Add transcription for: frames/MDA_TheSymbioteCreature_DownloadPirate.com_SYMBIOTE_CH017_IMPROVEMENTS_AND_REMESHING_1080p_mp4_frames.zip

transcriptions/frames/MDA_TheSymbioteCreature_DownloadPirate.com_SYMBIOTE_CH017_IMPROVEMENTS_AND_REMESHING_1080p_mp4_frames_transcription.json

@@ -0,0 +1,8 @@
+{
+  "text": " Hello guys, so this is going to be an improvement and tweaking lesson for our simulation and our machine. I've done a few changes here just to improve the overall sim and to minimize the flickering issue on the rendering stage. So I'll show what I've done. I really minimized the emission area by shrinking down the curvature with the RAM. And then what I also did is rebake the whole source. You see that it's really subtle, it's not like before. I could have left more particles for the emission, have more particles in the end. But this is enough for now. The only thing I did is think ahead for the simulation this time because what I did last time in my original simulation is using sub steps. And I didn't have any file caching nodes. So basically, if we ignore them like this, I went back and reset the sub steps to 1 and 2 in the solver. Because I realized that I need to sub step my collision sources and the volume source as well. So in order to do this, I actually just turned off the flip solver sub steps. I mean, I went back to the default ones. And I increased the overall dot network sub steps to 4. So I prefer to use these ones when we talk about a simulation like this. We have a really close up camera. And we need to see that the fluid is really sitting on the surface. But as you remember, if we look at our simulation, it's really fast moving, fast inflating. There's a lot of changes happening over a few frames. So we need proper sub-stepping. So the only thing that sub steps properly here is the geometry, the SOP that we have here. If we switch to it, you'll see that we have the time warp connected to this node connected to the time warp. So we're live interpolating this geometry source. But what we're not doing by baking this VDB is basically what we're not doing here is interpolating this VDB along with the geometry. So in order to do this, to do the proper interpolation, we have to save our VDBs and our sourcing in an interpolated manner. So the file cache nodes in the Houdini, I think 17 or 18 were changed and adjusted just so you could cache sub steps. So, if you're properly in the flip-dop, I set four sub-steps here, and I set the same in the file cache nodes, but the problem with this is the following. So basically, if we do another file cache here, I'll make another folder here, and I'll remove what I've done here, I'll come back to this. Let's say I set it to ff. for saving the sub steps you wouldn't use the f the dollar f expression you would use the ff because it represents the fractional frames and let's see what this does like five frames for example let's make two or three frames and I'll save it with four sub steps so what will happen here is we'll have this saving window and you see that it's rendering frame 1.5, then 175, and then the second frame, then 225. It divides every frame by 4, so you have increments of 0.25 frames every time. So for 3 frames it saved 9. So if we open up the folder, oh that actually worked. That's interesting because previously, I don't know if it's like a bug or something else, but previously when I was doing this fractional frame saving, it was actually giving me really terrible a bunch of decimal point numbers like this and the precision error at the end. So this could happen, really this could happen with a lot of these SOPs that are set in interpolation mode saving but in order to avoid this if you ever have this problem where it just yields you fractional frames with like 28 decimal points after the comma like if you have that issue you could fix it by using this expression so this is a Python expression that properly evaluates subframes and basically it searches for the current frame of the current node and it evaluates the this frame as an F3, I suppose, it's like a with three decimal points afterwards, but I'm not sure about this one, because I grabbed it from the Houdini forums and usually it, nobody gave an explanation, but this specific expression works really well and this is what I did here in the saving string. So after the name of your file, you insert this expression with the backticks. So if you third mouse click, if you mouse will click on the geometry file here, it will evaluate the name of your file. So if you go into enable playback at fractional frames here and you right click on this and you set it to four because we're at four sub steps, you will see that we're evaluating right now at 66.5 and if you go into the name over file here it evaluates as 66.5. So yeah this expression is life-saving if you have that issue with fractional frames but yeah apparently on this SOP or newly created SOP it works properly which is really interesting but yeah so I saved both the VDB that I had. That's interesting, apparently. We didn't save properly. I have another version that was saved properly like this. Let me find these fractional frames. It apparently didn't save it properly. Anyway, usually you would save the source, it saved the same way with four sub steps, with the spikin expression. And you can see that it evaluates at every fourth of a frame property. So if you step slowly, you would see that it's kind of flickering. But yeah, and we have four sub steps overall on the.NETROOK. Whenever you have fast moving colliders like this, fast moving sources, we would need to interpolate and save that with the sub steps. So we have the VDB of the geometry saved with sub steps, because it comes from the time warp. And then we have the geometry. This guy is interpolated live because it comes from the time warp. And it's sampled inside the collision source. I mean, not the collision source, but the collider in the static solver here. So it's using the deforming geometry and re-evaluating the subs to interpolate geometry. And the third is the source, which is interpolated also with four subs steps. and is sourced by the volume source here. So yeah, you have to match these sub steps that you have on the dot network in order to get the proper interpolation for this kind of simulation. So what else? I didn't change much afterwards. I just pushed out the particles after it was simulated. So yeah, the particles are here. There's some particles flowing down on the body. I did the UV texture thing with the camera optimization. So it removes clips on some of those particles. And in the particle fluid surface, I changed the method to spherical. Sometimes it works better for viscous fluids, especially when you have bubbles that you want to preserve inside the volume. But this affects more the flip tanks, where you have bubble paths. I mean not a bubble, but when you have internal bubbles forming, because an object fell into the volume of liquid and it creates like a narrow bubble after the collision. So this function really saves those bubbles. Usually if you use the average position, it will kill the bubbles. It will try to fill it up as fast as it can. So yeah, and yeah, I changed some parameters here. The attribute radius to 10 samples to 35. To better transfer the color. I cleaned up the mesh, and actually casted these guys and saved the mesh. So the mesh is like this. So yeah, this was transferred to my... this color is used by the shader. As you remember in the symbiote fluid you have the shader driving the blend between two colors. Two colors for the transmission and two colors for the roughness of the specular layer. So there's some areas that are less translucent and less refractive than the other ones. So this is cool. And then in the Octane Rop I turned off the Motion Blur and now I'll put some other passes, I think. Shading normal, UVs, Motion Vectors, just in case we can use it in comp with the real smart Motion Blur in After Effects. But yeah, that's about it. So the goal of this tutorial was to make something a little bit more unique than what you have on the market nowadays because I really haven't seen any tutorials concerning these soft-body simulations, especially value used for creatures and whatnot. But yeah, this is the type of tutorials that I aim to make in the future, and it all depends on how this one is going to work. But yeah, I enjoyed going over this one, and I hope you enjoyed it. And also learned something from the 3D part of this tutorial. and yeah let's go to the compositing stage. See you in the next one.",
+  "segments": [
+    {
+      "text": " Hello guys, so this is going to be an improvement and tweaking lesson for our simulation and our machine. I've done a few changes here just to improve the overall sim and to minimize the flickering issue on the rendering stage. So I'll show what I've done. I really minimized the emission area by shrinking down the curvature with the RAM. And then what I also did is rebake the whole source. You see that it's really subtle, it's not like before. I could have left more particles for the emission, have more particles in the end. But this is enough for now. The only thing I did is think ahead for the simulation this time because what I did last time in my original simulation is using sub steps. And I didn't have any file caching nodes. So basically, if we ignore them like this, I went back and reset the sub steps to 1 and 2 in the solver. Because I realized that I need to sub step my collision sources and the volume source as well. So in order to do this, I actually just turned off the flip solver sub steps. I mean, I went back to the default ones. And I increased the overall dot network sub steps to 4. So I prefer to use these ones when we talk about a simulation like this. We have a really close up camera. And we need to see that the fluid is really sitting on the surface. But as you remember, if we look at our simulation, it's really fast moving, fast inflating. There's a lot of changes happening over a few frames. So we need proper sub-stepping. So the only thing that sub steps properly here is the geometry, the SOP that we have here. If we switch to it, you'll see that we have the time warp connected to this node connected to the time warp. So we're live interpolating this geometry source. But what we're not doing by baking this VDB is basically what we're not doing here is interpolating this VDB along with the geometry. So in order to do this, to do the proper interpolation, we have to save our VDBs and our sourcing in an interpolated manner. So the file cache nodes in the Houdini, I think 17 or 18 were changed and adjusted just so you could cache sub steps. So, if you're properly in the flip-dop, I set four sub-steps here, and I set the same in the file cache nodes, but the problem with this is the following. So basically, if we do another file cache here, I'll make another folder here, and I'll remove what I've done here, I'll come back to this. Let's say I set it to ff. for saving the sub steps you wouldn't use the f the dollar f expression you would use the ff because it represents the fractional frames and let's see what this does like five frames for example let's make two or three frames and I'll save it with four sub steps so what will happen here is we'll have this saving window and you see that it's rendering frame 1.5, then 175, and then the second frame, then 225. It divides every frame by 4, so you have increments of 0.25 frames every time. So for 3 frames it saved 9. So if we open up the folder, oh that actually worked. That's interesting because previously, I don't know if it's like a bug or something else, but previously when I was doing this fractional frame saving, it was actually giving me really terrible a bunch of decimal point numbers like this and the precision error at the end. So this could happen, really this could happen with a lot of these SOPs that are set in interpolation mode saving but in order to avoid this if you ever have this problem where it just yields you fractional frames with like 28 decimal points after the comma like if you have that issue you could fix it by using this expression so this is a Python expression that properly evaluates subframes and basically it searches for the current frame of the current node and it evaluates the this frame as an F3, I suppose, it's like a with three decimal points afterwards, but I'm not sure about this one, because I grabbed it from the Houdini forums and usually it, nobody gave an explanation, but this specific expression works really well and this is what I did here in the saving string. So after the name of your file, you insert this expression with the backticks. So if you third mouse click, if you mouse will click on the geometry file here, it will evaluate the name of your file. So if you go into enable playback at fractional frames here and you right click on this and you set it to four because we're at four sub steps, you will see that we're evaluating right now at 66.5 and if you go into the name over file here it evaluates as 66.5. So yeah this expression is life-saving if you have that issue with fractional frames but yeah apparently on this SOP or newly created SOP it works properly which is really interesting but yeah so I saved both the VDB that I had. That's interesting, apparently. We didn't save properly. I have another version that was saved properly like this. Let me find these fractional frames. It apparently didn't save it properly. Anyway, usually you would save the source, it saved the same way with four sub steps, with the spikin expression. And you can see that it evaluates at every fourth of a frame property. So if you step slowly, you would see that it's kind of flickering. But yeah, and we have four sub steps overall on the.NETROOK. Whenever you have fast moving colliders like this, fast moving sources, we would need to interpolate and save that with the sub steps. So we have the VDB of the geometry saved with sub steps, because it comes from the time warp. And then we have the geometry. This guy is interpolated live because it comes from the time warp. And it's sampled inside the collision source. I mean, not the collision source, but the collider in the static solver here. So it's using the deforming geometry and re-evaluating the subs to interpolate geometry. And the third is the source, which is interpolated also with four subs steps. and is sourced by the volume source here. So yeah, you have to match these sub steps that you have on the dot network in order to get the proper interpolation for this kind of simulation. So what else? I didn't change much afterwards. I just pushed out the particles after it was simulated. So yeah, the particles are here. There's some particles flowing down on the body. I did the UV texture thing with the camera optimization. So it removes clips on some of those particles. And in the particle fluid surface, I changed the method to spherical. Sometimes it works better for viscous fluids, especially when you have bubbles that you want to preserve inside the volume. But this affects more the flip tanks, where you have bubble paths. I mean not a bubble, but when you have internal bubbles forming, because an object fell into the volume of liquid and it creates like a narrow bubble after the collision. So this function really saves those bubbles. Usually if you use the average position, it will kill the bubbles. It will try to fill it up as fast as it can. So yeah, and yeah, I changed some parameters here. The attribute radius to 10 samples to 35. To better transfer the color. I cleaned up the mesh, and actually casted these guys and saved the mesh. So the mesh is like this. So yeah, this was transferred to my... this color is used by the shader. As you remember in the symbiote fluid you have the shader driving the blend between two colors. Two colors for the transmission and two colors for the roughness of the specular layer. So there's some areas that are less translucent and less refractive than the other ones. So this is cool. And then in the Octane Rop I turned off the Motion Blur and now I'll put some other passes, I think. Shading normal, UVs, Motion Vectors, just in case we can use it in comp with the real smart Motion Blur in After Effects. But yeah, that's about it. So the goal of this tutorial was to make something a little bit more unique than what you have on the market nowadays because I really haven't seen any tutorials concerning these soft-body simulations, especially value used for creatures and whatnot. But yeah, this is the type of tutorials that I aim to make in the future, and it all depends on how this one is going to work. But yeah, I enjoyed going over this one, and I hope you enjoyed it. And also learned something from the 3D part of this tutorial. and yeah let's go to the compositing stage. See you in the next one."
+    }
+  ]
+}