Controlling Liquids Using Pressure Jump - Semantic Scholar
Controlling Liquids Using Pressure Jump Seung-Ho Shin* Korea University
Jung Lee Korea University
Abstract
This sketch presents a method to control liquids so that they flow into a target shape in a natural way. To avoid the oscillation and preserve the sharp detail of the target shape, we integrate the target-driven force into the projection step. So we require no postoptimization process for divergence-free condition at the liquid interface. Additionally, the target-driven force is calculated by enhanced ghost fluid method considering the pressure discontinuity around the liquid interface.
1 Introduction
Hong and Kim [2005] deal with interface discontinuities using the ghost fluid method (GFM). They consider surface tension at the interface between two immiscible fluids at the projection step and introduce a discontinuous viscosity condition. We add a targetdriven control force by using GFM that considers multiphase interfaces. The user may provide an image, 3D mesh data or sketches. Based on this input, the level-set of the target shape is constructed as a signed distance function, from which pressure jump values can be determined. Hong and Kim [2004] have suggested another efficient, simple method to control a fluid animation, in which a potential field is based on the shape of the target. An external force obtained from the negative gradient of the potential field enables the smoke to move towards the target shape. On the other hand, Shi and Yu [2005] control a liquid animation, using a feedback force with a velocity and a shape component. Because the feedback force is reduced by the projection step that makes the velocity field divergence-free, they perform an additional optimization process to make the shape component divergence-free, which avoids reducing the force. We use the same concept of shape feedback, but we do not need to worry about a reduced force because the force is added at the projection step, rather than as an external term.
Sun-Jeong Kim Hallym University
Chang-Hun Kim† Korea University
being chaged into the target shape of bunny. Small-scale details on the target shape are preserved by matching liquid interface such as ears of bunny. Figure 2 shows interesting liquid dynamics when floating bubbles put together, change Chinese dragon and drop with gravity. Like this, animator can represent much more effects of liquids easily. Our fluid simulation is divergence-free and robust, despite the addtion of the control force and the avoidance of any optimization process. We demonstrate realistic and smooth fluid motion within animation in which the fluid accurately assumes a target shape. This technique can easily implemented on existing fluid simulation pipelines.
Figure 1: The initial Venus shape is changed into the bunny shape. The resolution is 1283 .
2 Methods
In this sketch, the target shape is represented as a level-set data structure with a signed distance function represented by an adaptive octree grid. From this shape information, we can calculate the shape feedback force [Shi and Yu 2005]. In previous method [Hong and Kim 2005], surface tension causes a jump J in pressure across interface of liquid. Conversely, in this sketch, the magnitude of J is determined from the signed distance function of the target shape, and this pressure jump provides the force that makes the fluid assume the target shape. When a point on the liquid surface is outside the target shape, the magnitude of J is αφ t arg et , where φ t arg et is the value of the signed distance function of the target shape at the liquid interface, which is determined by interpolating between values of the signed distance function, φ t arg et at near nodes. But in the case of a point on the liquid surface, which is inside the target shape, the control force is added in the direction nomal to the liquid interface. The magnitude of J may be − σκ , which signifies a negative direction of the surface tension.
3 Results
We have successfully applied our method to control liquids with interfacial pressure discontinuity. Figure 1 shows three frames from a three dimensional controlled fluid animation which make source shape of Venus
Figure 2: The floating bubbles put together, change Chinese dragon and drop with gravity. The resolution is 1283 .
References
HONG, J.-M., and KIM, C.-H. 2005. Discontinuous fluids. ACM Transactions on Graphics, 24, 3, 915-920. HONG, J.-M., and KIM, C.-H. 2004. Controlling fluid animation with geometric potential. Computer Animation and Virtual Worlds 15, 147–157. SHI, L., and YU, Y. 2005. Taming liquids for rapidly changing targets. ACM SIGGRAPH/Eurographics Symposium on Computer Animation, 229-236. * [email protected] † [email protected]
Jung Lee Korea University
Abstract
This sketch presents a method to control liquids so that they flow into a target shape in a natural way. To avoid the oscillation and preserve the sharp detail of the target shape, we integrate the target-driven force into the projection step. So we require no postoptimization process for divergence-free condition at the liquid interface. Additionally, the target-driven force is calculated by enhanced ghost fluid method considering the pressure discontinuity around the liquid interface.
1 Introduction
Hong and Kim [2005] deal with interface discontinuities using the ghost fluid method (GFM). They consider surface tension at the interface between two immiscible fluids at the projection step and introduce a discontinuous viscosity condition. We add a targetdriven control force by using GFM that considers multiphase interfaces. The user may provide an image, 3D mesh data or sketches. Based on this input, the level-set of the target shape is constructed as a signed distance function, from which pressure jump values can be determined. Hong and Kim [2004] have suggested another efficient, simple method to control a fluid animation, in which a potential field is based on the shape of the target. An external force obtained from the negative gradient of the potential field enables the smoke to move towards the target shape. On the other hand, Shi and Yu [2005] control a liquid animation, using a feedback force with a velocity and a shape component. Because the feedback force is reduced by the projection step that makes the velocity field divergence-free, they perform an additional optimization process to make the shape component divergence-free, which avoids reducing the force. We use the same concept of shape feedback, but we do not need to worry about a reduced force because the force is added at the projection step, rather than as an external term.
Sun-Jeong Kim Hallym University
Chang-Hun Kim† Korea University
being chaged into the target shape of bunny. Small-scale details on the target shape are preserved by matching liquid interface such as ears of bunny. Figure 2 shows interesting liquid dynamics when floating bubbles put together, change Chinese dragon and drop with gravity. Like this, animator can represent much more effects of liquids easily. Our fluid simulation is divergence-free and robust, despite the addtion of the control force and the avoidance of any optimization process. We demonstrate realistic and smooth fluid motion within animation in which the fluid accurately assumes a target shape. This technique can easily implemented on existing fluid simulation pipelines.
Figure 1: The initial Venus shape is changed into the bunny shape. The resolution is 1283 .
2 Methods
In this sketch, the target shape is represented as a level-set data structure with a signed distance function represented by an adaptive octree grid. From this shape information, we can calculate the shape feedback force [Shi and Yu 2005]. In previous method [Hong and Kim 2005], surface tension causes a jump J in pressure across interface of liquid. Conversely, in this sketch, the magnitude of J is determined from the signed distance function of the target shape, and this pressure jump provides the force that makes the fluid assume the target shape. When a point on the liquid surface is outside the target shape, the magnitude of J is αφ t arg et , where φ t arg et is the value of the signed distance function of the target shape at the liquid interface, which is determined by interpolating between values of the signed distance function, φ t arg et at near nodes. But in the case of a point on the liquid surface, which is inside the target shape, the control force is added in the direction nomal to the liquid interface. The magnitude of J may be − σκ , which signifies a negative direction of the surface tension.
3 Results
We have successfully applied our method to control liquids with interfacial pressure discontinuity. Figure 1 shows three frames from a three dimensional controlled fluid animation which make source shape of Venus
Figure 2: The floating bubbles put together, change Chinese dragon and drop with gravity. The resolution is 1283 .
References
HONG, J.-M., and KIM, C.-H. 2005. Discontinuous fluids. ACM Transactions on Graphics, 24, 3, 915-920. HONG, J.-M., and KIM, C.-H. 2004. Controlling fluid animation with geometric potential. Computer Animation and Virtual Worlds 15, 147–157. SHI, L., and YU, Y. 2005. Taming liquids for rapidly changing targets. ACM SIGGRAPH/Eurographics Symposium on Computer Animation, 229-236. * [email protected] † [email protected]
