Persona Video Tutorial Series

Hi Guys,

My name is Benn and I’m currently working on Persona - UE4s animation toolkit. I’m looking to develop an animation tutorial series to explain the workflows we use and wanted to get some thoughts from the community on their direction.

What would you like to see? Is there anything particularly confusing or obscure about our animation tools? I’d just like some ideas so we make sure that we’re covering the right material.

Thanks!

Benn.

Something that would apply to those of us that are not in school, or too poor, to afford Max/Maya, would be nice… I know they’re industry standard, but Blender is all I can afford! :slight_smile:

I’ve signed up for the 3 year Maya/Max student editions. I figure if I’m still using them in 3 years not only will I know what I’m doing with them, but they’ll be worth paying for.

I’m learning from scratch, never done any of this stuff before, so it’s all going to be great info. It would be great to see some material on blending physics with animation. Really looking forward to seeing these videos! Many thanks…

Are you saying we should have blender tutorial? At this point we’re looking forward to hearing something in Persona that you found it difficult or not sure what to do.

I’d like to see a little more in depth info about the 1D blend space and such.

I guess I’m just not sure how to properly set it up. Right now I have something that sort of works:

  • Took the character in his walking pose, pulled the arm out (animation is for holding a torch out), and exported it as a single frame.

  • Set up the 1D blend space, and selected frame zero from the walk as the base blend (or whatever, don’t have it in front of me to tell the exact name).

  • The hold-torch animation is on the blend space, it blends with walk/jump etc properly.

The problem is that I don’t really know how to make it turn off when the value is set to 0 in the 1D blend space. The way mine is set up (and I tried duplicating how the aim offset stuff works to some degree) the hold torch animation blends in regardless of what I do, even in Persona with the value slider on the blend space pulled all the way down.

I’m not sure how to make it so that at 0 value on the blend space, it just gives no contribution from the torch holding animation.

I’d love to see a start-to-finish setup of a simple static blend pose like this. Maybe it’s there already and I missed it!

Thanks guys :slight_smile:

I like to see about retarget I how work with diferent models that share the same bone hierarchy.

If do you have a 4 Npc with different mesh but same bone hierarchy? Do you need 4 duplicates animations? and 4 animationsBlueprints equals, only changing the duplicaded animation reference?

I know this is a large favor, but could you add importing an animated character from iclone? Nobody I have found has used it so far except me, and I think once people see the quality models and animations that Iclone can bring in, it might gain traction.

Rather unfortunately I have not had enough time to play with it yet - I am coming from 2 years + using Mecanim (Unity) and have some pretty complex stuff I want to port over to it but won’t be doing it for another few weeks yet due to work load. I’ll post questions as soon as I can.

Done a bit more thinking - would this fall in the right area?

A overview of how to use a multipart character with animations, morphs for shape and facial animations etc?

Basically a Customisable character with multiple parts.

Still trying to wrap my head around Persona. So great to hear there are more tutorials in the works. Some more in-depth stuff about blending several animations properly would be great.

Tl;dr Procedural animations (like walk cycle from two poses).
I would be curious if it is possible to achieve out-of-box in Persona, or would I need to write some supporting code first.

Other than that I was really impressed about the multiple poses for different weapons, setup. I’m currently facing similar problem just multiplied by 2.
In my case I have different weapons, and on top of that dual wielding different weapons (where it make sense, although nothing wrong with Rambo style two M92 in each hand). I would be really interested in more detailed explanation about the setup used in Fortnite.

I’m also curious about those attachments setup, where attachment bones which Matthew mention, could be used to animate, well attachments without actually creating separate animation for each possible variation. It’s yet another problem I’m currently on, as my characters have lots of attachments (at least 6 weapon attachments, shoulders, random things like pouches). I’m currently trying to make all of them physicalized, which would not be bad solution, if I could get it working.

And besides that non physical solution might be just more efficient/fitted for some cases.

>>I would be curious if it is possible to achieve out-of-box in Persona, or would I need to write some supporting code first.

We’re working on making this easier. Currently it is difficult to do right out-of-box - i.e. spring blending. Our blending is set ranged from [0, 1].
Once we iron out some workflow issues, we’re going to post tutorial. But this won’t be happening at the first tutorial.

>>In my case I have different weapons, and on top of that dual wielding different weapons (where it make sense, although nothing wrong with Rambo style two M92 in each hand). I would be really interested in more detailed explanation about the setup used in Fortnite.

I asked Laurent to write “tricks and tips” for this kind of issues, but basically fortnite has one pose animation per weapon type. I think you can try layer both hand to be different animation depending but you might need some additive trick to make it look different between one weapon vs dual.

>> my characters have lots of attachments (at least 6 weapon attachments, shoulders, random things like pouches

For character attachment, it is easier if you just use spring controller than adding real physics. Usually looks more stable and better.

I’ll send this to Laurent and Matthew for them to join.

Multiple character parts can be done using MasterPoseComponent. That will require the partial bone hierarchical to be same, but if you have all body parts, it will follow what master does.

For master/slave, you can use blendshapes. You can use blendshapes for all animations. Make sure you enable “Import Morphtarget” when you import the mesh. The name of Morphtarget should match with name of BlendShape Curves.

Facial animations are a bit different topic. You can do this using just animation (scaled animation) or blendshapes - FBX Animation Pipeline in Unreal Engine | Unreal Engine 5.1 Documentation -.

We have a couple of third party who are working on facial capture system, so those plugins will be coming.

Thanks,

–Lina,

I’d love to see some nav mesh explaining while in these AI sectioned tutorials. It seems to confuse some people…I struggle with it sometimes as well. :stuck_out_tongue:

Hi iniside,

Our system in Fortnite works as follows:

  1. Base Layer is a 1 frame “weapon pose”. For aiming weapons, we actually have 3 frames (aiming up, center and down).
  2. Second layer is legs (and a bit of Spine) for locomotion. That includes idle, walk, run, jump, fall, land. We only blend that layer when there is movement going on. So idle, will actually be “weapon pose” layer full body.
  3. Third layer is an additive locomotion. We layer additive animation on top of the upper body. That is weapon sways when moving, a bit of bobbing when jumping and landing, etc. We also add a bit of leaning for strafes and turns.
  4. Fourth layer is for additive hit reactions and cringes. We use this when we don’t want to take control away from the player (like a full body root motion animation would do). So we can show hit reactions, but still allow player to jump, fire, etc.
  5. Fifth layer is for IK. We first retarget them, and then move hands to their IK positions. This will correct any side effect we have from retargeting, FK blending, etc.

When you dig in further there is a bit more complexity to these. But that’s the high level.
For example the weapon pose layer for aiming weapons as a couple of “Slots” to play AnimMontages. The first one is for firing animations, and is in “aiming space”. So we can author a single firing animation aiming center, and it will properly remap to your aim direction.
Another one is for “upper body”, and will cancel aiming animations. We use this for weapon reloads for example.
The nice thing about these, is that the additive locomotion layer will transform all these actions. So if you reload a weapon, and jump, you’ll have the jump “bobbing” effect applied to the reload animation. Or if you reload and run, you’ll have the run sways applied to the your reload animations.

Obviously mixing these kind of animations doesn’t always work well, so we have a curve for these layers, so we can tone them down if needed.
A good example of this is getting shot while being downsights. We tone down the hit react additive animation, because the weapon is so close to the face, and the camera is really close to the character. When he’s aiming from the hips, we can have the full effect of the additive hit react.

Hope that helps. As Lina said, I’ll try to work on some blog post about this. Or if there is enough interest, maybe do a live stream about it.
Post your questions and I’ll try to answer them the best I can.

[QUOTE=Laurent Delayen;109402]

Hope that helps. As Lina said, I’ll try to work on some blog post about this. Or if there is enough interest, maybe do a live stream about it.
Post your questions and I’ll try to answer them the best I can./QUOTE

+1 for the blog and detailed information on this layering setup. I watched the twitch stream which was fantastic btw, but I wanted to get a more readable version of the animations blueprints…state machines…ect… :slight_smile:

thanks for being Epic!

I’m very interested in Tutorials on Skeletal Animation similar to the reasons above. Entity Animation is a subject (along with 3D modeling) that I’ve being itching to get into as I was previously code-centric. I’m very happy about UE4’s support for Blender and look forward to creating procedural/real-time mocap/hybrid animation with Persona. I’m currently studying UE4 Persona Docs and reviewing Animation Content Examples & Morph Target Examples. I’m still a little foggy on terms Persona, Animontage, Blendspace, Animation Blueprints, etc. I’m working thru it with the Persona Overview in this UE4 Tutorials. Look forward to any addition Tutorials on this awesome software.

It would be nice a better explain for what is useful Animation Montage and, in general, when is worth using it instead of playing an animation directly (something like UDK PlayCustomAnim()).

Thanks.

I’d love to see how you guys would handle multiple part meshes with completely different skeletons that need to be fused together to create a resulting character. Something like a character that has an independently animated faced and hair (with entirely unrelated skeletons) that need to follow identical animations to the main body. As one cannot use child blueprints for different skeletons, I’d love to see how you guys would reduce the workload, fuse the parts together, and get them all playing nicely. Also something neat to see would be how you would do composite animation variants for different animations that don’t need a separate blueprint telling the mesh to use those composite animations - only when they are necessary (for example fusing different animations of a skeleton together, but only when certain meshes are using it).

I have a solution working on my end (which involves manually copy pasting every animation change to every blueprint), but it seems tremendously inefficient and certainly cannot be the ideal way to do this (maybe there’s something I’m missing, and blueprints can inherit animation data from each other?), so I’d like to see how more experienced hands would do it.

so why is this a sticky? not a very useful thread