Thanks for the suggestion!

I'm really interested in hearing about some examples, where this feature would be helpful, and not be something you can easily achieve already, with the tools available.
If you are able to share what kind of use cases this would help solve, it would better help us decide, if it's something we want to implement, and what kind of way we would potentially implement it:

1. Should bone scale both up and down?
2. Would a scaling limiter be required?
3. Should it scale the bone uniformly or only one one axis?

Those are some of the questions that needs to be answered for now

As Shiu mentioned, it would be helpful to hear more about the use cases where the new options would be applied.

Potential options:

1 bone IK
1) don't stretch,
2) stretch longer (not smaller),
3) stretch smaller (not longer),
4) stretch longer and smaller

2 bone IK
1) don't stretch
2) stretch longer (not smaller)

Currently we only have the 2 bone options. The 1 bone options could be 2 checkboxes, maybe Stretch and Compress.

A similar request was made here, additionally asking for uniform scale. This is turning IK constraint into a "distance-to-scale" constraint. While that is likely useful for some things, it may be better to have a separate constraint for tying together different transform properties rather than overload IK. IK is primarily about adjusting rotation so the tip of the child-most bone is closest to the target. By using a new, separate constraint for distance-to-scale it can be made more configurable (eg, limits and other options) without cluttering IK constraint. Still, it would be helpful if you can describe use cases for IK with uniform scale.

I forgot to add: you can simply use a small scale (or a short bone) with the current Stretch checkbox to achieve an effect similar to 1 bone IK #4 above. That makes the only setting not currently possible out of all 6 is 1 bone IK #3.

I'm mainly interested in the stretch/compress feature for single-bone uses. Previously I've used a path with a single bone (or sometimes multiple bones) set to chainscale in order to stretch those bones to a target. However, I want all of these bones to stay in a straight line (they are all at the same angle, linked like a chain). This is overly complicated to set up with a path, especially when you want those bones to be in a straight line, and there is currently no option for straight-line paths (at least not that I'm aware of).

This is turning IK constraint into a "distance-to-scale" constraint.

Yes, if you would prefer to make this a different type of constraint ("distance to scale"), that would be fine for my use. It is a bit of an edge case, so I can probably use the trick I mentioned in the other thread for now, where I set the scale of the bones I want to stretch/compress to a very small scale, and let the IK stretch them / compress them freely.

Any thoughts about my second question:

Is there any way that you could make it so that the IK-constrained bone could scale on both X and Y instead of just on the X as it is currently? Ideally this would also be a checkbox, labeled something like "scale both axes"

Thanks for the input guys! We'll look into the possibilities

For one bone suits me like it is done now, but with a small correction. This is the ability to scale the bone when the target is closer than the length of the bone.

Hi, Belpetr. To fix the above issue (the one shown in your video of the spider), select the bone you are stretching (in this case, the spider web strand), and set its X scale to something very small - like 0.001. It will still stretch to match the spider's location, but now it can also shrink down to the scale you set (0.001). With this setting, it should both stretch and compress to match the spider's location.

In 3.7.25-beta:

Image removed due to the lack of support for HTTPS. | Show Anyway

So promising, can't wait to test the stretch one!

3.7.25-beta is up, so you can try it now!

belpetr wrote

It would be great to be able to shift the child's bone

You've always been able to translate the child bone. The problem with scaling the parent bone is that it almost always causes the child bone to need to rotate to continue pointing at the target, especially with non-uniform scaling. Also, scaling the parent bone will shear the child bone, which is unlikely to be what you want. Maybe you'd be better off using two 1-bone IKs?

This is not a bug, it is an IK limitation.

If the IK mix is greater than zero and the parent IK bone has nonuniform local scale (meaning the X and Y scale have different values), then the child IK bone's local Y will snap to 0. Inheriting nonuniform scale does not trigger this snapping.

Spine's IK is the only I've ever seen that allows nonuniform scale at all. It was a very complex problem to solve and requires the child bone Y to be zero.

Your knee turning problem is not very clear to me. I saw your video above, but I'm not sure what you wanted by translating the child bone. Form your image it looks like you want the bones to bend the other way? Maybe you can animate the IK mix to zero, placing the FK bones so the leg is straight when the mix is zero, then flip the IK bend direction and animate the IK mix back to 100%.

Hello Belpetr, if you set two IK constraints with one bone that have the stretch and compressed options turned on, you can achieve the same effect of "Freeze "Joints On" while the first one just requires one IK with said options.

That's true Erika, though when you move the target of the lower 1-bone IK, it won't do IK for both bones.

There may be something we can do to improve translation of the second IK bone, like in the videos above. What happens now is the translation is applied to the FK pose, which tends to make the IK pose surprising.

I stumble in this threat looking through bug, And since the now stretch compress option is release, I try to experiment on building leg, arm without using the conventional way.

to make some blend between functionality and possibility.

Its a work around I think, but look close to what is the need