Add transcription for: frames/MDA_TheSymbioteCreature_DownloadPirate.com_SYMBIOTE_CH010_FEM_SETUP_1080p_mp4_frames.zip

transcriptions/frames/MDA_TheSymbioteCreature_DownloadPirate.com_SYMBIOTE_CH010_FEM_SETUP_1080p_mp4_frames_transcription.json

@@ -0,0 +1,8 @@
+{
+  "text": " So, long story short, I changed headsets during the recording of my FAM lesson. But apparently the sound was muted, so I didn't record any sound in the tutorial, which is an hour and a half long. I'll try to compress it into less than this time, because it was a little bit too long and I had some errors there, but I ended up doing a good simulation, a good wrinkle simulation. So the goal here is going to be making a nice wrinkle simulation on top of the development one. So I made a network box here, I'm gonna color it like something like this. I object merged the original quad mesh that I had before. And I'm gonna also copy and paste the object merge that contains the simulated Tetris. So I'll make a null here, which is going to be the null that points to the simulated tetrahedral mesh, geo tetras, animated, goes into the geometry. I will connect it to the clean version of it, so we have a minimum parameter of attributes here. So yeah, this is the simulated version, tetrahedral version, and this is the clean quads static mesh. So yeah, we'll jump right into creating our FEM simulation. The goal here is to create a core geometry that will drive the entire body and the two skin layers that are on top of it. In order to do this we'll be using the original mesh and we'll be transforming it into a volume first with the aid of VDB from Polygons. So we're going to transform this into a volume, a quite detailed volume with the resolution of 00012. So this time we'll be doing some some operations with this volume because we need to make three layers out of it. So first we'll do a convertVDV and convert this to geometry and you'll see why. So basically we're reconverting into geometry. This is gonna be our first layer, the top skin layer and now we'll do some VDB reshape operations, VDB reshape as the F. and we'll instead of dilating, which is by default, we'll erode it by one voxel and do another copy without dragging the node and eroding it by two voxels. So we have one geometry and we'll copy the convertWDB in both of these and we'll have the first skin layer, the second skin layer, and the third is going to be the core. Every time it's shrinking in volume, so the goal here is to make a remeshed versions of each of these layers and the remesh I'm going to sheet escape and disable this and set the parameters that I found that work. Ten iterations for the remesh and the smoothing to zero often when you have really precise geometry and you have tips like this very sharp areas the smoothing will just kill off in detail so if you if you just leave it at zero one or more it's gonna just round this off. So I'm gonna set to 0 and the size for remesh is gonna be 0.025. So you'll see why we're doing this, like let's remesh this and see what's happening. Basically I want to obtain a more detailed mesh than the original because if you look at the original, the original is like this, and you remesh it, it's quite pretty much the same but we can push it as far as we as far as we want because the remeshing detail that we set here is going to decide how much detail in your wrinkling you have. This layer, this skin is going to be the one that creates wrinkles. So for wrinkles you need some more geometry basically. So we're gonna disable this remesh node and copy it two times and we're gonna name these guys with nulls just so we can reference them after or just have them organized properly. cmbout skin level 1. Duplicate this and duplicate this again. This is going to be skin core. So we have two skin layers here and one internal core layer, which is going to control the whole body. We're going to use a lot of internal attributes that FemmSelva uses and one of them will be fusePID. I'm going to create an attribute with attribute create this time. Usually I don't use attribute create for creating the attributes, but for FAM it's quite mandatory because we have to have a default value of 0, minus 1, for the vertices that do not have any data on them. Basically minus 1 would mean that the values on this vertex are invalid, so I'm going to create an attribute called fusePID, which is a point attribute and an integer with a default value of minus 1, with a default value of minus 1 and initial value of 0. So this is mandatory, it's crucial to have this, and the size of this value is 1, so it's not a birthday, it's just an integer, plain integer value. And then we'll make an attribute triangle and create, on top of the attribute create, assign some values to this FusePID. I'll explain why we need this. So the FusePID will be equal to the ptnum. If you, let's display this attribute, and the attribute is going to be hate scholar by default, but we need just markers. You have a bunch of numbers here, which represent the ID of these points. So we'll also group this. We'll name it inner interior, I mean, It's going to be a point group, interior. Or maybe let's make it a primitive group, yeah. So interior. And we'll merge these guys together, all of them. This one, this one, this one. Merge them into one. Move this one here. Move that one there just for the order. And we have three objects merged into one. And you'll see why we have the fused PID here. So let's make a clip node. The clip will help us debug stuff inside the mesh because we'll be using it a lot. As you can see, we zoom into the edge and we have three layers here. Let's make a connectivity so you can see better. Connectivity, yeah. Connectivity, class, marker, color, random from attribute, and that's it. So as you can see, we have three colors. Purple, green, and red on the outside. The purple represents the inside, the core. Green and red are two skin layers. The goal in FEM in this simulation is to create two skin layers that are sitting on top of an internal core layer. The core layer will be controlled 100% by the motion of the vellum simulations we've done earlier, and the red and green layers will be simulated by FEM. So, yeah, this is for preview purposes. So, yeah, we have an interior group, which is a primitive group that contains all the triangles. Now we'll make a tetconform, which is a... you already know this node, and that conform will fill up our internal core with tetrahedrons. But don't forget that we disabled those press escape. We disabled those remesh nodes because we need triangles in order to make proper tetrahedrons, otherwise we'll have a lot of issues, new simulations. You see that it created three layers made out of the quantified mesh which is output by the convert VDB. So I'll come back when it just lets me stop this operation. So yeah, just let me let me back in. We have this retrohedral mesh. As you can see, it's quite detailed at the surface and it's simplified in the middle. The deeper it goes into the tissue and also I turned on the remesh and it's gonna take a while to remesh but again it's re-evaluating the triangles and the remesh into triangles but I'll show you an example of what usually happens when we use FEM for wrinkles. This is a simulation I've posted on Twitter before so this is my octopus churro that I use for a bunch of different experiments so basically yeah the motion is driven by vellum, but the wrinkles are done with fem, so it's a secondary pass. As you can see, there's quite some detail in the mesh, but the resolution of the mesh is quite high because, you know, Turing is a big guy, like two and a half meters wide, and there's a lot of geometry to cover. And yeah, so I think this is the best example for the system, because I used this exactly the same system for Turing. So yeah, now we've triangulated all these guys. So we have three layers, but I forgot to do something. I wanted to link this target size, copy parameter, copy parameter, paste relative references. And it's gonna re-value it, which is dumb in my mind. Paste relative references. And I'm going to multiply this value by, oh well, just go back and multiply this by 1.5 and make the internal core layer a little bit less detailed. So it triangulated this guy. I'll show you what happens when you have a fused BID attribute created on the core with the default value of minus 1 and the initial value of 0. of zero. So basically this minus one value defines the default value if there's no data assigned to this attribute. So if we go in here, we merge all these like triangulated meshes together, you'll see that only the vertices that were assigned some data, they have anything but minus one. So we go here, we display the fuse PID. So you see that our core layer has numbers, IDs, and the other two they don't. This views the ID if you change the minus two, for the initial value you put minus two. The meshes and the objects that don't have anything assigned to the vertices they will be minus two. So let's leave it at minus one. Femme understands that when it's minus one it's an invalid vertex for the... like this attribute doesn't have... there's no attributes on these vertices basically. So let's leave it like this and yeah, We've yet to conform this skin core. It's transforming it into tetrahedrons, as you've seen before. These are some really expensive operations, especially when you have big meshes and detailed meshes and big monsters. We will add another group operator here. We'll call it inner. So this is going to be the inner part of the mesh. It's going to be a point group. we'll be adding the step conform here now. So basically, let me do the flow, the entire flow here, and then we're going to save this geometry because I don't want to spend time evaluating these things. So the step conform, the only difference between this step conform and this one, will add surface triangles. So it's a lot of layer of polygons on top of the tetrahedral mesh, on the outside. Okay, so now we'll be group promoting. our interior points. So the interior will be promoted to the points because it was a primitive attribute and will last that interior group. Okay, and now I'll come back when I evaluate this.conform. I'm not sure what's happening. So it re-evaluated the whole thing. Now, because we merged three triangular meshes into one with some fused PID attributes, some of them are invalid, which are minus 1, we're going to clip this to see what's happening inside. This is crucial just to debug what's happening in the mesh. And you'll see clearly that we have some PID attributes, only on that connection between the core layer and the two-skin layers. So you see the two-skin layers are clearly defined. And this separate kind of surface, this is the surface triangles that we added here in the tetcoform. So everything looks good for now. What we're doing now is promoting the group. And yeah, so we're promoting the interior group. So now the interior group was a primitive attribute. And now it's become a point attribute and we're blasting the interior. So let's see what it does. And now we're left with the two skin layers plus the surface triangles that are on the outside. So we basically removed the tetrahedrons that are outside. So the next step would be merging this inner group that we had before with the reminder of our last operation. So we'll also add a, we'll duplicate this group, we'll call it outer. And the outer group is gonna be also points and we'll merge these guys together, the outer and the inner. And now we have below this clip node over. This is how I usually work. I just move my clip node lower and lower every time we're evaluating and see what's happening in the next operation. So we have the internal core, the skin layer, and the surface triangles. Okay, and for optimization purposes, we'll just save this guy. And I already made a file cache in my previous attempt to record this tutorial. I called it geocached, GweSympioFM type mesh v01bGoC and I'm going to save it. So now we have this really nice mesh, which has 892,000 tetrahedrons, 143,000 polygons on the surface, a fuse PID attribute, a bunch of groups that we'll be using during the simulation. And the next step would be creating an ID attribute. So this is not going to be a fuse PID, but just an ID attribute. we actually can go back and steal this attribute triangle and instead of use BID make it an ID and we'll create in a safe manner again an attribute with an attribute create attribute create will create again an internal attribute that is recognized by Femm solver which is called solid shape stiffness and it's a float and as a default value we'll use one and the initial value will be also one so any part of the mesh that doesn't have this value assigned this attribute created on it will have a default value of one and let's blast this inner part and only the inner part so we'll leave everything will remove everything except the inner part so we are left with the core and I want to assign a higher solid shape stiffness to the spot of the body. So if you see we disable it, we have the skin removed basically, this blast node. So we removed all the trihidrons that are outside. And we'll group expand. We'll group expand the surface points. So we're going to group expand the surface points. So the surface points of our inner primitive group. So what's going to happen here is we're going to remove only the surface of the tetrahedral core basically. And... Yeah, so... blast. Oh yeah, it blasted the inner part, but I should have blasted the outer. So basically we're leaving only the outer part and the group expanding by the surface points and blasting the surface points. I just made a mess. So surface points. Yeah, so basically I'm going to set the clip, see what's happening there. or blasting the outer shell basically. And the group expanding this guy. And we're applying... Yeah. So what I'm doing here is blasting everything except the outer layer. group expanding and then group expanding means that we're kind of expanding into the inside the mesh and then we're blasting the surface points. What happens here is we're gonna apply a different stiffness to our first row vertices that are inside the core and you'll see why. fsolid shape stiffness equals and I'll make a challenge. I'm gonna get the name of stiff. Click on the slider button, set it to 4 and we're going to copy this by ID to our original mesh, which is the entire mesh. And we'll set ID as a matching attribute. Whenever you do this kind of stuff, you can create an ID attribute before, do some operations like blasting and working on a separate part of the mesh, and then you can copy the attributes back to your entire mesh by ID. It will copy all the data that you applied on this part of the mesh onto the original mesh. This works also very well when you have IDs on an animated geometry, you time shift it to one, and you do operations and then you copy back your attributes to the moving geometry. We'll do it a little bit later also, you'll see how it works. And we're gonna copy the solid shape stiffness. Let's see what we've done here. And the clip node is not supposed to be here, but it will be here. And you'll see what happened with our solid shape stiffness. Let's visualize this attribute. It's going to be a number. it to 4. We'll see. Yeah, so I did the mistake here. Instead of, we're deleting basically the outer, the skin layer. So we're left with the core. This is what I wanted to do initially. I was a little bit confused by on my own setup. So we're expanding the selection of the surface points, which is the surface of this core layer, and we're expanding it by 1. So it's going to expand the group by one row of vertices, and we're removing the surface points. So we're basically trying to select the next row of vertices inside the volume of our core mesh. This is what we've done. We applied the solid shape stiffness to the inside of that layer of vertices, copied the solid shape stiffness to the original mesh. Now we're clipping the mesh just to see what happens. So as you can see, at the boundary between the core mesh and the two skin layers, we have this attribute which is higher than the solid shape stiffness on the outside. So it's four inside and one outside. This is gonna act as a stiffness inside the Fem solver. It's going to override entirely the shape stiffness inside the Fem hybrid object that we'll create. And next we'll create an attribute triangle again. Let's copy this one so we don't really create it again. How we, not the best one, we'll remove this parameter. Except remove this guy. I'm lazy just to create another attribute triangle. I've done this setup a few times prior to this. So we'll create a pin to animation attribute, which is an internal attribute recognized by Femm also, which says to Femm that this specific group, which is gonna be the inner one, which is inside this inside core mesh, will be obeying the animation that comes from Vellum. The blue one, which is a skin layer, will be simulated, which will not be, basically, it will be attached to this red boundary, but it's not going to be pinned to animation. So if you want to visualize this, pinned to animation, if you want to visualize any attribute, you just hover above the node, click I, click your onto your attribute, and then it's going to be added to this list if you didn't know about it. So yeah, where pin to animation is between one and zero. Now it's random by attribute, but I want to make it ramp attribute. So the red zone is going to be pinned to animation. The blue one will be obeying following the red part because there will be internal collisions. Next we'll make an attribute called... let's rename these guys because set stiffness core. This one set ID. Then this will be set pin to an IN. this will be setRest and setRest. And in this attribute triangle, it's a special one. Let me remove the group that we used before, disable the display of attributes, and I'll create two attributes called vMaterialP. It's a vector attribute. And vBaseP. So, yeah, these attributes they represent the rest state of the mesh on the first frame. It's used internally by FEM to know the initial state of the mesh. And if there's no material P, it's going to switch to base P as a fallback attribute, if it doesn't find material P. So, and the opposite, usually if it finds one of these, everything is going to go smoothly. these are rest attributes as if we're creating a rest position. So the rest, as you remember, it's an O that creates a rest attribute. It takes the position from the first ring of the timeline and applies it into a separate attribute. Same goes for them, but these are specifically named attributes that you have to obey basically and follow. We have this master transform that we used before and we already referenced it here in order to create a clean animated mesh, the quads that are animated, so we're going to copy it here. When you copy a reference note that it has referenced channels, it's going to be referencing the original, so I'm copying the master transform and now we'll do some more groups. I know there's a lot of groups so bear with me here. We'll create the group called tetrahedrons. A great thing that was added to this group operator is the geometry filter. So we'll use the geometry filter to select only the tetrahedrons. So all tetrahedrons are selected here. No polygon meshes are selected here. So if you see the groups, the groups will be containing 992,000 primitives, exactly like the tetrahedron group here. same we'll create another group which will be called polygons and instead of tetrahedrons here we'll use polygons and the polygons group has 143,000 as the polygons field here so all right and now we'll be copying this guy over naming them naming them set collisions as whole and will create two other attributes that are internally used by FEM and they'll be applied to primitives this time and we're applying this attribute to all the primitives in the system, collision ID minus one and I will copy this guy over and name it exterior collision ID. So we're basically applying invalid values to all primitives in this attribute. So all the primitives will have this as a collision attribute, which means that collisions are basically disabled everywhere inside the mesh. And we'll duplicate this guy again, set collisions. we're going to re-enable all the collisions only for the, we'll leave just the exterior collision ID, but we'll set to zero. So basically zero means that the, the trahedrons will be colliding between themselves and the outside colliders. So, yeah. So it's also a primitive attribute. So let's see. Yeah, we don't have to visualize this because we know what happened here. So now, yeah, so the tets, yeah, the tets will be colliding with all the outside the trichegons and other objects. And now we'll be quite ready to connect this thing to the Femm solver. So let's create this Femm solver. The network in the first input goes our geometry. And in the second one, you'll see what goes here. We're gonna pull our animated Tetris and the time shift this guy, remove the channels, make a point before. And we're gonna pull the geometry that comes from the master transform here, connect the time shift to the second input and the animated live geo to the third input. Yeah, so it works. Our tetrahedral mesh is following the valium simulation. And what we're going to do here is work on the FemSolver first, and then we'll come back to this point before. I will actually connect the second input. So the first input is a static tetrahedral geometry with all the shenanigans and attributes and whatever. And the second one is the animated version of it. See the animated version and the static version. Now we'll go inside the Fem solver, but first we'll name it FemRinkles. Enable it, go inside, now let's create a Fem Hybrid object. Why a hybrid object? It's because we have a geometry containing polygons and tetrahedrons at the same time. We'll disable particle, we're in rod, we don't need these, but we'll use the shell and the solid, which makes sense because we have a shell, with a skin and the solid. The skin is the outer polygon layer, which is controlled by these attributes and the solid is the old tetrahedrons in the mesh that are controlled by these parameters. I'll change just a few parameters. The shape stiffness of our shell, let's say a thousand, and just for the beginning parameters, the shape stiffness doesn't really matter because this is the solid shape stiffness that we defined earlier. So this is, let's put 16, but it doesn't really matter. It's like a, it doesn't act as a multiplier, the value that we've set before. So, and the value stiffness can be 100. This will define the stiffness of the tetrahedrons contained inside the two skin layers that we created. As for the repulsion, we're going to set this repulsion at the highest possible limit. Let's say, I mean, it can go higher, but I'll leave it at this, and the friction at zero. I don't want to have any friction on contact. So now we go into the deformation tab, and in the deformation tab, we have to tell them which object is going to be deformed, which is the initial, basically, geometry that we are connecting. So let's type in a back tick, op, input path. Path. Two dots represent the, basically the current geometry. It goes into the solver. And because we're basically going back a level, if you can type a slash there, but it's not going to make a big difference. So we have this, let's connect it directly to the output. And we see that we are pulling the static geometry. is getting solved here. And we need a target deformation because we really want to deform this mesh by this guy. So this is the live geometry that is in the second input. So the second input has the index of one. So it's all good. Let's add the FEMS solver here. I'm going to think for a while, read the point before I realize there's movement there and we'll change the solve method to, we leave it to Gnl. The Gnl is a solver that just, yeah, the Gnl is a solver that is a little bit faster and needs less computation steps to solve stuff. I use Gnl, I don't, GSL has some limits, you can read about that in the documentation, but I'll also change the integration to the second order. This is another parameter that defines how things are solved inside of M, But yeah, I'll let you read about these more in the documentation, which is quite amazing if I may say so. Because, yeah, Houdini's documentation is top-notch compared to other software. So yeah, the Fepsolev's set. In the collisions here, we'll also check self-collide each component, connect component. So there will be internal collisions between the tetrahedrons of the skin and so on. We can disable this and it's going to be a little bit lighter, but you'll have penetrations. have penetrations. If there's a lot of deformation in your mesh, you'll have a lot of penetrations. We'll leave everything here, I'm not going to change anything. Just let it be like this and we'll add some gravity even if we don't really need a lot of it because our skin layer is already linked to the core layer and we don't... it's not gonna play a big role here except if you have scenarios where you want to have a lot of jiggly fat that is getting pulled by the gravity and you have hanging pieces like or sax or fat or organic tissue. Yeah, the gravity is going to play a big role there. But here we have a really tight skin around the an internal structure like the core layer. So let's also add a constraint, which is called the FemmFuse constraint. Allotted before the gravity. And the FemmFuse constraint is a special one. Because it will allow us to connect our skin to our core using the fusePID that we created earlier. So let's name our Fem object as a symbiote. Wrinkles just so we can reference it properly. We can have a bunch of Fem objects here merged together and solved by the same Fem solver. So this is why we have to name them properly because in the constraints like the Fem fuse constraint, you'll be able to recognize the objects in the list better. So we're gonna connect the symbiote wrinkles to symbiote wrinkles, but instead of using point groups, we'll be using a identifier point attribute. And by default, it's set to fuse BID, the attribute that we created earlier. So fuse BID here, fuse BID here. It's a good thing that they have two fields for the fuse BID because maybe you change the fuse BID into something else on the other piece of geometry and you can set two different attributes here. And the type of the constraint is gonna be hard. And when you use the identifier point attribute here, you don't have any guides showing. So these parameters are not going to work. Yeah, so we're kind of set here. The only thing left to do is let's create a blast right here before the Pintu animation. And what I'm going to do is select this arm in the middle. Let's do it again like this. Click enter, delete non-selected, and we're going to work on this single arm. You can see that there's two skin layers outside and one core layer here. the let's test if our simulation even works. I'll save this as a as this guy because this is the scene I was trying to record let's call it 8a. Oh yeah you didn't see the so basically I'm saving it as a this guy 8a saved now the scene is saved let's activate our solver and see if the wrinkles are even working because of the procedural nature of our software we can really clip just a piece of geometry and see if anything works here and as you can see already there's some wrinkles already in the beginning on the second frame we already have the wrinkles creating so you have a bunch of wrinkles on the arms and the detail of these wrinkles is defined by the remesh nodes up top they will define how detailed is the skin on those arms and yes it's quite cool. We have the system working straight away which is great we don't have to debug anything. In the previous attempt I had to debug a few things but now it's working fine so you see that when it compresses the arm creates a bunch of wrinkles and it's going to look good as in that octopus we've seen earlier so if you compare the 0.4 mesh with the wrinkled Fem mesh. You can see that it's solving the last frame. So yeah, this is the big difference between the two meshes. So let's do one thing. I'm going to disable this, but I'll show you what I'm going to do with this point before inversion. I will add a... Because my goal is what? Is to create a Wrinkle Simulations only on the arms. I will simulate the entire body with the wrinkles, although I can set the simulation in a way where only the arms are getting covered in wrinkles, by pitting to animation everything except the arms. But I prefer to have some control over the areas that are affected. So yeah, we'll create an attribute triangle here. And the attribute triangle will contain a rest attribute. This time is going to be a simple rest. And this rest is going to follow the animation. As you can see, there's a little clock here. It means that the rest changes at each frame. So I will attribute copy this rest attribute by ID. ID, rest. So the rest attribute from the point before the animated version that comes from Valem will be transferred to our Fem simulation after it's done. And we'll time-shift this guy. Set it to frame 1. And we have to do one more thing. We'll object merge our clean Quad Smash. And we'll do a quick attribute paint. a really quick hatch of paint. And so I'm going to paint only the arms. And the tail. I want some wrinkles there also. Although there's not going to be wrinkles. I already painted some kind of wrinkles here in ZBrush, but that's OK. Control Shift Drag just to decrease the size. And just limit myself here for the arms end. Yeah, that's not bad. Yeah, that's cool. So this is gonna be a mask, an attribute called mask. I'll attribute blur this guy. So my goal is to make a blend between the original value simulation and the wrinkled Fem simulation using this mask. So only the arms will be, we'll have this wrinkly thing. I'll attribute blur the mask just to see it. So yeah, we're maybe 32, yeah, let's say 32. Let's leave it at that. And I'll copy this master transformals here, just so we have the same transformation as the other stuff. And we will use this time-shifted the Femm-Rinkels simulation to attribute transfer, our mask, because we have a static mesh here and we have a static mesh here on the first frame. So this is how we do the attribute transfer. But afterwards, points, points. Afterwards, I'll be doing an attribute copy again. Back to the left geometry, back to the animated simulation. So attribute copy here, I'll call it mask. And we'll do some optimization attribute cast. and we'll actually cast these guys V velocity and mask. We downgrade it to 16 bit, save a little bit of space as we did before and make a clean right here. In the clean, I'll actually remove pretty much everything except V, ID, rest and mask. mask. And as for the groups, any group that contains the name, the word surface, I'll remove it. And now I'll just add a output node, I'll call it, I paste it from the previous attempt. I'll call it simfem, v0.1, simsimp.fem, v0.1, f5. And yeah, I will enable this wrinkle simulation and let's simulate a few frames. Save this project. I'll simulate like 10 frames and come back. So I sold a few frames here. Now we see that we have an ID attribute mask, rest and velocity and what I was saying when I said that I will blend the original mesh with the developed mesh. I wanted to show you how to do it. So yeah, so I simulated a few frames. As you can see, there's already some wrinkles here being created. This is the Femmesh that was starting to get wrinkly here. And if we compare it to the original mesh, you can see that there's no wrinkles on the arm. but here we have some wrinkles, but we have issue arising here in the bubbles. So this is why we painted the mask. I don't want to have any wrinkles on the bubbles. This is why we're going to get the tetrahedral mesh. In order to get the final mesh as we had it here, this is the animated clean mesh, but doesn't contain any wrinkles. So how do we do this? First, I want to blend the, the two versions of our simulation. So let's activate this, go to the frame 3 where we have this wrinkle thing and let's do something interesting here. Let's do a point deform. Let's copy the time shift and let's get our static quad mesh as we did before after the vellum sim. So we have this, oh yeah, the static quad mesh goes into the point point-to-form, the tension goes into our fem sim and the second input and the animate version into the third input. So we will subdivide this mesh once because we need some detail there and the settings for the point-to-form will be a little bit different. So 0, 0, 1, 5 and 5 because when you average the maximum points is 100 it's going to average a lot the deformation so we don't want So we'll set it to 5 in order for it to respect the wrinkles. So now we have wrinkles on our quad mesh, but only because we have the subdivide. Without the subdivide, we'll lose a lot of detail. When we add the subdivide, you'll obtain back, you'll just get the detail back. So, and we'll insert a little attribute triangle, which is the main trick here. That's a nice workflow for blending two versions of the simulation. because we did this whole thing with the rest attribute. So the simulation, baked simulation contains the rest, which is the animate position of the vellum simulation vertices or points. So basically this Femme has a version of the simulation which is not wrinkled in the rest attribute that we saved inside the final Femme simulation because of it, and because of the fact that we have the mask attribute that we painted on the static quad mesh and it transferred it through this workflow to the FEM simulation. Because of these two things, we can write this expression. VP equals lerp, VP, V rest, one minus mask. So what it does, it uses the mask to blend our wrinkly FEM simulation and the clean non-wrinkly valem simulation using the mask. So if you visualize the mask, where the red is, We have the Fem simulation where the blue is, it's developed. So using this version of the simulation, yeah, using this version, we're going to point to form our quad mesh and get this. So yeah, that's it. Does disable this mask thing? And you can see that we have a wrinkly sim here. Yeah, so one thing we'll add here is... We don't have any color here, so in order to get the color we can do maybe another thing because we got the color and the velocity from the villain sim, which is not a bad idea. So let's copy these guys here and let's connect this and our vellum sim that was animated. Yep. So we're basically copying the color and the velocity from the original vellum sim, which is not bad. and the vectors are pointing downwards because the mesh is floating onto the ground, which is correct. And we're actually bitcasting these guys the v, the n, and the cd. So we're saving a little bit of space here. And we'll basically get rid of the group delete. We'll remove all the groups for the asterisk. And we'll also maybe remove the mask because we don't need it anymore. Let's do it with the clean. So clean. I like this node because it's universal. But it considers all the primitives as degenerate. it so let's uncheck this. Yeah let's remove everything except a few like CD, CD, NP. Yeah, and the UV is also. We want to save those. But something happened on the way. Yeah, no, everything's fine. We're actually casting this thing. But yeah, I'm going to copy this output. just name it differently, it's going to be called like this, sim main clean, sim main clean quad rinkled high, and set the range to one to 300, we're simulating 300 frames in general. And yeah, the fem sim setup is done. I will simulate this whole thing and come back to you in the second lesson, and in the next lesson, about the flip simulation, which is going to be built on top of this rinkled simulation. So yeah, it looks cool. It's quite long, but I hope you understood the whole setup and how the workflow works with these kind of sims. Thanks for listening and watching. See you in the next one.",
+  "segments": [
+    {
+      "text": " So, long story short, I changed headsets during the recording of my FAM lesson. But apparently the sound was muted, so I didn't record any sound in the tutorial, which is an hour and a half long. I'll try to compress it into less than this time, because it was a little bit too long and I had some errors there, but I ended up doing a good simulation, a good wrinkle simulation. So the goal here is going to be making a nice wrinkle simulation on top of the development one. So I made a network box here, I'm gonna color it like something like this. I object merged the original quad mesh that I had before. And I'm gonna also copy and paste the object merge that contains the simulated Tetris. So I'll make a null here, which is going to be the null that points to the simulated tetrahedral mesh, geo tetras, animated, goes into the geometry. I will connect it to the clean version of it, so we have a minimum parameter of attributes here. So yeah, this is the simulated version, tetrahedral version, and this is the clean quads static mesh. So yeah, we'll jump right into creating our FEM simulation. The goal here is to create a core geometry that will drive the entire body and the two skin layers that are on top of it. In order to do this we'll be using the original mesh and we'll be transforming it into a volume first with the aid of VDB from Polygons. So we're going to transform this into a volume, a quite detailed volume with the resolution of 00012. So this time we'll be doing some some operations with this volume because we need to make three layers out of it. So first we'll do a convertVDV and convert this to geometry and you'll see why. So basically we're reconverting into geometry. This is gonna be our first layer, the top skin layer and now we'll do some VDB reshape operations, VDB reshape as the F. and we'll instead of dilating, which is by default, we'll erode it by one voxel and do another copy without dragging the node and eroding it by two voxels. So we have one geometry and we'll copy the convertWDB in both of these and we'll have the first skin layer, the second skin layer, and the third is going to be the core. Every time it's shrinking in volume, so the goal here is to make a remeshed versions of each of these layers and the remesh I'm going to sheet escape and disable this and set the parameters that I found that work. Ten iterations for the remesh and the smoothing to zero often when you have really precise geometry and you have tips like this very sharp areas the smoothing will just kill off in detail so if you if you just leave it at zero one or more it's gonna just round this off. So I'm gonna set to 0 and the size for remesh is gonna be 0.025. So you'll see why we're doing this, like let's remesh this and see what's happening. Basically I want to obtain a more detailed mesh than the original because if you look at the original, the original is like this, and you remesh it, it's quite pretty much the same but we can push it as far as we as far as we want because the remeshing detail that we set here is going to decide how much detail in your wrinkling you have. This layer, this skin is going to be the one that creates wrinkles. So for wrinkles you need some more geometry basically. So we're gonna disable this remesh node and copy it two times and we're gonna name these guys with nulls just so we can reference them after or just have them organized properly. cmbout skin level 1. Duplicate this and duplicate this again. This is going to be skin core. So we have two skin layers here and one internal core layer, which is going to control the whole body. We're going to use a lot of internal attributes that FemmSelva uses and one of them will be fusePID. I'm going to create an attribute with attribute create this time. Usually I don't use attribute create for creating the attributes, but for FAM it's quite mandatory because we have to have a default value of 0, minus 1, for the vertices that do not have any data on them. Basically minus 1 would mean that the values on this vertex are invalid, so I'm going to create an attribute called fusePID, which is a point attribute and an integer with a default value of minus 1, with a default value of minus 1 and initial value of 0. So this is mandatory, it's crucial to have this, and the size of this value is 1, so it's not a birthday, it's just an integer, plain integer value. And then we'll make an attribute triangle and create, on top of the attribute create, assign some values to this FusePID. I'll explain why we need this. So the FusePID will be equal to the ptnum. If you, let's display this attribute, and the attribute is going to be hate scholar by default, but we need just markers. You have a bunch of numbers here, which represent the ID of these points. So we'll also group this. We'll name it inner interior, I mean, It's going to be a point group, interior. Or maybe let's make it a primitive group, yeah. So interior. And we'll merge these guys together, all of them. This one, this one, this one. Merge them into one. Move this one here. Move that one there just for the order. And we have three objects merged into one. And you'll see why we have the fused PID here. So let's make a clip node. The clip will help us debug stuff inside the mesh because we'll be using it a lot. As you can see, we zoom into the edge and we have three layers here. Let's make a connectivity so you can see better. Connectivity, yeah. Connectivity, class, marker, color, random from attribute, and that's it. So as you can see, we have three colors. Purple, green, and red on the outside. The purple represents the inside, the core. Green and red are two skin layers. The goal in FEM in this simulation is to create two skin layers that are sitting on top of an internal core layer. The core layer will be controlled 100% by the motion of the vellum simulations we've done earlier, and the red and green layers will be simulated by FEM. So, yeah, this is for preview purposes. So, yeah, we have an interior group, which is a primitive group that contains all the triangles. Now we'll make a tetconform, which is a... you already know this node, and that conform will fill up our internal core with tetrahedrons. But don't forget that we disabled those press escape. We disabled those remesh nodes because we need triangles in order to make proper tetrahedrons, otherwise we'll have a lot of issues, new simulations. You see that it created three layers made out of the quantified mesh which is output by the convert VDB. So I'll come back when it just lets me stop this operation. So yeah, just let me let me back in. We have this retrohedral mesh. As you can see, it's quite detailed at the surface and it's simplified in the middle. The deeper it goes into the tissue and also I turned on the remesh and it's gonna take a while to remesh but again it's re-evaluating the triangles and the remesh into triangles but I'll show you an example of what usually happens when we use FEM for wrinkles. This is a simulation I've posted on Twitter before so this is my octopus churro that I use for a bunch of different experiments so basically yeah the motion is driven by vellum, but the wrinkles are done with fem, so it's a secondary pass. As you can see, there's quite some detail in the mesh, but the resolution of the mesh is quite high because, you know, Turing is a big guy, like two and a half meters wide, and there's a lot of geometry to cover. And yeah, so I think this is the best example for the system, because I used this exactly the same system for Turing. So yeah, now we've triangulated all these guys. So we have three layers, but I forgot to do something. I wanted to link this target size, copy parameter, copy parameter, paste relative references. And it's gonna re-value it, which is dumb in my mind. Paste relative references. And I'm going to multiply this value by, oh well, just go back and multiply this by 1.5 and make the internal core layer a little bit less detailed. So it triangulated this guy. I'll show you what happens when you have a fused BID attribute created on the core with the default value of minus 1 and the initial value of 0. of zero. So basically this minus one value defines the default value if there's no data assigned to this attribute. So if we go in here, we merge all these like triangulated meshes together, you'll see that only the vertices that were assigned some data, they have anything but minus one. So we go here, we display the fuse PID. So you see that our core layer has numbers, IDs, and the other two they don't. This views the ID if you change the minus two, for the initial value you put minus two. The meshes and the objects that don't have anything assigned to the vertices they will be minus two. So let's leave it at minus one. Femme understands that when it's minus one it's an invalid vertex for the... like this attribute doesn't have... there's no attributes on these vertices basically. So let's leave it like this and yeah, We've yet to conform this skin core. It's transforming it into tetrahedrons, as you've seen before. These are some really expensive operations, especially when you have big meshes and detailed meshes and big monsters. We will add another group operator here. We'll call it inner. So this is going to be the inner part of the mesh. It's going to be a point group. we'll be adding the step conform here now. So basically, let me do the flow, the entire flow here, and then we're going to save this geometry because I don't want to spend time evaluating these things. So the step conform, the only difference between this step conform and this one, will add surface triangles. So it's a lot of layer of polygons on top of the tetrahedral mesh, on the outside. Okay, so now we'll be group promoting. our interior points. So the interior will be promoted to the points because it was a primitive attribute and will last that interior group. Okay, and now I'll come back when I evaluate this.conform. I'm not sure what's happening. So it re-evaluated the whole thing. Now, because we merged three triangular meshes into one with some fused PID attributes, some of them are invalid, which are minus 1, we're going to clip this to see what's happening inside. This is crucial just to debug what's happening in the mesh. And you'll see clearly that we have some PID attributes, only on that connection between the core layer and the two-skin layers. So you see the two-skin layers are clearly defined. And this separate kind of surface, this is the surface triangles that we added here in the tetcoform. So everything looks good for now. What we're doing now is promoting the group. And yeah, so we're promoting the interior group. So now the interior group was a primitive attribute. And now it's become a point attribute and we're blasting the interior. So let's see what it does. And now we're left with the two skin layers plus the surface triangles that are on the outside. So we basically removed the tetrahedrons that are outside. So the next step would be merging this inner group that we had before with the reminder of our last operation. So we'll also add a, we'll duplicate this group, we'll call it outer. And the outer group is gonna be also points and we'll merge these guys together, the outer and the inner. And now we have below this clip node over. This is how I usually work. I just move my clip node lower and lower every time we're evaluating and see what's happening in the next operation. So we have the internal core, the skin layer, and the surface triangles. Okay, and for optimization purposes, we'll just save this guy. And I already made a file cache in my previous attempt to record this tutorial. I called it geocached, GweSympioFM type mesh v01bGoC and I'm going to save it. So now we have this really nice mesh, which has 892,000 tetrahedrons, 143,000 polygons on the surface, a fuse PID attribute, a bunch of groups that we'll be using during the simulation. And the next step would be creating an ID attribute. So this is not going to be a fuse PID, but just an ID attribute. we actually can go back and steal this attribute triangle and instead of use BID make it an ID and we'll create in a safe manner again an attribute with an attribute create attribute create will create again an internal attribute that is recognized by Femm solver which is called solid shape stiffness and it's a float and as a default value we'll use one and the initial value will be also one so any part of the mesh that doesn't have this value assigned this attribute created on it will have a default value of one and let's blast this inner part and only the inner part so we'll leave everything will remove everything except the inner part so we are left with the core and I want to assign a higher solid shape stiffness to the spot of the body. So if you see we disable it, we have the skin removed basically, this blast node. So we removed all the trihidrons that are outside. And we'll group expand. We'll group expand the surface points. So we're going to group expand the surface points. So the surface points of our inner primitive group. So what's going to happen here is we're going to remove only the surface of the tetrahedral core basically. And... Yeah, so... blast. Oh yeah, it blasted the inner part, but I should have blasted the outer. So basically we're leaving only the outer part and the group expanding by the surface points and blasting the surface points. I just made a mess. So surface points. Yeah, so basically I'm going to set the clip, see what's happening there. or blasting the outer shell basically. And the group expanding this guy. And we're applying... Yeah. So what I'm doing here is blasting everything except the outer layer. group expanding and then group expanding means that we're kind of expanding into the inside the mesh and then we're blasting the surface points. What happens here is we're gonna apply a different stiffness to our first row vertices that are inside the core and you'll see why. fsolid shape stiffness equals and I'll make a challenge. I'm gonna get the name of stiff. Click on the slider button, set it to 4 and we're going to copy this by ID to our original mesh, which is the entire mesh. And we'll set ID as a matching attribute. Whenever you do this kind of stuff, you can create an ID attribute before, do some operations like blasting and working on a separate part of the mesh, and then you can copy the attributes back to your entire mesh by ID. It will copy all the data that you applied on this part of the mesh onto the original mesh. This works also very well when you have IDs on an animated geometry, you time shift it to one, and you do operations and then you copy back your attributes to the moving geometry. We'll do it a little bit later also, you'll see how it works. And we're gonna copy the solid shape stiffness. Let's see what we've done here. And the clip node is not supposed to be here, but it will be here. And you'll see what happened with our solid shape stiffness. Let's visualize this attribute. It's going to be a number. it to 4. We'll see. Yeah, so I did the mistake here. Instead of, we're deleting basically the outer, the skin layer. So we're left with the core. This is what I wanted to do initially. I was a little bit confused by on my own setup. So we're expanding the selection of the surface points, which is the surface of this core layer, and we're expanding it by 1. So it's going to expand the group by one row of vertices, and we're removing the surface points. So we're basically trying to select the next row of vertices inside the volume of our core mesh. This is what we've done. We applied the solid shape stiffness to the inside of that layer of vertices, copied the solid shape stiffness to the original mesh. Now we're clipping the mesh just to see what happens. So as you can see, at the boundary between the core mesh and the two skin layers, we have this attribute which is higher than the solid shape stiffness on the outside. So it's four inside and one outside. This is gonna act as a stiffness inside the Fem solver. It's going to override entirely the shape stiffness inside the Fem hybrid object that we'll create. And next we'll create an attribute triangle again. Let's copy this one so we don't really create it again. How we, not the best one, we'll remove this parameter. Except remove this guy. I'm lazy just to create another attribute triangle. I've done this setup a few times prior to this. So we'll create a pin to animation attribute, which is an internal attribute recognized by Femm also, which says to Femm that this specific group, which is gonna be the inner one, which is inside this inside core mesh, will be obeying the animation that comes from Vellum. The blue one, which is a skin layer, will be simulated, which will not be, basically, it will be attached to this red boundary, but it's not going to be pinned to animation. So if you want to visualize this, pinned to animation, if you want to visualize any attribute, you just hover above the node, click I, click your onto your attribute, and then it's going to be added to this list if you didn't know about it. So yeah, where pin to animation is between one and zero. Now it's random by attribute, but I want to make it ramp attribute. So the red zone is going to be pinned to animation. The blue one will be obeying following the red part because there will be internal collisions. Next we'll make an attribute called... let's rename these guys because set stiffness core. This one set ID. Then this will be set pin to an IN. this will be setRest and setRest. And in this attribute triangle, it's a special one. Let me remove the group that we used before, disable the display of attributes, and I'll create two attributes called vMaterialP. It's a vector attribute. And vBaseP. So, yeah, these attributes they represent the rest state of the mesh on the first frame. It's used internally by FEM to know the initial state of the mesh. And if there's no material P, it's going to switch to base P as a fallback attribute, if it doesn't find material P. So, and the opposite, usually if it finds one of these, everything is going to go smoothly. these are rest attributes as if we're creating a rest position. So the rest, as you remember, it's an O that creates a rest attribute. It takes the position from the first ring of the timeline and applies it into a separate attribute. Same goes for them, but these are specifically named attributes that you have to obey basically and follow. We have this master transform that we used before and we already referenced it here in order to create a clean animated mesh, the quads that are animated, so we're going to copy it here. When you copy a reference note that it has referenced channels, it's going to be referencing the original, so I'm copying the master transform and now we'll do some more groups. I know there's a lot of groups so bear with me here. We'll create the group called tetrahedrons. A great thing that was added to this group operator is the geometry filter. So we'll use the geometry filter to select only the tetrahedrons. So all tetrahedrons are selected here. No polygon meshes are selected here. So if you see the groups, the groups will be containing 992,000 primitives, exactly like the tetrahedron group here. same we'll create another group which will be called polygons and instead of tetrahedrons here we'll use polygons and the polygons group has 143,000 as the polygons field here so all right and now we'll be copying this guy over naming them naming them set collisions as whole and will create two other attributes that are internally used by FEM and they'll be applied to primitives this time and we're applying this attribute to all the primitives in the system, collision ID minus one and I will copy this guy over and name it exterior collision ID. So we're basically applying invalid values to all primitives in this attribute. So all the primitives will have this as a collision attribute, which means that collisions are basically disabled everywhere inside the mesh. And we'll duplicate this guy again, set collisions. we're going to re-enable all the collisions only for the, we'll leave just the exterior collision ID, but we'll set to zero. So basically zero means that the, the trahedrons will be colliding between themselves and the outside colliders. So, yeah. So it's also a primitive attribute. So let's see. Yeah, we don't have to visualize this because we know what happened here. So now, yeah, so the tets, yeah, the tets will be colliding with all the outside the trichegons and other objects. And now we'll be quite ready to connect this thing to the Femm solver. So let's create this Femm solver. The network in the first input goes our geometry. And in the second one, you'll see what goes here. We're gonna pull our animated Tetris and the time shift this guy, remove the channels, make a point before. And we're gonna pull the geometry that comes from the master transform here, connect the time shift to the second input and the animated live geo to the third input. Yeah, so it works. Our tetrahedral mesh is following the valium simulation. And what we're going to do here is work on the FemSolver first, and then we'll come back to this point before. I will actually connect the second input. So the first input is a static tetrahedral geometry with all the shenanigans and attributes and whatever. And the second one is the animated version of it. See the animated version and the static version. Now we'll go inside the Fem solver, but first we'll name it FemRinkles. Enable it, go inside, now let's create a Fem Hybrid object. Why a hybrid object? It's because we have a geometry containing polygons and tetrahedrons at the same time. We'll disable particle, we're in rod, we don't need these, but we'll use the shell and the solid, which makes sense because we have a shell, with a skin and the solid. The skin is the outer polygon layer, which is controlled by these attributes and the solid is the old tetrahedrons in the mesh that are controlled by these parameters. I'll change just a few parameters. The shape stiffness of our shell, let's say a thousand, and just for the beginning parameters, the shape stiffness doesn't really matter because this is the solid shape stiffness that we defined earlier. So this is, let's put 16, but it doesn't really matter. It's like a, it doesn't act as a multiplier, the value that we've set before. So, and the value stiffness can be 100. This will define the stiffness of the tetrahedrons contained inside the two skin layers that we created. As for the repulsion, we're going to set this repulsion at the highest possible limit. Let's say, I mean, it can go higher, but I'll leave it at this, and the friction at zero. I don't want to have any friction on contact. So now we go into the deformation tab, and in the deformation tab, we have to tell them which object is going to be deformed, which is the initial, basically, geometry that we are connecting. So let's type in a back tick, op, input path. Path. Two dots represent the, basically the current geometry. It goes into the solver. And because we're basically going back a level, if you can type a slash there, but it's not going to make a big difference. So we have this, let's connect it directly to the output. And we see that we are pulling the static geometry. is getting solved here. And we need a target deformation because we really want to deform this mesh by this guy. So this is the live geometry that is in the second input. So the second input has the index of one. So it's all good. Let's add the FEMS solver here. I'm going to think for a while, read the point before I realize there's movement there and we'll change the solve method to, we leave it to Gnl. The Gnl is a solver that just, yeah, the Gnl is a solver that is a little bit faster and needs less computation steps to solve stuff. I use Gnl, I don't, GSL has some limits, you can read about that in the documentation, but I'll also change the integration to the second order. This is another parameter that defines how things are solved inside of M, But yeah, I'll let you read about these more in the documentation, which is quite amazing if I may say so. Because, yeah, Houdini's documentation is top-notch compared to other software. So yeah, the Fepsolev's set. In the collisions here, we'll also check self-collide each component, connect component. So there will be internal collisions between the tetrahedrons of the skin and so on. We can disable this and it's going to be a little bit lighter, but you'll have penetrations. have penetrations. If there's a lot of deformation in your mesh, you'll have a lot of penetrations. We'll leave everything here, I'm not going to change anything. Just let it be like this and we'll add some gravity even if we don't really need a lot of it because our skin layer is already linked to the core layer and we don't... it's not gonna play a big role here except if you have scenarios where you want to have a lot of jiggly fat that is getting pulled by the gravity and you have hanging pieces like or sax or fat or organic tissue. Yeah, the gravity is going to play a big role there. But here we have a really tight skin around the an internal structure like the core layer. So let's also add a constraint, which is called the FemmFuse constraint. Allotted before the gravity. And the FemmFuse constraint is a special one. Because it will allow us to connect our skin to our core using the fusePID that we created earlier. So let's name our Fem object as a symbiote. Wrinkles just so we can reference it properly. We can have a bunch of Fem objects here merged together and solved by the same Fem solver. So this is why we have to name them properly because in the constraints like the Fem fuse constraint, you'll be able to recognize the objects in the list better. So we're gonna connect the symbiote wrinkles to symbiote wrinkles, but instead of using point groups, we'll be using a identifier point attribute. And by default, it's set to fuse BID, the attribute that we created earlier. So fuse BID here, fuse BID here. It's a good thing that they have two fields for the fuse BID because maybe you change the fuse BID into something else on the other piece of geometry and you can set two different attributes here. And the type of the constraint is gonna be hard. And when you use the identifier point attribute here, you don't have any guides showing. So these parameters are not going to work. Yeah, so we're kind of set here. The only thing left to do is let's create a blast right here before the Pintu animation. And what I'm going to do is select this arm in the middle. Let's do it again like this. Click enter, delete non-selected, and we're going to work on this single arm. You can see that there's two skin layers outside and one core layer here. the let's test if our simulation even works. I'll save this as a as this guy because this is the scene I was trying to record let's call it 8a. Oh yeah you didn't see the so basically I'm saving it as a this guy 8a saved now the scene is saved let's activate our solver and see if the wrinkles are even working because of the procedural nature of our software we can really clip just a piece of geometry and see if anything works here and as you can see already there's some wrinkles already in the beginning on the second frame we already have the wrinkles creating so you have a bunch of wrinkles on the arms and the detail of these wrinkles is defined by the remesh nodes up top they will define how detailed is the skin on those arms and yes it's quite cool. We have the system working straight away which is great we don't have to debug anything. In the previous attempt I had to debug a few things but now it's working fine so you see that when it compresses the arm creates a bunch of wrinkles and it's going to look good as in that octopus we've seen earlier so if you compare the 0.4 mesh with the wrinkled Fem mesh. You can see that it's solving the last frame. So yeah, this is the big difference between the two meshes. So let's do one thing. I'm going to disable this, but I'll show you what I'm going to do with this point before inversion. I will add a... Because my goal is what? Is to create a Wrinkle Simulations only on the arms. I will simulate the entire body with the wrinkles, although I can set the simulation in a way where only the arms are getting covered in wrinkles, by pitting to animation everything except the arms. But I prefer to have some control over the areas that are affected. So yeah, we'll create an attribute triangle here. And the attribute triangle will contain a rest attribute. This time is going to be a simple rest. And this rest is going to follow the animation. As you can see, there's a little clock here. It means that the rest changes at each frame. So I will attribute copy this rest attribute by ID. ID, rest. So the rest attribute from the point before the animated version that comes from Valem will be transferred to our Fem simulation after it's done. And we'll time-shift this guy. Set it to frame 1. And we have to do one more thing. We'll object merge our clean Quad Smash. And we'll do a quick attribute paint. a really quick hatch of paint. And so I'm going to paint only the arms. And the tail. I want some wrinkles there also. Although there's not going to be wrinkles. I already painted some kind of wrinkles here in ZBrush, but that's OK. Control Shift Drag just to decrease the size. And just limit myself here for the arms end. Yeah, that's not bad. Yeah, that's cool. So this is gonna be a mask, an attribute called mask. I'll attribute blur this guy. So my goal is to make a blend between the original value simulation and the wrinkled Fem simulation using this mask. So only the arms will be, we'll have this wrinkly thing. I'll attribute blur the mask just to see it. So yeah, we're maybe 32, yeah, let's say 32. Let's leave it at that. And I'll copy this master transformals here, just so we have the same transformation as the other stuff. And we will use this time-shifted the Femm-Rinkels simulation to attribute transfer, our mask, because we have a static mesh here and we have a static mesh here on the first frame. So this is how we do the attribute transfer. But afterwards, points, points. Afterwards, I'll be doing an attribute copy again. Back to the left geometry, back to the animated simulation. So attribute copy here, I'll call it mask. And we'll do some optimization attribute cast. and we'll actually cast these guys V velocity and mask. We downgrade it to 16 bit, save a little bit of space as we did before and make a clean right here. In the clean, I'll actually remove pretty much everything except V, ID, rest and mask. mask. And as for the groups, any group that contains the name, the word surface, I'll remove it. And now I'll just add a output node, I'll call it, I paste it from the previous attempt. I'll call it simfem, v0.1, simsimp.fem, v0.1, f5. And yeah, I will enable this wrinkle simulation and let's simulate a few frames. Save this project. I'll simulate like 10 frames and come back. So I sold a few frames here. Now we see that we have an ID attribute mask, rest and velocity and what I was saying when I said that I will blend the original mesh with the developed mesh. I wanted to show you how to do it. So yeah, so I simulated a few frames. As you can see, there's already some wrinkles here being created. This is the Femmesh that was starting to get wrinkly here. And if we compare it to the original mesh, you can see that there's no wrinkles on the arm. but here we have some wrinkles, but we have issue arising here in the bubbles. So this is why we painted the mask. I don't want to have any wrinkles on the bubbles. This is why we're going to get the tetrahedral mesh. In order to get the final mesh as we had it here, this is the animated clean mesh, but doesn't contain any wrinkles. So how do we do this? First, I want to blend the, the two versions of our simulation. So let's activate this, go to the frame 3 where we have this wrinkle thing and let's do something interesting here. Let's do a point deform. Let's copy the time shift and let's get our static quad mesh as we did before after the vellum sim. So we have this, oh yeah, the static quad mesh goes into the point point-to-form, the tension goes into our fem sim and the second input and the animate version into the third input. So we will subdivide this mesh once because we need some detail there and the settings for the point-to-form will be a little bit different. So 0, 0, 1, 5 and 5 because when you average the maximum points is 100 it's going to average a lot the deformation so we don't want So we'll set it to 5 in order for it to respect the wrinkles. So now we have wrinkles on our quad mesh, but only because we have the subdivide. Without the subdivide, we'll lose a lot of detail. When we add the subdivide, you'll obtain back, you'll just get the detail back. So, and we'll insert a little attribute triangle, which is the main trick here. That's a nice workflow for blending two versions of the simulation. because we did this whole thing with the rest attribute. So the simulation, baked simulation contains the rest, which is the animate position of the vellum simulation vertices or points. So basically this Femme has a version of the simulation which is not wrinkled in the rest attribute that we saved inside the final Femme simulation because of it, and because of the fact that we have the mask attribute that we painted on the static quad mesh and it transferred it through this workflow to the FEM simulation. Because of these two things, we can write this expression. VP equals lerp, VP, V rest, one minus mask. So what it does, it uses the mask to blend our wrinkly FEM simulation and the clean non-wrinkly valem simulation using the mask. So if you visualize the mask, where the red is, We have the Fem simulation where the blue is, it's developed. So using this version of the simulation, yeah, using this version, we're going to point to form our quad mesh and get this. So yeah, that's it. Does disable this mask thing? And you can see that we have a wrinkly sim here. Yeah, so one thing we'll add here is... We don't have any color here, so in order to get the color we can do maybe another thing because we got the color and the velocity from the villain sim, which is not a bad idea. So let's copy these guys here and let's connect this and our vellum sim that was animated. Yep. So we're basically copying the color and the velocity from the original vellum sim, which is not bad. and the vectors are pointing downwards because the mesh is floating onto the ground, which is correct. And we're actually bitcasting these guys the v, the n, and the cd. So we're saving a little bit of space here. And we'll basically get rid of the group delete. We'll remove all the groups for the asterisk. And we'll also maybe remove the mask because we don't need it anymore. Let's do it with the clean. So clean. I like this node because it's universal. But it considers all the primitives as degenerate. it so let's uncheck this. Yeah let's remove everything except a few like CD, CD, NP. Yeah, and the UV is also. We want to save those. But something happened on the way. Yeah, no, everything's fine. We're actually casting this thing. But yeah, I'm going to copy this output. just name it differently, it's going to be called like this, sim main clean, sim main clean quad rinkled high, and set the range to one to 300, we're simulating 300 frames in general. And yeah, the fem sim setup is done. I will simulate this whole thing and come back to you in the second lesson, and in the next lesson, about the flip simulation, which is going to be built on top of this rinkled simulation. So yeah, it looks cool. It's quite long, but I hope you understood the whole setup and how the workflow works with these kind of sims. Thanks for listening and watching. See you in the next one."
+    }
+  ]
+}