimental nature of houdini's networks support this concept time and again. Houdini showcases the artists' true talent and quickly becomes a transparent solution, except when a bug is encountered. Off to support to get it fixed… In a couple days! Even sooner if it impacts production seriously.
Generally, if Maya fails, Houdini is there to pick up and move the shots to completion but at a cost as the Houdini teams are generally challenged with the most difficult shots to begin with.
Alias have and will continue to spend millions to help develop their market presence, but in the end, is that money being spent on developing the package for Feature Film? Maya artists don't seem to care all that much as the product is affordable now and the character tools are robust enough . It is a completely other story for the CTO's of those large companies who see maintenance fees spent away in areas not critical to their own needs. This is evolving right now as a serious dynamic in many of the large shops. How much cheaper can Maya get. Free? If Alias sees the Film market as a lost-leader, then yes.
We need more competent Houdini artists in this community, plain and simple. :?
Another thing about Houdini and PRMan is, that Houdini's output to RIB is way better than Pixar's own RAT.
The best on the market actually.
am waiting for the prove.
If you can read rib, the proof is right there. Read the RenderMan companion and Advanced RenderMan books. Then read our rib. It is readable! It makes sense. Try reading MTOR's rib… :wink: They both create nice pictures through PRMan. One is in a format that you can actually maintain efficiently inside a pipeline. Read as: “Large shops like maintainable rib and small facilities simply don't have the resources (talent) to work at that level.”
One interesting development in recent months is that some new features in PRMAN seem to be predicated on supporting RAT and MTOR. Is this in the best interest of generating clean rib? Does PIXAR see this as a requirement for RAT and MTOR? Given the audience that uses these tools, probably not that great a concern. Feature Film artists? Unknown right now.
Go to a SIGGRAPH RenderMan user group, when they ask the perrenial question: “Who writes their own shaders”, half the group raises their hands. It just so happens that many of them are waring the familar black Houdini “T-ees”. The other half giggles and laughs that they don't have to write code. >>> Maya users… :x I guess those PRMAN users are proud of their ignorance. In a way, you don't want an artist writing his/her own shaders so MTOR is a blessing.
Maya and Houdini are great tools. Houdini just let's you do more without learning to read or write code. From an animator's point of view, this could be quite liberating depending on what they want to do with their characters. The plethora of MEL scripts on the web are great, but they do have a cost if they fall over in some cases.
What do you want to do with your character today?
Technical Discussion » Canning animation for Copy SOP
- old school
- 9 posts
- Offline
If you want to access attributes bound to geometry but you are using a file SOP to load the geometry in, then you will have to add a “signature” to the data so that SOPs can directly access the attributes.
Right after the File SOP, isert an attribute create SOP and put the name of the actual attribute as found in the geo info (MMB-hold on tile icon) in the “Name” field, set Type to be Vector then use $LIFE1 $LIFE2 0 in the last three fields.
I bet this is why you can't directly access $LIFE.
Right after the File SOP, isert an attribute create SOP and put the name of the actual attribute as found in the geo info (MMB-hold on tile icon) in the “Name” field, set Type to be Vector then use $LIFE1 $LIFE2 0 in the last three fields.
I bet this is why you can't directly access $LIFE.
Houdini Lounge » Influence Object Skinning
- old school
- 9 posts
- Offline
I don't know about influence objects much but to push skin around, use objects with metaballs in them.
Use the object merge SOP to bring them in to your object with the skin and then use the bulge SOP to push the skin around. Please make sure that the transform parameter is set to “this object” for H5.5 and for H6, just put a dot “.” in the field to indicate “this object” (if you know shell paths then . means right here and .. means one level above me).
You can simulate sliding and stuff by moving either the metaball or the skin, or both around. Only the space that the two occupy is that counts. You can push the metaball right through the surface and it will keep on pushing. The other SOP is the Magnet SOP with similar behaviour to the bulge but with more controls.
Are maya influence objects stuck to the skin? I remember hearing that you need to specify the points to affect.
Use the object merge SOP to bring them in to your object with the skin and then use the bulge SOP to push the skin around. Please make sure that the transform parameter is set to “this object” for H5.5 and for H6, just put a dot “.” in the field to indicate “this object” (if you know shell paths then . means right here and .. means one level above me).
You can simulate sliding and stuff by moving either the metaball or the skin, or both around. Only the space that the two occupy is that counts. You can push the metaball right through the surface and it will keep on pushing. The other SOP is the Magnet SOP with similar behaviour to the bulge but with more controls.
Are maya influence objects stuck to the skin? I remember hearing that you need to specify the points to affect.
-
- Quick Links