PDA

View Full Version : Character Animation: deal breakers



jin choung
11-19-2007, 10:48 PM
howdy,

so there are several threads right now talking about CA in lw and ALLLLLLL the stuff it needs/we want....

i thought it might be informative to try a different tack -

WHAT FEATURES, WITHOUT WHICH, MAKE LW CA A NON-STARTER FOR YOU? AS IN DISQUALIFIES IT AS A CONTENDER.

if you think that lw is a capable CA app, then you would have nothing to add.

if you have more than... i'd say 5, you're missing the point of this thread. and provide reason why you feel it to be so.

here are my 3 (yep, narrow it down to this and i have just 3):

1. ROBUST undo/redo system in layout. EVERYTHING reliably undoable and those FEW things that are not undoable are DOCUMENTED AS SUCH!

not having a good undo system makes you TIMID and discourages EXPERIMENTATION. to a degree that you don't properly understand until you've been in an environment that does let you undo reliably and well. this is VERY VERY VERY BAD.

every system, every sub-system inside of layout must be reliably undoable.

2. REFERENCE A COMPLETE RIG - so in one scene, you have a rig with all controls, expressions and drivers set up - no animation. and that completed rig can be referenced into an animation scene someone can start animating.

this allows an animator to get started on a preliminary rig even while the rig is not yet complete... and then, simple re-reference when the rig is updated or completed.

rig revision should not be a CATASTROPHIC prospect.

requiring that one process be complete before another down the pipeline can begin is desirable but hardly ever possible.

3. fast interaction. with all expressions and drivers and iks and deformations in place, and with several rigs in the scene, lw must remain fast and interactive. of course poly count will make a difference but interactivity here needs to be competitive if not overwhelmingly better.

more importantly, performance must scale predictably! if i take out all deformation in my scene and replace with parented proxies, i should see FASTER PERFORMANCE!

the difference between a real tool and a toy is usability. if it has lots of features that you can't realistically use because it slows things down too much, it's just a toy.

that's it for me. just 3 that are deal breakers.
jin

pooby
11-20-2007, 02:07 AM
When I'm doing a rigging (and when I say rigging, I mean just stringing a buch of stuff together to make something, It could be a skeleton, or it could be an Ocean, or anything)
I approach it in a similar way to building something out of Lego technical.

Lego technical is a great design example , because it's designed so that
1- all the parts snap together (IE easy.. no screws and bolts etc, you just snap)
2- there is a coherence to all the parts, they are designed as a set to fit one another.
3-Thirdly, there are all the parts you would need. A comprehensive set.

1- In LW, things are NOT made simple and Easy for you. Just things like the impossibilty of multiselecting and adjusting parameters. Or, linking channels together, you have to go into the Envelope and choose from a limited and confusing range of modifiers.

2- There is NO design or Set of tools in LW, when it comes to rigging. Just a miscelaneous selection of half baked tools that were obviously designed by completely different people with completely different interfaces, and not designed to be used in conjunction with any of the others.

3- there are large gaps in the toolset. Ie no lattices, or animatable constrainable curves etc. It's like having Lego technical, but without any chains, univeral joints, and only one size of cog etc. you get the idea.


If XSI is Lego technical

LW is like having a construction kit made from a few bits of Duplo, a couple of bits of Meccano,stickle bricks, playmobile, maybe one piece of Lego technical ...
- and 3rd parties supply some rolls of sellotape.

jin choung
11-20-2007, 03:07 AM
cool.

but to clarify in my mind then, the lack of a UNIFIED DESIGN is a deal breaker for you? in terms of lw CA?

i have posted relentless rants about the need for software to be DESIGNED and be simple and consistent and elegant... i am AAAAAAALLLLL about that.

but frankly, i've got little hope of it getting addressed.... i mean we got different mousing and keying in different modal windows for heaven's sake!

so the "frankenstein's monster" aspect of lw is something i've just come to... sigh... ... deep breath... accept... ugh.

but it's a deal breaker for ya? if so, cool. i just wanted to be clear.

jin

sammael
11-20-2007, 03:09 AM
I completely agree with pooby whenever I try to rig in Lightwave (which is as rarely as I can get away with to the point that I forget what I did last time) I find that I never feel that I am doing it correctly. It kind of feels like there is no proper or standard way to rig a character you just have to keep hacking away until you come up with something that is vaguely workable.

It is a long drawn out process to come up with a basic functional rig.
I want to see a basic solution which is quick to set up that just works straight off the bat as a decent IK rig complete with pre set limits and handles. Of course something which can be expanded upon with extra bones further adjustment of settings etc but it would be nice just to be able to slap on a rig and test character deformations. None of this naming every bone, typing in every IK limit, adjusting rotations with dodgy tools, setting up handles, trying to remember settings and tool locations etc etc then you accidently change one wrong setting and bam your rig is screwed beyond reapir... hope you remembered to save!

Yep there are plugins but they are **** compared to what I have seen in other packages, max's Biped for example makes rigging a much quicker and easier task also it gives you the feeling that what you have done is correct which to me is important. I don't like doing things one of infinite different ways none of which quite feels quite right or practical.

sammael
11-20-2007, 03:21 AM
Yep so for me it's just the 1 a quick solution to rig setup. There would probably be many more but I rarely get past this first milestone and by the time I do I'm so over it that I cant be bothered animating anything.

ben martin
11-20-2007, 04:58 AM
1 - Decent deform engine - So you can see in real time (without slideshow effect) the deformations/movements in the mesh.

2 - RIG made easy - Just merge Modeler and Layout so people can RIG and set-up/Fine tune the RIGS in the same enviorment/3D space (not switch between Modeler Layout).
Very important note: Skelegons hierarchies in Modeler are easy to draw; this method cold be used to draw bones directly instead of skelegons that must be converted latter ( Any RIG tools must be completely integrated in Layout if MODELER /LAYOUT merge won't be possible/desirable, for some strange reason, right away).

3 - IK solver made easy - Just select the "origin/IK chain start" bone and CTRL + "Click" the chain end bone. Easy as that! Sure that a numeric panel to lock /fine tune IK chains rotation… etc… should be accessible the usual way: hit "N"!

4 - VERY! VERY! VERY IMPORTANT - A decent DOPE-SHEET: something like the image bellow!

Unfortunately Newtek seems not to give a damn about our opinions/needs regarding CA! :thumbsdow

pooby
11-20-2007, 04:58 AM
To give you an example, I spent about two months trying to make an, admittedly quite complex, rig for a goblin character in LW. On top of that I was using techniques that I'd spent years desingning to get around LW's lack of tools.
It worked, and I was quite pleased with it, but still, I couldn't get it to deform the model as well as I'd like. In the end, Another job came up and I dropped the project.
Recently I found the model again and decided to have a go with it in XSI. Its been absolutely straightforward to make a rig in a few days that's incredibly more powerful than the LW one. I haven't had to do ANY workarounds or anything awkward, and It deforms perfectly. ( Not to mention that I can just drop mo-cap onto it )
LW in it's current incarnation isn't just bad for rigging. It's atrocious. I didn't realise how bad until I learnt something else.
Anyway, I'm sure Newtek are acutely aware of how bad it is, so I doubt that much of what makes up the animation toolset will still be in LW in 5 years time or maybe even sooner.

Dirk
11-20-2007, 05:20 AM
What I would like to know is in how far spezialized tools like a Walkcycle Generator or better dynamic parenting (example: wrestling characters) are there in other applications and to what extend they are used.

Elmar Moelzer
11-20-2007, 06:49 AM
For me the absolute dealbreaker is that the character animation specific tools from Modeler are not present in Layout.
It would be really nice to be able to adjust bones like we do with Skelegons in Modeler, directly in Layout, without the need for recreation of the entire hirarchy etc. Also it is IMHO a necesseity to have some modeling capabilities in Layout in order to make changes to the object for fixing deformations.
E.g. I want to be able to model a morphmap on the already deformed object with the morphmap only containing the changes as they would appear on the BASE- Model. Getting good deformations on character- models is very difficult otherwise, IMHO.
For me the hardest part of the rigging process is making the character deform right. Right now there are very limited tools for this. Having Modeling tools and the ability to create endomorphs directly on the bone deformed mesh, would be a big improvement. Also it would be very cool if we could finally adjust weightmaps directly in Layout and see those updated on the bone deformed mesh (even though weightmaps can be a pita to use and I always try to avoid using them wherever possible).
A better dopesheet would be very welcome. Having that integrated with a more solid Non Linear Animation system would be most welcome.
CU
Elmar

Sil3
11-20-2007, 07:59 AM
Having Modeling tools and the ability to create endomorphs directly on the bone deformed mesh, would be a big improvement. Also it would be very cool if we could finally adjust weightmaps directly in Layout and see those updated on the bone deformed mesh (even though weightmaps can be a pita to use and I always try to avoid using them wherever possible).
A better dopesheet would be very welcome. Having that integrated with a more solid Non Linear Animation system would be most welcome.
CU
Elmar

To be able to use Shapes (Morphs) to really correct deformations and let them be what we initially correct modeled, LW needs to apply those Morphs on TOP of all other Deformers, or else things will be same they are now, Morphs first Bones on Top, things never look like we modeled them...

Sorry for bringing XSi in here, but the Secondary Shape Modeling in XSi exists exactly to do this, it is powerfull, since we can Layer Deformers on Top of each other without screwing the Deformations.

I can have a Character deformed by Bones and Morphs, later on Corrective Morphs on top of it and still have room for other Deformers to make things like Elastic Girl kinda setups, try to do something like this in LW... I dont know if it is impossible, but easy and straigh to do in a timely manner im sure it is not. Of course that for all this to work, Bones cannot be the only Deformer items.

Weightmaps are a PITA when the way to make and adjust them is splited like in LW, if everything was under the same roof and not glued together without the slightiest interaction, yes they became not a PITA but a F****** Nightmare... no wonder majority of people run away from them in LW

Verlon
11-20-2007, 08:33 AM
1. Streamline the process. Its like learning three, four, or five different applications (Layout, Modeler, rigging, morph mixer, expressions, IK booster, etc...) . The learning curve is so steep you need a helicopter. As others have said make it possible to do some bone surgery in Layout if not integrating the two applications (and I don't think integration is needed). Why do we need "convert skelegons to bones? Do we EVER make anything out of skelegons besides bones?

2. faster interaction. If it takes too long to redraw the screen, you really can't do anything. This should probably be #1.

3. Something to GREATLY simplify walk cycle generation. If Poser can do it, Lightwave can surely do it. Character studio's "just lay out the foot prints" would be cool (I think...haven't played with it much really), but I actually like the Poser walk designer. It is sort of like morph mixer, but with walk cycles (add a little 'brisk' and some hip sway...maybe lighten up on the stiffness....now go edit your keyframes). Double props if it allows expansion for quadrupeds and other more exotic rigging.

Chris S. (Fez)
11-20-2007, 10:52 AM
I haven't touched CA in Lightwave in years. Here is what I need from Newtek to consider going back to Lightwave for CA:

1) Fastest deforming subds in the solar system.
2) Smartskin-Deformations in Layout.
3) Stable, flexible, intuitive rigging (canned bendy bones and stretchy-IK please!)

Give me those three, along with robust scene-management/render passes that are Fprime compatible, and I will switch back immediately.

eyelandarts
11-20-2007, 11:02 AM
non linear tweening of morphs.

pooby
11-20-2007, 11:38 AM
non linear tweening of morphs.

how would that work? You can do it now with midpoint morphs, but how would you describe the non-liner path each vertex would have to make?

jin choung
11-20-2007, 11:53 AM
how would that work? You can do it now with midpoint morphs, but how would you describe the non-liner path each vertex would have to make?

well we don't have MTSE for endomorphs (YET!!!!) so that would be my way....

jin

eyelandarts
11-20-2007, 12:45 PM
Well if there where MTSE for endomorphs then that would go a long way to making it easier but it would still require several hand made morphs to approximate the resolution of the curve. Imagine you are using morphs to create the movement of a swing starting from its highest position until it reaches the other side (180) in one smooth movement. Even if you made two morphs (one in the middle, parrallel to the swing armature and another at the other side) the tweening between those points in the morphs would sill be linear and your swing would be terribly deformed. Ok not too bad its just a swing and we can add more hand made inbetweens but now imagine you are making a character speak and you want the muscles to move in arches (as it should). Are you going to create a bunch of in betweens for every phenome? and what if you are going from one extreme facial expresion to another, are you going to make custom tweens for each time that happens? It would be much easier if the tweening was handled internally as does LW in the Graph editor using TCB.

I believe smartmorph does this though I havent tried it ( http://smorph.polas.net/smorph/ ) . It looks like an excellent advance morphing package but LW should still have some of this functionality included.

pooby
11-20-2007, 02:17 PM
You can do pretty much everything Smartmorph could do in the displacement nodal editor. Plus you can do tons more. Just requires more manual input.

bobakabob
11-20-2007, 02:52 PM
I'm hoping for an XSI style editable generic biped / quadroped armature you can autorig with IK options with the minimum of hassle. It should be accessible enough for artists who want to animate to a tight deadline and customiseable enough to satisfy specialised rigging engineers. Imo XSI have the balance right. Surely Newtek can do this too.

Messiah style 'smart bones' would be a dream - In Messiah they are wonderfully fast and intuitive to work with, though weight maps are presently very useful and flexible in LW.

Greater mocap flexibility. BVH is presently the only file format LW can handle.

Some kind of Max / Biped style editable walk cycle generator would be a bonus for deadline doom and appeal to newbies. Even Poser (cough) has a clever if basic solution.

Finally remember the days Newtek pulled some dazzling innovation out of the hat and rocked the competition? Even Daz Studio (ahem) has developed a unique CA tool: an 'instant mocap' solution operated by the animator's mouse. Sure it's primitive but a marvel of lateral thinking and capable of cool spontaneous results.

After all these years rigging characters from scratch in LW is the only out of the box solution to starting CA - a daunting prospect to newbies despite the wealth of tutorials out there. Yes there are some great autorigging plugins but this shouldn't be the case in 2007.

Mipmap
11-20-2007, 03:49 PM
I'd like the process to be a lot more simple in regard to the actual setup process.

We model the things in Modeler, so let us continue from there and to the skelegon thing in Modeler, set our weight maps and smart skins up in there, with a tool that lets us bend the bones and see deformation so we can repair it. From there allow us to export our rigs to a separate file, or save the character with skeleton and all so it is one complete unit. Oh, and also allow us a function to easily mirror settings on the opposite bone if we wish to. This way you do smart skin repairs quickly on one side of the body, and it will do them for the opposite bones, for the polygons on that side of the body.

Also allow skeletons to be loaded and easily resized to fit new characters just as you can skelegons by clicking the point\joint and moving it. This way a good skeleton can be reused and refit to new characters easily.

That way all the setup would be done in one program in a simple productive way. By the time the model was taken to Layout we'd have already been able to verify that all the parts move and deform correctly.

In regards to animation:

Increase the point\click accuracy so we don't click the wrong things. Give us those overlayed dots on joints to easily and quickly click and move joints to animate. Supply a biped model with the program that is production quality. For reference, just buy a copy of Hash.

Cageman
11-20-2007, 03:54 PM
From ontop of my head:

Deformers, Constraints and Set Driven Key are the most fundamental tools I use in rigging. None of them work in an intuitive, easy way. Set Driven Key isn't even working the way it should work (bad design). The lack of connectivity in LW is also a limiting factor for technical rigs. Connection Editor in Maya combined with Hypershade is a truly powerful system. Something like that in LW would be really cool stuff. I experimented with a rig that used 15-20 connections with Relativity and it went pretty heavy. Compared to a rig I did in Maya that has hundreds of connections and still flies (even with the high-res object visible). Relativity is nifty and easy to use, but very slow.

eyelandarts
11-20-2007, 04:21 PM
You can do pretty much everything Smartmorph could do in the displacement nodal editor. Plus you can do tons more. Just requires more manual input.

eye yes but then again manual input IS the problem when it comes to this.

3D|Dave
11-20-2007, 06:18 PM
Faster OpenGL is a MUST! A simple 50K poly model with bones is too slow. in XSI the same model is realtime with or without bones.
Non destructive setup, neutral pose.
Parent in place method changed to match translation/rotation (yes like XSI).
Bone creation and editing like XSI.

Start there and we will talk later.

Verlon
11-20-2007, 06:26 PM
yeah, I think the FIRST priority has to be faster response.

What good do ANY of the other changes do if it takes too long for the screen to update?

pooby
11-21-2007, 06:07 AM
yeah, I think the FIRST priority has to be faster response.

What good do ANY of the other changes do if it takes too long for the screen to update?


With LW, its often bone deformation or other internal calcualtion that slows it down rather than just open gl drawing..
A new animation system should fix that.
LW style bones are always going to be slower than XSI or Maya, as each one has a falloff warping zone around it whereas in XSI the vertices are just bound to the 'bones' making it faster. there is no deforming ability in the bones themselves. In fact you don;'t need bones at all, you could use nulls if you wanted.

dballesg
11-21-2007, 06:18 AM
You can do pretty much everything Smartmorph could do in the displacement nodal editor. Plus you can do tons more. Just requires more manual input.


Hi,

To not hijack this thread, Pooby can you explain how you will do non linear morphs in the displacement nodal editor on another thread?

Thanks.
David

pooby
11-21-2007, 09:53 AM
You don't need the displacement editor for that, just use Cycler in the channel modifier of the morph to set it up and link a few morphs to a nulls x channel or something.

beverins
11-21-2007, 10:19 AM
1) fix the SDK hooks so that people can make plugins to make better solutions than you could think up yourself
2) redefine your outlook towards plugin developers. i.e. they should never feel as if they are ignored.
3) fix your data transport systems - FBX, BVH, OBJ, MDD - so that they talk seamlessly with other programs. The oeprative word being SEAMLESSLY. I want to be able to move a scene file of any complexity back and forth from any other program (i.e. Messiah, Motionbuilder, zbrush, maya, xsi, 3dsmax, etc) without any hassle whatsoever. Please note, the usual wording is "minimum of fuss", but I recommend reaching farther than that and going to "zero hassle", "invisible", "seamless", "smooth as butter" - you get the idea.

Yeah, Id like to see better CA tools inside LW, but the fact of the matter is that you guys have too much to fix to get that going now. Eventually, you will have a CA system in LW that not just catches up, but is once again a leader - I have confidence in you guys to do that - but in order for you to accomplish that, the program should be fixed NOW to allow for this sort of smooth interaction with other software.

I shouldn't have to beg on a forum like this for someone to send me FBX version 6, then translate using Motionbuilder's FBX translator. I shouldn't have to use vertex baking via a program that has never exited beta (point oven).

If you guys can't do CA right this minute, that I understand. So in the meantime, work on the tools that allow seamless transitioning.

And if this means that you have to go to the doors of companies like Autodesk, Luxology and Avid and face their dismissive stances, then so be it. Suck it up and go ask to play nice. Bad blood is secondary to business. Like it or not, people DO use other programs. Maybe you should have an .LXO exporter for LW scene files, for example. Things like that. Hastings and Ferguson might have been the ones that left you in the lurch of having a program that you have been trying to fix and unravel since version 7.5, but I think its high time to work with them again and get things moving. Autodesk says that Newtek are the ones that have to make a working FBX translator, being that the SDK is free.

How about working with the people who make Blender? I mean, the damn thing is free. One would think that having a rock-solid import/export for Blender would be a tremendous benefit. Have it communicate directly with Layout much in the same way Vue and Messiah do. WORK WITH THESE PEOPLE. You know, and I'm sure that some money thrown their way wouldn't make them less likely to help.

If you can't reinvent the wheel, make it easier for the users to go use the wheels that do work, and then return to use your superior rendering and surfacing (IMHO)

beverins
11-21-2007, 10:35 AM
OK, you know what, Newtek?

I will make this real easy for you.

See this guy? http://physbam.stanford.edu/~fedkiw/

1) go visit him
2) ask him to name his price for making what you see on his page NATIVE to Lightwave, as well as EASY (well, as easy as you can make it) to use.
3) Pay his price, and allow him to make his magic.

Seems you get
1) Quasistatic skinning deformations
2) superior CFD based muscle deformations
3) possible breakthroughs in rigging

plus you get all those CFD dynamics which NO other software yet has

DO IT NOW! Sometimes you have to mortgage the house, the car, the wife and the children to make a positive change. I can guarantee you that if you spend money to get people like this (and yes, he will be very expensive) Lightwave will not only be winning Emmies, but it will be winning OSCARS. Trust me on that one.

beverins
11-21-2007, 10:46 AM
Oh, and on the "working with Blender" thing.

Collada is a step forward, and definetely should be undertaken... but I'm talking about a seamless instant update between Lightwave and Blender much like Xstream or Messiah does, or your own Hub program.

and yes, *YOU* SHOULD DO IT. Not plugin developers, not the people who work on Blender.

YOU. NEWTEK. MAKE IT HAPPEN.

jin choung
11-21-2007, 11:32 AM
re: physbam...

oh please... be somewhere in the realm of remotely realistic. ILM and PIXAR are using this guy for all of their high end fluid effects (poseidon, harry potter, pirates of the carribean 3, ratattouille). conversely, little shops do not and cannot use it.

can you imagine the price discrepancies that make it so?

it's not in maya, xsi, max or houdini. you think lw can afford to put it in?!

this is not going to happen.

jin

3D|Dave
11-21-2007, 11:45 AM
The new 9.3.1 has performance increases for bone deformations vs 9.3 so thats a movement in the right direction!


Thanks Pooby, you hit the nail right on the head about LW bones and deformations.

Ztreem
11-21-2007, 11:51 AM
Beverins: As far as I know, NT is allready working on FBX and Collada support. That's what NT had said here at the forum, maybe if you do a search you can find the thread.

For animation, not only CA but all animation. All motion modifiers have to see each other and play together even motions of a child that only follow the parents motion must show up in the graph editor.

eyelandarts
11-21-2007, 11:57 AM
Hmmm, one more thing. Really good lattice deformers that work with all other lightwave stuff (targeting,parenting,expressions,etc)

hrgiger
11-21-2007, 12:43 PM
how would that work? You can do it now with midpoint morphs, but how would you describe the non-liner path each vertex would have to make?

I've always desired endomorphs to work off of a slider in modeler. You would create the endomoprh as usual, but then you would adjust the points at different spots along the slider. In a typical Lightwave endomorph as we have today, all points are adjusted at 100% morph. But then you could slider the slider down to 50% and create the midpoint. Now when you scrub the slider, it interpolates the points between the adjustments you have made along a typical Lightwave curve, much like keys are tweened in Layout.

hrgiger
11-21-2007, 12:52 PM
Some of the dealbreakers for me...

1. No new deformation tools. We need a way to adjust our meshes on the fly during the animating process that does not involve modeler in any way shape or form. Whether it's latices, spline deformers or manual adjustment of points that are all able to be tied to the bending either through an expression or bone property.

2. Slow interaction. Multiple character scenes that have characters with many bones and expressions/motion modifiers shouldn't see a severe slowdown when it comes to posing the characters.

3. Adding news tools or fixing old ones that still don't work with other existing tools. I say if it doesn't talk to the other one, throw it out until it does.

ericsmith
11-21-2007, 01:13 PM
I've always desired endomorphs to work off of a slider in modeler. You would create the endomoprh as usual, but then you would adjust the points at different spots along the slider. In a typical Lightwave endomorph as we have today, all points are adjusted at 100% morph. But then you could slider the slider down to 50% and create the midpoint. Now when you scrub the slider, it interpolates the points between the adjustments you have made along a typical Lightwave curve, much like keys are tweened in Layout.

That is a really cool idea. It would require a change to the .lwo format to accomodate it, but it would be worth it in my opinion.

Although as I think about it, I wonder if it could work if under the hood each saved "keyframe" morph state was a new vmap. They'd just have to figure out a way to save the curve data in the object, or perhaps you'd have to export it as a seperate file. I could live with that.

Eric

bobakabob
11-21-2007, 03:30 PM
Some of the dealbreakers for me...

1. No new deformation tools. We need a way to adjust our meshes on the fly during the animating process that does not involve modeler in any way shape or form. Whether it's latices, spline deformers or manual adjustment of points that are all able to be tied to the bending either through an expression or bone property.

2. Slow interaction. Multiple character scenes that have characters with many bones and expressions/motion modifiers shouldn't see a severe slowdown when it comes to posing the characters.

3. Adding news tools or fixing old ones that still don't work with other existing tools. I say if it doesn't talk to the other one, throw it out until it does.

Agreed... Character animators have been crying out for the ability to edit / animate / manipulate points directly in Layout for years.

Iaian7
11-21-2007, 04:00 PM
That is a really cool idea. It would require a change to the .lwo format to accomodate it, but it would be worth it in my opinion.

Although as I think about it, I wonder if it could work if under the hood each saved "keyframe" morph state was a new vmap. They'd just have to figure out a way to save the curve data in the object, or perhaps you'd have to export it as a seperate file. I could live with that.

Eric

That'd be great! And as you say, little to no change to the .lwo format, simply a change in how the morphs are interpreted by the Morph Mixer (granted, updating how they are edited in Modeler would be very, very helpful, especially in previewing and fixing in-between issues when blending from one morph to another). Implemented via, say, a serial number appended to morphs of the same name, much like an image sequence. Blink1, Blink2, Blink3, and viola... you'd have to ability to approximate curved morphs by setting multiple keyframes for one slider!

Not to mention, there are many, many times when I needed a series of morphs. Being able to "group" morphs into a single motion, such as a leaf unfolding as a plant grows, could be invaluable. You can, in a way, blend morphs by using Gradients in the nodal displacement editor (attaching a different morph to each gradient key), but it's hardly easy, and certainly not as effective as a dedicated morph map sequence setup.

Of course... this would be even better as a morph mixer that operates in more than one dimension... much like the 2D palette-based animation controls posted to the forums a couple weeks ago, being able to mix to and from multiple morphs would be amazing. Of course, that'd require quite a bit more than just morph map sequences... so maybe not that feasible. :D It's late, I'll shut up now.

jin choung
11-21-2007, 05:35 PM
That is a really cool idea. It would require a change to the .lwo format to accomodate it, but it would be worth it in my opinion.

Although as I think about it, I wonder if it could work if under the hood each saved "keyframe" morph state was a new vmap. They'd just have to figure out a way to save the curve data in the object, or perhaps you'd have to export it as a seperate file. I could live with that.

Eric


MTSE... multiple target single envelope - old idea but this is a really elegant workflow for implementing it into modeler! (not to mention we don't have mtse for endomorphs yet... this would be a great way to add it)

make it so that if you intervene during the slide from 0 - 100, you end up creating a another morph target at the position the slider is at...

jin

hrgiger
11-21-2007, 06:09 PM
Yes, a single endomorph with multiple targets could be very powerful. Sure it could do simple non-linear shapes like the closing of a lid over a round eyeball or something simple like that but you could do whole body motions and with the ability to hold the shape you desire during the whole process of the morph.

jin choung
11-21-2007, 08:29 PM
man, i'm kind of concerned that this idea won't be adopted now... it's SUCH an elegant solution in creating MTSEs with endomorphs. and while you're playing with a slider so you get immediate feedback and can monitor how your inbetween shapes are affecting your "sequence".

you should email them hr. cuz it's a truly beautiful solution you propose.

jin

wacom
11-21-2007, 09:10 PM
Talk about flogging a dead horse. This subject should only be rehashed as a reminder to NewTek that this "little" part of LW is broken (well or not really there) and to silence the odd ball IKbooster freak.

That being said...

Oh man- I don't want to even waste my time on this- it just sucks and needs to be fixed, and there are WAY to many good examples out there of solutions to even go into it.

Maybe we should just put up links to the demo downloads of other applications for Newtek to review HOW some people do it and often do it WAY better?

OK...two deal breakers for me...

The rigging tools in general require voodoo not found in other applications. Call the witch doctor! I really can't say it more simply than that.

And...The lack of a history stack/operator stack/link engine (like houdini for instance), kind of suck it up in this day in age. I likes to makes me own tools see- not have NewTek MAKE me use them a certain way...

I'll stop there- those alone are very powerful things to add/change/make/correct.

If I sound cynical at this point it's because I have another 3D app. open that I'm animating in...and man...I'd be impressed if NewTek could make 10 even 1/4 good to animate in...because that would be 10x as good as what scraps are clanking around right now...

I don't know if morphs are the real solution to all our fears. Node it up baby! That's what I say at least...

Right now it's like the modeler and nodal/new rendered engine are in love on two desert islands...but there is this huge body of water between them...called "animation" with a giant sea monster called the "HUB" devouring any coherent idea that might finally unify the two and produce the lovely offspring we call "The 21st century, 3D Animation tools set".

IKbooster is the equivalent of trying to go toob'n in these waters...bring several six packs and swim like hell!

hrgiger
11-21-2007, 09:58 PM
man, i'm kind of concerned that this idea won't be adopted now... it's SUCH an elegant solution in creating MTSEs with endomorphs. and while you're playing with a slider so you get immediate feedback and can monitor how your inbetween shapes are affecting your "sequence".

you should email them hr. cuz it's a truly beautiful solution you propose.

jin

I sent an email to LW-features with the proposal.

Stooch
11-24-2007, 12:59 PM
1) Everything needs to talk to each other. SCripting must be fully comprehensive and affect EVERYTHING, same goes for SDK. This is the most important thing you can do for lightwave PERIOD, not just CA.

2) Scene referencing. Again this is such an important feature that it goes way beyond CA but is critical for it as well.

3) Dynamics. I want this to be fast and solid. Again this isnt just CA but the entire physics solver needs to be rethought, unified and blazingly fast. Hair and Cloth should be driven by this. Again this affects far more than just CA but is also critical. Who wants a bald, naked character in 2008? Not me.


Notice how my requests arent just targeting CA alone, i feel that aiming only at CA right now is a fatal mistake for LW. you guys are too far behind to be taken serious as a CA tool even if you do make it work. You need to play to your strengths.

Stooch
11-24-2007, 01:13 PM
I didnt mention interaction speed because i believe NT already understands this and are making strides in this area.

Cageman
11-24-2007, 02:10 PM
Notice how my requests arent just targeting CA alone, i feel that aiming only at CA right now is a fatal mistake for LW. you guys are too far behind to be taken serious as a CA tool even if you do make it work. You need to play to your strengths.

I think we have just witnessed ONE of the strengths getting a major update; Rendering. I think/hope NT will continue to refine and add things to the renderer such as a GI cache that works with network rendering, new arealights, softshadows for all lights.

Ok... on topic again:

The whole motionmodifier thing should become one fat Nodal-interface. Anyone who has played with Item Motion Node (a motionmodifier using a node-interface) knows what I'm talking about. Pooby has done some great examples of what can be done using those tools (thirdparty). Having it integrated could really change the way LW works for the user, because the thirdparty tools that does this already makes a huge difference, but have their set of limitations.

I find it interresting to see your comment about "Play to your strengths"... what do you mean by that? In my work, I've noticed some really nice timesavers using LW compared to Maya, mostly they are related to how vertex maps works in LW compared to Maya, and the "no need for weights mentality" for bones in Layout. Should these things be kept or removed/changed? Can they be kept when things will change? Why is it that in Maya you need to rebind a character when adding a bone/joint, when you don't need to do that in LW? My personal oppinion on this is; change whaterver you want to the better, but leave vertex maps as is, and also leave the bones "no need for weights" as is as well...

hrgiger
11-24-2007, 02:52 PM
1)
Notice how my requests arent just targeting CA alone, i feel that aiming only at CA right now is a fatal mistake for LW.

It has been stated numerous times by Newtek that different areas of LW are being worked on simultaneously so I seriously doubt that a focus on CA is all that would be going on for an upcoming release.

jin choung
11-24-2007, 02:55 PM
you don't need to rebind in maya 2008. also, in lw, you do have to rest bones again... same thing, just that lw's bones are embedded in a mesh and is a "subobject" of an object rather than independent objects on the same hierarchy level as objects, cameras, lights, etc....

all things considered, i'd prefer bones to be independent and not nested objects.

"no need for weights" is fine to keep on board... unless that's what's causing some skinned animation to be sooooooo slow. if it is, then keep it but only have the detrimental speed effects ONLY affect such non weighted skinning.

i would also like LIMITED REGION bones be better... regions can be visible in all views (min and max), there are options on how overlapping regions interact, the falloff curves between min and max is adjustable, and once regions are nicely setup, the equivalent weights can be BAKED INTO WEIGHTMAPS... would be a great way to start the weighting process. blender has this feature.

as for nodes, it would be nice if nodes can be the backbone of the entirety of lw. such that not only is every part of lw talking to every other part and accessible through scripting as stooch specifies but also exist as nodes and can be wired up to create script like behaviors. in maya, it is faster to manipulate relationships through hard coded nodes than expressions.

but i do digress here.

again, i'm going into all the things that would be nice for lw to have.... but this is not bare minimum, deal breaker for me. now we're aiming for the moon.

jin

jin choung
11-24-2007, 03:02 PM
re: playing to one's strengths

yah, i do think this is why it is unfair to criticize newtek for working on its renderer so much this last round.

they are doing just that, playing to their strengths... playing to where they can make the biggest impact in market and make the most money.

absolutely right, no matter how much time they invest in CA, it's not likely to be more time than xsi and alias are spending... and it's not likely to supplant either as the industry leaders in CA.

taken to the extreme then, newtek could just convert lw into a renderer.

but hopefully, they're not taking THAT route and they are aiming to have a viable all-around solution that can start gaining mindshare again.

besides, it's a crowded marketplace for renderers now - there's no lack of third party renderers at attractive price points including free.

so all around package, while getting crowded, is still the smaller pond to be in.

jin

Stooch
11-24-2007, 09:11 PM
It has been stated numerous times by Newtek that different areas of LW are being worked on simultaneously so I seriously doubt that a focus on CA is all that would be going on for an upcoming release.

yes i am fully aware of that, newtek has also stated many times that they are working on CA so why are you guys still bitching about it? hmmm???

I for one dont want to see too much development time to be wasted on CA.

Stooch
11-24-2007, 09:15 PM
I find it interresting to see your comment about "Play to your strengths"... what do you mean by that?

jesus have you been reading my posts?

i mean exactly what I stated.

CA is NOT a strength for LW and probably never will be, read the rest of my points about SDK, that should open a door for a CA system from a third party. Not sure how much clearer i can get.

p.s. while using ligthwave and maya at work i found maya to be one huge time saver in comparison to LW. Sure there are times when i save a bit of time with LW over maya, but that quickly becomes a fart in the wind.

Stooch
11-24-2007, 09:27 PM
Sure there are times when i save a bit of time with LW over maya, but that quickly becomes a fart in the wind.

this refers to CA of course.

By the way newtek, loving your dumbass post edit timelimit. i see you like to keep your forums as flexible as LW. keep it up! you are winning over many hearts and minds for sure!

i guess this goes hand in hand with your philosophy on non destructive workflow right?

(look at that, had to make 3 posts when could have just edited one.)

Cageman
11-24-2007, 11:57 PM
you don't need to rebind in maya 2008. also, in lw, you do have to rest bones again... same thing, just that lw's bones are embedded in a mesh and is a "subobject" of an object rather than independent objects on the same hierarchy level as objects, cameras, lights, etc....

all things considered, i'd prefer bones to be independent and not nested objects.

"no need for weights" is fine to keep on board... unless that's what's causing some skinned animation to be sooooooo slow. if it is, then keep it but only have the detrimental speed effects ONLY affect such non weighted skinning.

i would also like LIMITED REGION bones be better... regions can be visible in all views (min and max), there are options on how overlapping regions interact, the falloff curves between min and max is adjustable, and once regions are nicely setup, the equivalent weights can be BAKED INTO WEIGHTMAPS... would be a great way to start the weighting process. blender has this feature.

Ok.. nice to see Maya 2008 has this feature... we may move to that version in a year or so. We just recently moved from Maya 7.01 to Maya 8.5.

Autocreation of weights is avaliable in LW as well (could be way smoother though), but the way you describe it sounds way more userfriendly and flexible, and would go hand in hand with a merged Layout/Modeler. On the theme of bones, I would like to see a more refined way of saving rigs and bonesinformation to file (just about every bit of information a bone has). Loading a rig onto a new character should become more streamlined and if following a rule of namingconventions both on weights and bones, the loaded rig should be able to find its weights without any manual labor.

Animationlayers would be very usefull as well. I miss them in Maya/LW after using Motionbuilder for some time. Master Channels can, to some extent, do what I want, but that way of working is far from intuitive and leaves many things to be desired.

Cageman
11-25-2007, 12:01 AM
jesus have you been reading my posts?

i mean exactly what I stated.

CA is NOT a strength for LW and probably never will be, read the rest of my points about SDK, that should open a door for a CA system from a third party. Not sure how much clearer i can get.

Ok... I was kind of confused if you wanted NT to focus on both things or if they should leave the CA to third party through a more open app.

jin choung
11-25-2007, 12:09 AM
Autocreation of weights is avaliable in LW as well (could be way smoother though), but the way you describe it sounds way more userfriendly and flexible, and would go hand in hand with a merged Layout/Modeler.

right, the current method basically takes the ^8, ^16, ^32 etc. infinite falloff influences and bakes them to weight maps.

but yah, it would be nice to have the gui of the "forcefield hot dogs" be viable just to use on their own (right now, you can only see them in certain views and/or redraw methods) and to make them bakeable into weight maps.

actually, user defined "hot dogs" would be nice too... let the user draw out an arbitrary forcefield shape and set that as a bone's hot dog.

(alright, while i'm making my christmas wishlist - you have hotdog max and hotdog min boundaries... then you can specify the falloff curve between min and max... AND - when you have two bone hot dog influences intersect, you can specify a weight between 1 and zero governing which hotdog gets more influence (that value between 1 and 0 being multiplied to the min/max falloff somehow... of course).

now THAT would be hot. with THAT you might not need to use weightmaps. and with that you could bake it all down to weightmaps if you needed it for export to game engine or other app.

oh well, yeah, this is all my pie in the sky wishes... but not my deal breakers.

i still only have 3 deal breakers. lots of wishes. 3 deal breakers.

jin

jin choung
11-25-2007, 12:18 AM
re: maya 2008 - yeah, we're in the middle of a project and while it would be nice to upgrade, and they seem to have bug fixes and stuff that we need right now, it's too risky.

also, it's too motherfing expensive! if you're ilm or something, the cost of a seat may be negligible. alas, that isn't the case for most folk. upgrade pricing is highway robbery.

and for people with limited funds, lw wins the "cost of ownership" battle every time. i've said it before, i'll say it again, lw is the only affordable 3d app (well blender and animation master are affordable as are other lesser apps but they're not in the same league imo).

jin

Cageman
11-25-2007, 01:11 AM
On the theme of bones, I would like to see a more refined way of saving rigs and bonesinformation to file (just about every bit of information a bone has).

I may have to further explain my point with this. Lets say we don't get a working solution for scene-referencing ala Maya, we will have the issue of re-importing things from animationscenens and...well alot of manual labor. Old news, eh?

So, if Item Propertis had this little button that when pressed, a UI pops up letting me choose what information to save, load or referece. So, lets say I animate a character in Scene A. When I finish, I simply export all the information to a file. I then create Scene B, C and D, all having the same character loaded without bones, just the modelfile. The difference here is that it's referencing or loading the rest of the information from the file I exported in Scene A. If/when I change the animation in Scene A (and save the information), Scene B, C and D will automaticly get updated on a scenereload. In Scene B, C and D I should be able to determine what I want to load from the informationfile. Lets say it loads all information by default, but I'm only interrested in the rig+animation, I should be able to reference just that.

cresshead
11-25-2007, 02:20 AM
character animation>> deal breakers

the main deal breakers are the following for most people:-
1.being crap at character animation
2.lack of good training videos on 'how to animate'
3.time you spend on a shot, trying to get it looking good.

everything else is negotiable...with the base app such as lightwave or plugins or using another app to rig/animate and then bring in or comp the result.

jin choung
11-25-2007, 03:01 AM
actually, i think the heart of what we're trying to get at here is what is the bare minimum for you such that your number 1 point is no longer true.

jin