Splines - using bias

[Caroline]

When I model, I never use the bias - I always fill out with splines. I am sure that I should be using the bias handles, but I always seem to get creases instead of the nice curves I am looking for.

 

Do "real" modellers always use bias to get definition? What's the secret to stopping that little handle from slipping out of control all the time?

 

Is the secret to create the mesh as far as possible and only then tweak?

[steve392]

Hi Caroline Im far from a good modeler but I allways model best as I can then use the handels if I have to ,but spearingly

[Rodney]

Caroline,

I'll let others with more experience address the aspects of more detailed models but here is a little background to fill out the subject. My models thus far haven't been very detailed so the issues you mention don't effect me as much.

 

While its long been available, adjusting Bias on splines is a fairly recent practice.

The reason for this is that prior to v11 adjusting Bias would often lead to unnecessary popping when the adjusted splines were animated. Hash Inc worked out these issues and the popping is a thing of the past.

 

Many old time users will still shy away from Bias manipulation for the reason I just mentioned. Many may not know or be able to fully accept that they can and in many cases should tweak Biases.

 

but I always seem to get creases instead of the nice curves I am looking for

 

I might be reading your words wrong here but do you mean to say that anytime you tweak Bias handles you get creases? Can you post example meshes?

 

You shouldn't be seeing many creases unless you create them. (Peaking Bias control handles etc)

 

Edit: It occurs that you may be talking about decals. If that is the case then adjusting Bias after the decal has been applied can definitely cause artifacting/creasing.

[itsjustme]

I'm not what I would consider a "real" modeler, but I've done a modest amount of it.

 

For organic models, I use very little bias tweaking...and only in areas that don't move or move very little. The reason is that the tweaked bias will maintain its' position as it is moved. It won't pop, but it also won't move like an untweaked bias. Sometimes that isn't a problem, but in a lot of areas it is. If the model is purely for a still image, then the tweaking won't matter. For an average human character, I tweak the bias in the eye (there's a link to my eye making tutorial in my signature), teeth, ear, sometimes the top of the skull, finger nails, tips of the fingers and I generally reduce the percentages along the edge of the nose to get a slight crease.

 

For mechanical models, I tweak the bias quite a bit...also on parts that don't need to bend, but that (bending parts) is more rare in mechanical models.

 

Hope that helps, Caroline.

[Caroline]

Lol, Steve, your female is what I am trying to achieve.

 

I did not realise that biases had a problem in the past. So that is probably why we do not hear so much about them.

 

By creases I don't mean the sort you get from wrong splinage, just an uncomfortable not-quite curve. Like I was just doing a rib cage and the handles fly off every which way. So I just wondered if that was what everybody gets

 

I shall try Steve's way and try and patiently get the mesh right before I tweak. I just want the "I've had enough of this model - finish it now" button

[Rodney]

Caroline,

When you get bored and have a lot of time to kill (yeah right!) my suggestion would be to embrace your fears.

 

By this I mean... set out to tweak a model's Bias drastically... intentionally.

Hit the P key to Peak and adjust each side of the Bias handle. Whack it back to curved via the O key. Turn the model around and view it from all sides. Just jump in there and tweak, tweak, tweak.

 

You'll either never want to never tweak Biases again... or you'll have established a better workflow with Biases.

 

Where I've found Bias adjustment most useful is before Lathing or Extruding. A:M then does the calcuations for you and generates smooth or beveled curves (depending on the Bias level) that apparently we can't do too well as humans. Make A:M work for you.

 

The reason is that the tweaked bias will maintain its' position as it is moved. It won't pop, but it also won't move like an untweaked bias. Sometimes that isn't a problem, but in a lot of areas it is.

 

David,

I've never heard it put that way but that makes a lot of sense to me. Thanks for that. I feel the need to investigate.

[robcat2075]

There may be a way to get smooth without bias. I've been impressed with Will Sutton's models that seem to have no bias adjustments at all.

 

But when I'm using other people's models bias adjustment seems to be the only solution.

 

 

The reason is that the tweaked bias will maintain its' position as it is moved. It won't pop, but it also won't move like an untweaked bias.

 

Can you show an example that does this consistently? I've seen this sometimes but it went away after saving and reopening the model.

[itsjustme]

The reason is that the tweaked bias will maintain its' position as it is moved. It won't pop, but it also won't move like an untweaked bias.

 

Can you show an example that does this consistently? I've seen this sometimes but it went away after saving and reopening the model.

 

Here is an example Project and a short video clip of what happens when the CP's are moved around.

bias_example_project.prj

bias_example.mov

[Caroline]

The reason is that the tweaked bias will maintain its' position as it is moved

 

Yes, as demonstrated in your movie, that's what I've been getting, and don't like the unnatural movements.

 

Aha, Rodney - I have never moved the bias of a peaked CP, nice that they go separately. You are right - I should "just do it" . I suspect, as David says, that things like biased muscles won't animate as well, although I guess the bias could be changed in a smartskin movement, if it is important.

 

I shall go and pull apart some more models.....

[Rodney]

David,

I'm looking at your movie and for the life of me I can't think of how that movement is a bad thing. If you move the CPs you want the splines to keep their shape... right? Aren't the movements more natural if/when the Bias don't automatically adjust?

 

I'm missing something simple so bear with me.

I've only looked at the movie. I'll look at the project file.

[Kamikaze]

I believe I read that one can use bias adj. in a smartskin, may help with muscle definition in animations, like animals? I may have to give it a try!, Im using a lot of smartskin on my deer model but havent attempted any bias adj. yet.

[Rodney]

Not entirely sure why I'm posting this here. It does have something to do with Bias tweaking... but not animated Biases.

 

I started tweaking and this environmental thingy was the result.

The cool blue stream was surely my subconsciousness trying to balance the ridiculousness of today's politics. Will November never come?

 

If you look in the lower right window you'll see no grid was injured in the process of tweaking these Biases. All tweaks were made in an Action and the original model remains unharmed. My Biases may be tweaked but they try to be fair and impartial.

 

Hair material and patch images/decals were placed on the grid as bipartisonally as possible.

 

To all you nonbelievers in Bias adjustment... accept your fate... become one with the Bias!

[itsjustme]

David,

I'm looking at your movie and for the life of me I can't think of how that movement is a bad thing. If you move the CPs you want the splines to keep their shape... right? Aren't the movements more natural if/when the Bias don't automatically adjust?

 

I'm missing something simple so bear with me.

I've only looked at the movie. I'll look at the project file.

 

Let's say you adjust a bias to cause a crease for some reason, that crease is not going to go away when that area is flexed...if anything, it will become more pinched. The part of the spline that is adjusted will maintain exactly the adjustment that you model in there...it won't look natural in most areas (other than the places I mentioned in a previous post in this thread).

 

I have rigged quite a few models that have made liberal use of bias tweaking...in every case, I had to remove that tweaking (and either re-model sections or move things around as much as I could when the model had already been textured) in order to get good looking movement. I've seen quite a few people attempt to use bias tweaking to overcome bad modeling. Using it sparingly, and in only the appropriate areas, is the way to go...at least in my mind. Your mileage may vary.

[Caroline]

My current thoughts - static surfaces - bias good, animated joints - bias bad.

 

Like fingers - around the fingers - bias good, along the length of the fingers - bias bad. Would that be right?

 

River banks - bias good

 

And in hybrid's werewolf model, in the 6-pack area, bias bad. (I'm watching your comments there too, David )

 

Even though there are distinct lines (maybe not a crease) where there is a bias used, I think the decal will be enough to disguise it.

[itsjustme]

My current thoughts - static surfaces - bias good, animated joints - bias bad.

 

That's the "rule of thumb" I use.

 

Like fingers - around the fingers - bias good, along the length of the fingers - bias bad. Would that be right?

 

Generally true, Caroline...there are some things to think about here though. If you use any bias tweaking on the fingers to reduce the number of splines and round them up, you'll have to make the bases of the fingers a little different so that the five-point patches look right (check Squetchy Sam's fingers...they weren't made with that in mind, the only concern was low spline count) and you couldn't squish the fingers while animating realistically very easily.

 

River banks - bias good

 

Absolutely.

 

And in hybrid's werewolf model, in the 6-pack area, bias bad. (I'm watching your comments there too, David )

 

Exactly, the bias tweaking would interfere with rigging the stomach area and give you un-natural movement.

[Rodney]

David/Caroline,

There is no doubt you've seen the effects of Bias adjustment up close and personal and know the effects it'll have on moving meshes. Point taken and accepted.

 

It'll take me a few more years to actually experience your perspective but I'll get there.

 

The whole Alpha, Gamma and Magnitude thing has historically had me confused.

Early on I never used it for anything. Others on the animaster mailing list scared me away from it. Then I saw experienced modelers use it very effectively (in static models). The point I'd like to make, because some will read all of this as adjusting Bias is bad for everything, is that as long as the model isn't likely to be animated its safe to tweak away. At the low end of production where some error is acceptable more Bias adjustments will be acceptable in animation as well.

 

I have to gain considerably more experience in setting up meshes for animation to be able to even begin to comment about that aspect. I'm glad you can speak to it intelligently.

[largento]

I can say that when I first encountered A:M, I saw the splines/CPs/Bias Handles like paths/points/bezier curve handles that I was used to in Illustrator.

 

That was pretty disastrous for me. I was trying to use the bias handles like I would use the bezier curve handles in Illustrator and the results were *not* the same.

 

Modeling became much easier for me when I turned off the bias. I don't even peak CPs very often. If I need a sharp edge, I just end the spline and add a new spline. With it being non-continuous, I get the sharp edge I'm wanting.

 

Granted, I'm not really doing anything that has ultra-fine detail... nor mechanical models that require extreme accuracy.

[HomeSlice]

I spent many many hours experimenting with the strengths and limitations of using the various bias adjustments - both in static models and animated ones. My conclusion is that it is more trouble than it is worth. I use the Magnitude settings a bunch. They work very well in both static and animated models. But adjusting Gamma and Alpha settings just created more work for me, even with static models. The problem was that I would spend a lot of time tweaking bias handles, then later I would go back and move the splinage around (as I always do) and the CPs that had been adjusted would pop and move in ways that were not very pretty. So, I would readjust them. Then I would go back later and tweak the splinage again (as I always do - did I say that already?) and have to redo all the alpha and gamma settings again ... ad infinitum. So if you want to play with the bias handles, even with a static model, make sure you are absolutely done tweaking it before you start adjusting alpha and gamma values.

[robcat2075]

I'm quite doubtful about the central premise here. Here's why...

 

bias0002MP4.mov

[itsjustme]

I'm quite doubtful about the central premise here. Here's why...

 

Hmmm, you're right about it not being locked down, Robert. However, if bias tweaking is used in a bending joint, it's going to cause problems if not done correctly...and I have run into none that were done correctly in the models I have rigged (almost as frustrating as having the model already textured). I still think you are safer avoiding bias tweaking in areas that will move quite a bit in the vast majority of cases. There may be modelers that have the ability to use bias tweaking in a "rigging friendly" manner, but I haven't encountered them yet.

[robcat2075]

if not done correctly

but that's true of anything in CG, not just bias adjustments.

 

 

Based on your comments I think the problem biases happen when they have been used to force splines into a shape that should have been arrived at by good CP placement instead.

 

When I've used bias to un-crease my own models it's always been very small adjustment of 0 to maybe 3 degrees and i haven't had a problem with moving those around.

 

I bet on your own models ( because you know where CPs ought to be) biases wouldn't be a problem.

[itsjustme]

if not done correctly

but that's true of anything in CG, not just bias adjustments.

 

Sure.

 

Based on your comments I think the problem biases happen when they have been used to force splines into a shape that should have been arrived at by good CP placement instead.

 

In about 99% of the instances that I have run across...one that wasn't the case sticks out in my head at the moment. That one was an experienced modeler trying to add a crease in a muscle...it looked good static, but didn't work in movement.

 

When I've used bias to un-crease my own models it's always been very small adjustment of 0 to maybe 3 degrees and i haven't had a problem with moving those around.

 

That small of an adjustment could work. The adjustments I've had to deal with were not generally that small.

 

I bet on your own models ( because you know where CPs ought to be) biases wouldn't be a problem.

 

Possibly...I haven't come across an instance where I've had to try it though. Good spline layout is going to do the job in almost all cases...in every case of modeling I've come across so far, at least. You never know though, so I'll leave the qualifier in there. I think the only time I might resort to bias adjustment (other than top of the head, teeth, finger nails, finger tips, edge of the nose, eye, ear, or some other non-moving part) would be if I couldn't change the spline layout (the character already being textured or something) on someone else's model or if I was aiming for a very low spline count.

 

On my own models, I go for the most squashing and stretching ability I can get...I like having options. That means that I'm going to be pushing and pulling CP's in all kinds of directions and the modeling has to hold up under it. It's also one of the reasons I use CP Weighting 99.99% of the time and avoid Smartskin except in very rare circumstances.

 

The biggest thing I want to avoid is giving an inexperienced modeler/rigger the impression that tweaking biases whenever they run into a problem is the way to go.

[robcat2075]

It's also one of the reasons I use CP Weighting 99.99% of the time and avoid Smartskin except in very rare circumstances.

 

The biggest thing I want to avoid is giving an inexperienced modeler/rigger the impression that tweaking biases whenever they run into a problem is the way to go.

 

I think adjusting bias in modeling and doing it in rigging are two different issues. I haven't had many reasons to adjust bias in an action or a pose as your sample did. I only use bias in the modeling stage.

 

I agree that CP weighting is the main tool for rigging. But I wouldn't tell modelers that bias is bad. I'd tell them that bias is last thing you should adjust if you need to smooth a mesh and that making the mesh right is the first goal.

[itsjustme]

I agree that CP weighting is the main tool for rigging. But I wouldn't tell modelers that bias is bad. I'd tell them that bias is last thing you should adjust if you need to smooth a mesh and that making the mesh right is the first goal.

 

Anything is fine in moderation, the difficult thing is knowing when things are necessary. Very experienced modelers have had problems using bias adjustments so that they will work when a character moves.