Add transcription for: frames/MDA_TheSymbioteCreature_DownloadPirate.com_SYMBIOTE_CH013_FLIP_RECAP_AND_MESHING_1080p_mp4_frames.zip

transcriptions/frames/MDA_TheSymbioteCreature_DownloadPirate.com_SYMBIOTE_CH013_FLIP_RECAP_AND_MESHING_1080p_mp4_frames_transcription.json

@@ -0,0 +1,8 @@
+{
+  "text": " Hey guys, here we are after the simulation. What I did here, I made a few adjustments in the BELIM simulation and then we simulated the FEM, we simulated the FLIP. So I made a few adjustments in the FLIP also, I'll show you what I've done, but these are two different simulations. The first one on the left is the one that I did earlier. As you can see the sacs, those bubbles are inflating less than on the right side. I like more when it's more exaggerated, like on the right side, so I feel like it's more interesting and more dramatic. You can see that the liquid reacts better to bubbles that are inflating more, and you see that it's kind of rolling off the body. And the colors of the liquid represent the variable viscosity of the flip simulation. So yeah, basically we have two two simulations now. The one on the right is the V2. So I'm pretty happy with this one on the right. Nothing, I think we'll continue with this one on the right. I'll show you what I've done with the flip and what I've done with the vellum to make the bubble inflate more. So let's go back to the vellum. We go into the solver here. To evaluate one frame and we'll go inside, see what's happening. So the only thing I changed here in the bubbles, constraint properties node. I just went into the max inflate channel and I switched it from three to eight. And this is why the bubbles are inflating dramatically, much more. So I resimulated the whole thing like this and then I made the wrinkles pass based on that value simulation. And I got the V2. I usually use a master control to use the version, like to change the versions, but this is a sim that doesn't, we don't have too many like iterations here to use that, so I'm comfortable using just overwriting the whole thing. So as you can see, the bubbles are inflating more. And if you look at the preview, even the left arm is bending much more because, oh yeah, I forgot to mention that in the vellum sim here in the animate contract, it was a thousand and I resimulated the 10,000, which increases the stiffness of those dead fiber constraints, and makes the dead fiber constraints contract more and pull more than previously. So the version on the right is where the creature moves much further than on the left because there's more contraction in the arms. This is why the left arm is bending much more and it's kind of twisting also. So the travel distance is longer and bigger. So it moves a little bit faster because of that higher contraction of the muscles. So yeah, when I was done with this vellum, and it was baked into this Vellum I.O. If you look here, we had a V2. Maybe, oh yeah, because I saved it there. But yeah, the original Vellum was this one. And if you want to see the difference, so basically this was the previous SIM. You see, it moves it a little bit further than previously, so if you compare the two you'll see that the previous version doesn't move that far. and the newer version where there's more contraction and bigger bubble inflation multipliers, it moves much further. And the wrinkles are more dramatic here. So yeah, so this is the version with more subdivisions. This is the one that I... like this one has the subdivision level 1, the one here, and I'm using it for rendering. So this is going to be the body which I'll render and apply a displacement map to but this one the low-poly version without the sub-divide I used it for the flip simulation so it went through the time warp as I described before I went through this VDB to polygons I baked it so we can read it really fast during the sim so you see that VDB is really moving here a lot and also the big change that I did here I inserted a measured curvature node here because this node contains more parameters that allow me to control better the curvy in convexity. The curvature here is the output of this node is better than the VDB analysis that I used before. So what I did here in the emission magic node, I basically used the concavity instead of the volume sample that we had before which used the second input the VDB analysis. I used the concavity node and I fit it. I just inverted the values and I made a ramp here as you can see here and it's more refined I think and you see the white areas if you disable the lighting you see that the white is not as present as the previous volume sample because if you put the volume sample you see it covers a bigger area so the emission area is bigger than than I need because if we leave it, I can't really control this area with any ramps or it doesn't shrink it too much when I use a ramp on it. So this is why I use this alternative for the curvature. So this is the measure curvature node from labs, from Houdini labs, it's a great node and uses an internal measure node that calculates the curvature. So yeah, so this was combined with the emit attribute, which is the selection of four faces pointing up. If you remember, we did this emit attribute, which is this guy, and it was blurred a little, and we combine it internally with the concavity. It's multiplied by the concavity, and it's multiplied also by a animated curl noise. So the output, you can see the output here in the emission magic. So this is the output map, And it's really not a big area of emission. Like if you consider that the emission will happen like at each frame, even a small area like this, a white area like this will emit a lot of particles over time. So this is why you have to shrink it as much as we can just to leave it as faint as possible. And afterwards I blur it by one iteration, I copy it to the moving mesh and the moving mesh has the same kind of attribute. Then we scatter some points and then we push them out. As you remember, we push them by normals just so we emit from a little distance from the surface. We convert them to VDBs. And this VDB then is baked into a sequence. So you have a baked sequence here. If we switch to the V2 of the sequence. Yeah. So and then we use points from volume to scatter points around with the point separation, which is equal to the point separation inside the flip object. So this is the final point separation I used. And still, it yields a lot of particles, like 7 million particles, which is kind of overkill for this kind of sim. But still, it yields a lot of detail. So our goal now is to measure this thing. I baked the flip. So what I baked here, I cleaned up the entire sim because it saves a lot of parameters, like attributes that we don't need. And I end up with three attributes, position the last thing in viscosity. And just as a reminder, in the emission magic, we saved the viscosity as an attribute, but it was remapped from 0 to 1. 0 to 1 was the original range, and then it was fitted to the values that are here, viscosity minimum, viscosity maximum. So let's copy these guys. copy the minimum value, go into color here, paste it into the relative reference for the minimum and then copy this guy, the max value color and paste this as a relative reference so it's mapped between 1 and 3, so the colors are correct yeah, so this is a great setting for the previous, but let me duplicate this guy and make it grayscale So this is the color that we're going to use for meshing, and we're going to transfer this color to the mesh. So let's lay down a particle fluid mesh, fluid surface, and let's connect it here. And yeah, as you can see, the parameters are completely variable. So let's skip this, the transfer attributes for now. And the rule of thumb for me in these kind of simulations is that I used a point separation of 0,0,0,15. And this is the parameter that I'm going to use as a reference in the particle. Something like this. So you can see that it meshes our thing here. Let's drag this a little bit further in the timeline, see what it does. So it can make some mesh here. But this is just, I just changed this guy to, yeah, so this is the fluid that we get right now. And we're going to use the filtering. So instead of experimenting with the meshing over the whole, the entire mesh, I'm going to just use regions here. And I'll use the bounding box, maybe not the bounding, yeah, bounding box, but make it, I'll disable this node and set the parameters first. I usually, yeah, so click enter when the node is selected, press enter and then we'll just shrink down this bounding box. So yeah, let's enable all of these guys. I usually use all of them. And also, you need compressed fluid, so we don't need that because we're not using oceans. The adaptivity may be 0, 0, 0, 1, and for the transfer attributes V and CD. And the attribute radius 2. And I usually, like for the particle separation in the particle fluid surface node, I usually use a smaller value than the particle separation used in the flip. So let's put it 0, 0, 0, 1. Let's compare it to the, let's template this liquid surface. What I like about this liquid, you can see that there's a strands of liquid flowing down. And what I did, I added some additional parameters here in the flip to achieve the liquid behavior. In the flip, I added this gas surface tension, like micro solver. It's already kind of included in the flip solver nowadays. So if you go into the surface tension, this is it. but I like to have it as a separate microsolver. So visually it makes more sense for me. So I used the default parameters except the surface tension itself. I set it to zero, it was zero 15 before, but I set it to zero two just for the final. I want the bubbles to create like a lot of surface tension when the liquid is on the ground. And another thing is for the collisions, I brought in the, like if you remove this and you leave it without, like you leave this field empty, what's going to happen is the flip simulation will not know anything about the velocity of this object. There will be collisions happening using this volume representation of the collision object, but the collisions will be kind of in place, so the liquid will slide over the VDB. It's not going to flow around the moving surface, so it's going to be weird. So you have to bring in the geometry representation with the velocity vectors into the soft path right here. So and also when you check the use deforming geometry and revaluate revaluate sops to interpolate geometry, it's going to use those interpolated frames in between frames in order to calculate collisions during fast moving collisions. I mean, calculate collisions during fast moving, you know, simulations. So this is really important, the geometry that you put here in the sub-pad. And what else? Oh yeah, and the flip solver, we, I checked in the use open CL and self pressure with that activity. The use open CL is going to accelerate the calculations using the GPU. And also in the viscosity, which that's important, really important, in the flow precision, I lowered from 64 to 32. This will, like the 32 bit precision for viscosity solving is enough. There's only enough precision here. So it accelerates the calculation of viscous simulations. So this is also important. And what else? In the sub steps, I use two and four. Because we're using viscosity, I want the viscosity to be solved properly. So usually when you have introduced this question to SIM, you have fast moving colliders like this, you need to increase these sub steps. Because sometimes the default sub steps of one and two, minimum one and maximum two, sub steps are not enough to satisfy the viscosity requirements and the liquid is not viscous enough, even if the viscosity is high. So you need to increase those until you get the viscosity you need. So this is like a question of experimenting and trying. But yeah, I use these ones. So yeah, this is why we get that behavior as you've seen here. You see that there's a lot of rolling liquid and you have some bubbles kind of rolling over themselves because there's a strong surface tension here. It's an interesting effect. So we'll use that viscosity to transfer it to the fluid surface. So let's go here somewhere and calculate. Let's find a frame where there's some intersection. And I'll template the... You see that the attribute transfer is kind of terrible because we don't have enough sample here. 50. Let's increase the attribute radius and the add-up to 25. The add-up to 25 may be too much, so it may yield some flickering in the mesh, especially with viscosity viscous particles. Also in the flip sim, when you're doing viscous simulations, going into the particle motion into the receding and uncheck the recede particles. The recede particles will create particles where the flip solver thinks there's empty spaces, so it's going to generate new particles and it's quite difficult to track those and there's going to be popping in the simulation, especially when you use some viscous particles and you'll see that they appear from nowhere, especially when when it's a really slow moving viscous particle. So yeah, I turn off the reseeding for viscous simulations. So you see that the relaxation of the mesh is quite high. So let's keep improving this. Let me increase this lower. Oh, zero, zero, like three zeros. And we have a mesh that gets closer to our, and in the filtering maybe for the final filtering, use one and maybe three in between because what happens here the default workflow before when we didn't have this node a particle fluid surface we had to do VDB reshape nodes so basically you dilate the VDB you smooth it somehow with a VDB smooth SDF and then you shrink it back to the original size and then you apply a final smooth you can use also masks but this works well when you have really calm surfaces and there's like a shark breaching the surface and you want to calm down the flat surface around the interaction area. So I'm not going to use that here. And we can also subtract our VDB. So in the collision object and volumes you can connect your VDB, the collision VDB, and subtract collision volumes. So basically it will remove. Let me see if it's... There's penetration here. Maybe because the VDB that we used is a little bit too... If I shrink it, VDB reshape as the F. We erode it. Just by a little notch. I think we'll not use it because it's going to eat out a little too much of our mesh and if we connect these guys together There's some particles going inside the Yeah, so what we can do is push them out of the mesh, which is not that hard we can do that Let's make a little VOP that does this so attribute VOP We'll use This guy, so basically this black and white colored guy This was for the preview purposes and this one is going to be for our final Mesh and in the attribute VOP you connect the collision volume and we want the collision volume in the particles to respect the Collision volume, so let's volume sample Volume gradient So the volume sample gets us let's select both of them go set select the second input and the sample position goes into position and what I want to do is push out our particles that are going below zero so they have negative values inside the volume so let's compare volume sample compare less than zero and we'll make a two-way switch this is a classical push out from the volume setup so it's a two-way switch and if the condition is true if the particle is going underneath like inside the volume. We want to use a multiply here and I want to multiply constant maybe invert the vector and multiply this by the volume sample and the two-way switch is using condition input 1 if the condition is true. So and if the condition is false we're just using the original position and we're connecting this to the position. So let's check what's happening here. Some particles go crazy. So let's see. The VDB is here. Instead of the inverse, let me multiply constant and then trust the invert minus one. So, yes, so the particles that are inside are kind of going crazy and they disappear. And let's connect this compared to, so now you see that the volume sample is compared to zero, so it's less than. You can see that the white particles are inside the volume. So I really want to push them, let's add some normals using the gradient from file. Yeah, so let's display the normals. So press D, go into the visualizer and into the guides and the scale normal will just set it to... So you can see that it displays the normals, the volume gradient as a normal now, and it looks okay. Just there's an issue with the volume sampling. So we multiply the inverse gradient by the volume sample. Yeah, and changing the position. If we invert these guys, yeah. So, I'll be right back. So yeah, because the volume gradient is a, we should add the volume gradient multiplied by the volume sample or the inverse volume gradient to the volume sample and add it to the current position. I don't know what I was thinking about. This is what we're connecting to our first input. So now, if you check properly, you can see that this was before, now this is after. So all the particles that are inside the volume, they get pushed out of the... So you can see that. This is before and this is after. So let's remove this and we're good to go. Now, our simulation obeys the respects more the surface. This happens when you don't have enough sub steps. So like if I was increasing the sub steps at the flip sim level, like at the dope network level, This would calculate, will subdivide even more the sub steps that we had in the flip solver. So it would yield more precision because we have really fast moving arms here and they kind of throw the liquid really far. If you look, you see in the beginning, the liquid is really thrown through the, into the direction of movement. So yeah, this is why we may probably need to increase the sub step, but we can work with the attribute valve that fixes the issue. Particles push out. And yeah, we'll go back to our, we'll remove the VDB reshape. And in the particle fluid surface, let's see what happens. Let's remove the normals. And also let's also remove the bounding box. Let's calculate the entire mesh just to see how it looks like. It'll take just a little, few seconds. This is the mesh that we have now. This black and white attribute will use it in shading because we will need to apply a color. Let's merge this together with the... Instead of this particle stream, I'll use the mesh just so I could... Yeah, so maybe there's a little too much space between the mesh and the object, I mean, the creature. But I think if we go back here in the subtract, I'll set the collision offset to zero just so I would remove the gap more. So yeah, you can see the difference now. Like if you put a little gap here, you'll see that it's out the inside of the mesh and it subtracts more than we need. So I'll set it to zero. I don't want to do any kind of... I might even... Let's do it in a negative way. See what it does. And if there's a difference... Yeah, there's a little difference, but let's keep it at zero. It's fine. Keep in mind that we're going to use some displacement at the render time. And this displacement will push out the mesh a little to the outside. So we'll have that gap filled with displaced polygons. So that's not a big deal. What I like about the SIM is that we have these bubbles forming. And they will inherit that color from the viscosity attribute. So yeah, this is going to be cool to render. So yeah, the mesh is, I think the mesh, the meshing is done here. Let's see how many polygons are there. And it's a polysoup. And I don't like polysoups because they're not really used too much. And I prefer to use. And also because octane renderer that I'm going to use for rendering this doesn't like polysups. Usually it converts it back to polygons. And if I remember correctly, the polysups are quite, they're a little bit faster to read and manage, but yeah, some renders are having a hard time with the polysups. I'll use the surface polygons here. Nothing is gonna change visually, but it's going to change the way the renderer uses the mesh. Yeah, so it takes a little bit of time. it takes 17 seconds for this mesh to calculate. So the meshing is alright, let's duplicate this guy. And I'm going to mesh this, rename this, same flip, mesh, and keep the version as version 2. Oh, that's interesting. Oh yeah, I love when that happens. I love when that happens. So let me reload this desktop. We have a nice setup here, all split by colors. I'll save this as a flip-sim-mesh. Chapter 11. Yeah, so yeah, that's cool. I'll save that mesh and come back to you and we'll start rendering this puppy. So yeah, see you in the next one guys.",
+  "segments": [
+    {
+      "text": " Hey guys, here we are after the simulation. What I did here, I made a few adjustments in the BELIM simulation and then we simulated the FEM, we simulated the FLIP. So I made a few adjustments in the FLIP also, I'll show you what I've done, but these are two different simulations. The first one on the left is the one that I did earlier. As you can see the sacs, those bubbles are inflating less than on the right side. I like more when it's more exaggerated, like on the right side, so I feel like it's more interesting and more dramatic. You can see that the liquid reacts better to bubbles that are inflating more, and you see that it's kind of rolling off the body. And the colors of the liquid represent the variable viscosity of the flip simulation. So yeah, basically we have two two simulations now. The one on the right is the V2. So I'm pretty happy with this one on the right. Nothing, I think we'll continue with this one on the right. I'll show you what I've done with the flip and what I've done with the vellum to make the bubble inflate more. So let's go back to the vellum. We go into the solver here. To evaluate one frame and we'll go inside, see what's happening. So the only thing I changed here in the bubbles, constraint properties node. I just went into the max inflate channel and I switched it from three to eight. And this is why the bubbles are inflating dramatically, much more. So I resimulated the whole thing like this and then I made the wrinkles pass based on that value simulation. And I got the V2. I usually use a master control to use the version, like to change the versions, but this is a sim that doesn't, we don't have too many like iterations here to use that, so I'm comfortable using just overwriting the whole thing. So as you can see, the bubbles are inflating more. And if you look at the preview, even the left arm is bending much more because, oh yeah, I forgot to mention that in the vellum sim here in the animate contract, it was a thousand and I resimulated the 10,000, which increases the stiffness of those dead fiber constraints, and makes the dead fiber constraints contract more and pull more than previously. So the version on the right is where the creature moves much further than on the left because there's more contraction in the arms. This is why the left arm is bending much more and it's kind of twisting also. So the travel distance is longer and bigger. So it moves a little bit faster because of that higher contraction of the muscles. So yeah, when I was done with this vellum, and it was baked into this Vellum I.O. If you look here, we had a V2. Maybe, oh yeah, because I saved it there. But yeah, the original Vellum was this one. And if you want to see the difference, so basically this was the previous SIM. You see, it moves it a little bit further than previously, so if you compare the two you'll see that the previous version doesn't move that far. and the newer version where there's more contraction and bigger bubble inflation multipliers, it moves much further. And the wrinkles are more dramatic here. So yeah, so this is the version with more subdivisions. This is the one that I... like this one has the subdivision level 1, the one here, and I'm using it for rendering. So this is going to be the body which I'll render and apply a displacement map to but this one the low-poly version without the sub-divide I used it for the flip simulation so it went through the time warp as I described before I went through this VDB to polygons I baked it so we can read it really fast during the sim so you see that VDB is really moving here a lot and also the big change that I did here I inserted a measured curvature node here because this node contains more parameters that allow me to control better the curvy in convexity. The curvature here is the output of this node is better than the VDB analysis that I used before. So what I did here in the emission magic node, I basically used the concavity instead of the volume sample that we had before which used the second input the VDB analysis. I used the concavity node and I fit it. I just inverted the values and I made a ramp here as you can see here and it's more refined I think and you see the white areas if you disable the lighting you see that the white is not as present as the previous volume sample because if you put the volume sample you see it covers a bigger area so the emission area is bigger than than I need because if we leave it, I can't really control this area with any ramps or it doesn't shrink it too much when I use a ramp on it. So this is why I use this alternative for the curvature. So this is the measure curvature node from labs, from Houdini labs, it's a great node and uses an internal measure node that calculates the curvature. So yeah, so this was combined with the emit attribute, which is the selection of four faces pointing up. If you remember, we did this emit attribute, which is this guy, and it was blurred a little, and we combine it internally with the concavity. It's multiplied by the concavity, and it's multiplied also by a animated curl noise. So the output, you can see the output here in the emission magic. So this is the output map, And it's really not a big area of emission. Like if you consider that the emission will happen like at each frame, even a small area like this, a white area like this will emit a lot of particles over time. So this is why you have to shrink it as much as we can just to leave it as faint as possible. And afterwards I blur it by one iteration, I copy it to the moving mesh and the moving mesh has the same kind of attribute. Then we scatter some points and then we push them out. As you remember, we push them by normals just so we emit from a little distance from the surface. We convert them to VDBs. And this VDB then is baked into a sequence. So you have a baked sequence here. If we switch to the V2 of the sequence. Yeah. So and then we use points from volume to scatter points around with the point separation, which is equal to the point separation inside the flip object. So this is the final point separation I used. And still, it yields a lot of particles, like 7 million particles, which is kind of overkill for this kind of sim. But still, it yields a lot of detail. So our goal now is to measure this thing. I baked the flip. So what I baked here, I cleaned up the entire sim because it saves a lot of parameters, like attributes that we don't need. And I end up with three attributes, position the last thing in viscosity. And just as a reminder, in the emission magic, we saved the viscosity as an attribute, but it was remapped from 0 to 1. 0 to 1 was the original range, and then it was fitted to the values that are here, viscosity minimum, viscosity maximum. So let's copy these guys. copy the minimum value, go into color here, paste it into the relative reference for the minimum and then copy this guy, the max value color and paste this as a relative reference so it's mapped between 1 and 3, so the colors are correct yeah, so this is a great setting for the previous, but let me duplicate this guy and make it grayscale So this is the color that we're going to use for meshing, and we're going to transfer this color to the mesh. So let's lay down a particle fluid mesh, fluid surface, and let's connect it here. And yeah, as you can see, the parameters are completely variable. So let's skip this, the transfer attributes for now. And the rule of thumb for me in these kind of simulations is that I used a point separation of 0,0,0,15. And this is the parameter that I'm going to use as a reference in the particle. Something like this. So you can see that it meshes our thing here. Let's drag this a little bit further in the timeline, see what it does. So it can make some mesh here. But this is just, I just changed this guy to, yeah, so this is the fluid that we get right now. And we're going to use the filtering. So instead of experimenting with the meshing over the whole, the entire mesh, I'm going to just use regions here. And I'll use the bounding box, maybe not the bounding, yeah, bounding box, but make it, I'll disable this node and set the parameters first. I usually, yeah, so click enter when the node is selected, press enter and then we'll just shrink down this bounding box. So yeah, let's enable all of these guys. I usually use all of them. And also, you need compressed fluid, so we don't need that because we're not using oceans. The adaptivity may be 0, 0, 0, 1, and for the transfer attributes V and CD. And the attribute radius 2. And I usually, like for the particle separation in the particle fluid surface node, I usually use a smaller value than the particle separation used in the flip. So let's put it 0, 0, 0, 1. Let's compare it to the, let's template this liquid surface. What I like about this liquid, you can see that there's a strands of liquid flowing down. And what I did, I added some additional parameters here in the flip to achieve the liquid behavior. In the flip, I added this gas surface tension, like micro solver. It's already kind of included in the flip solver nowadays. So if you go into the surface tension, this is it. but I like to have it as a separate microsolver. So visually it makes more sense for me. So I used the default parameters except the surface tension itself. I set it to zero, it was zero 15 before, but I set it to zero two just for the final. I want the bubbles to create like a lot of surface tension when the liquid is on the ground. And another thing is for the collisions, I brought in the, like if you remove this and you leave it without, like you leave this field empty, what's going to happen is the flip simulation will not know anything about the velocity of this object. There will be collisions happening using this volume representation of the collision object, but the collisions will be kind of in place, so the liquid will slide over the VDB. It's not going to flow around the moving surface, so it's going to be weird. So you have to bring in the geometry representation with the velocity vectors into the soft path right here. So and also when you check the use deforming geometry and revaluate revaluate sops to interpolate geometry, it's going to use those interpolated frames in between frames in order to calculate collisions during fast moving collisions. I mean, calculate collisions during fast moving, you know, simulations. So this is really important, the geometry that you put here in the sub-pad. And what else? Oh yeah, and the flip solver, we, I checked in the use open CL and self pressure with that activity. The use open CL is going to accelerate the calculations using the GPU. And also in the viscosity, which that's important, really important, in the flow precision, I lowered from 64 to 32. This will, like the 32 bit precision for viscosity solving is enough. There's only enough precision here. So it accelerates the calculation of viscous simulations. So this is also important. And what else? In the sub steps, I use two and four. Because we're using viscosity, I want the viscosity to be solved properly. So usually when you have introduced this question to SIM, you have fast moving colliders like this, you need to increase these sub steps. Because sometimes the default sub steps of one and two, minimum one and maximum two, sub steps are not enough to satisfy the viscosity requirements and the liquid is not viscous enough, even if the viscosity is high. So you need to increase those until you get the viscosity you need. So this is like a question of experimenting and trying. But yeah, I use these ones. So yeah, this is why we get that behavior as you've seen here. You see that there's a lot of rolling liquid and you have some bubbles kind of rolling over themselves because there's a strong surface tension here. It's an interesting effect. So we'll use that viscosity to transfer it to the fluid surface. So let's go here somewhere and calculate. Let's find a frame where there's some intersection. And I'll template the... You see that the attribute transfer is kind of terrible because we don't have enough sample here. 50. Let's increase the attribute radius and the add-up to 25. The add-up to 25 may be too much, so it may yield some flickering in the mesh, especially with viscosity viscous particles. Also in the flip sim, when you're doing viscous simulations, going into the particle motion into the receding and uncheck the recede particles. The recede particles will create particles where the flip solver thinks there's empty spaces, so it's going to generate new particles and it's quite difficult to track those and there's going to be popping in the simulation, especially when you use some viscous particles and you'll see that they appear from nowhere, especially when when it's a really slow moving viscous particle. So yeah, I turn off the reseeding for viscous simulations. So you see that the relaxation of the mesh is quite high. So let's keep improving this. Let me increase this lower. Oh, zero, zero, like three zeros. And we have a mesh that gets closer to our, and in the filtering maybe for the final filtering, use one and maybe three in between because what happens here the default workflow before when we didn't have this node a particle fluid surface we had to do VDB reshape nodes so basically you dilate the VDB you smooth it somehow with a VDB smooth SDF and then you shrink it back to the original size and then you apply a final smooth you can use also masks but this works well when you have really calm surfaces and there's like a shark breaching the surface and you want to calm down the flat surface around the interaction area. So I'm not going to use that here. And we can also subtract our VDB. So in the collision object and volumes you can connect your VDB, the collision VDB, and subtract collision volumes. So basically it will remove. Let me see if it's... There's penetration here. Maybe because the VDB that we used is a little bit too... If I shrink it, VDB reshape as the F. We erode it. Just by a little notch. I think we'll not use it because it's going to eat out a little too much of our mesh and if we connect these guys together There's some particles going inside the Yeah, so what we can do is push them out of the mesh, which is not that hard we can do that Let's make a little VOP that does this so attribute VOP We'll use This guy, so basically this black and white colored guy This was for the preview purposes and this one is going to be for our final Mesh and in the attribute VOP you connect the collision volume and we want the collision volume in the particles to respect the Collision volume, so let's volume sample Volume gradient So the volume sample gets us let's select both of them go set select the second input and the sample position goes into position and what I want to do is push out our particles that are going below zero so they have negative values inside the volume so let's compare volume sample compare less than zero and we'll make a two-way switch this is a classical push out from the volume setup so it's a two-way switch and if the condition is true if the particle is going underneath like inside the volume. We want to use a multiply here and I want to multiply constant maybe invert the vector and multiply this by the volume sample and the two-way switch is using condition input 1 if the condition is true. So and if the condition is false we're just using the original position and we're connecting this to the position. So let's check what's happening here. Some particles go crazy. So let's see. The VDB is here. Instead of the inverse, let me multiply constant and then trust the invert minus one. So, yes, so the particles that are inside are kind of going crazy and they disappear. And let's connect this compared to, so now you see that the volume sample is compared to zero, so it's less than. You can see that the white particles are inside the volume. So I really want to push them, let's add some normals using the gradient from file. Yeah, so let's display the normals. So press D, go into the visualizer and into the guides and the scale normal will just set it to... So you can see that it displays the normals, the volume gradient as a normal now, and it looks okay. Just there's an issue with the volume sampling. So we multiply the inverse gradient by the volume sample. Yeah, and changing the position. If we invert these guys, yeah. So, I'll be right back. So yeah, because the volume gradient is a, we should add the volume gradient multiplied by the volume sample or the inverse volume gradient to the volume sample and add it to the current position. I don't know what I was thinking about. This is what we're connecting to our first input. So now, if you check properly, you can see that this was before, now this is after. So all the particles that are inside the volume, they get pushed out of the... So you can see that. This is before and this is after. So let's remove this and we're good to go. Now, our simulation obeys the respects more the surface. This happens when you don't have enough sub steps. So like if I was increasing the sub steps at the flip sim level, like at the dope network level, This would calculate, will subdivide even more the sub steps that we had in the flip solver. So it would yield more precision because we have really fast moving arms here and they kind of throw the liquid really far. If you look, you see in the beginning, the liquid is really thrown through the, into the direction of movement. So yeah, this is why we may probably need to increase the sub step, but we can work with the attribute valve that fixes the issue. Particles push out. And yeah, we'll go back to our, we'll remove the VDB reshape. And in the particle fluid surface, let's see what happens. Let's remove the normals. And also let's also remove the bounding box. Let's calculate the entire mesh just to see how it looks like. It'll take just a little, few seconds. This is the mesh that we have now. This black and white attribute will use it in shading because we will need to apply a color. Let's merge this together with the... Instead of this particle stream, I'll use the mesh just so I could... Yeah, so maybe there's a little too much space between the mesh and the object, I mean, the creature. But I think if we go back here in the subtract, I'll set the collision offset to zero just so I would remove the gap more. So yeah, you can see the difference now. Like if you put a little gap here, you'll see that it's out the inside of the mesh and it subtracts more than we need. So I'll set it to zero. I don't want to do any kind of... I might even... Let's do it in a negative way. See what it does. And if there's a difference... Yeah, there's a little difference, but let's keep it at zero. It's fine. Keep in mind that we're going to use some displacement at the render time. And this displacement will push out the mesh a little to the outside. So we'll have that gap filled with displaced polygons. So that's not a big deal. What I like about the SIM is that we have these bubbles forming. And they will inherit that color from the viscosity attribute. So yeah, this is going to be cool to render. So yeah, the mesh is, I think the mesh, the meshing is done here. Let's see how many polygons are there. And it's a polysoup. And I don't like polysoups because they're not really used too much. And I prefer to use. And also because octane renderer that I'm going to use for rendering this doesn't like polysups. Usually it converts it back to polygons. And if I remember correctly, the polysups are quite, they're a little bit faster to read and manage, but yeah, some renders are having a hard time with the polysups. I'll use the surface polygons here. Nothing is gonna change visually, but it's going to change the way the renderer uses the mesh. Yeah, so it takes a little bit of time. it takes 17 seconds for this mesh to calculate. So the meshing is alright, let's duplicate this guy. And I'm going to mesh this, rename this, same flip, mesh, and keep the version as version 2. Oh, that's interesting. Oh yeah, I love when that happens. I love when that happens. So let me reload this desktop. We have a nice setup here, all split by colors. I'll save this as a flip-sim-mesh. Chapter 11. Yeah, so yeah, that's cool. I'll save that mesh and come back to you and we'll start rendering this puppy. So yeah, see you in the next one guys."
+    }
+  ]
+}