KR20230102287A - A method of predicting a support of 3D printing using shadow projection - Google Patents

Abstract

The present invention relates to a method for predicting a support body for 3D printing using a shadow projection method, and in a method for predicting the amount of a support structure (support) supporting an output object when the output object is printed using 3D printing, the output Calculate the volume of the surface of the object acquired by the camera and the volume of the invisible surface, and estimate the consumption of the support structure by integrating the pixel values of the tomography of the output object obtained by the camera. Outputs the output object with the minimum amount of In addition, the volume of the surface of the output object and the volume of the invisible surface obtained by the camera converts the object data of the output object from the +/- sign of the normal vector z component value of the mesh triangle pixel to the alpha pixel, beta pixel, TC pixel, After acquiring the NV pixels, the consumption of the support structure is predicted through their integration.

Description

A method of predicting a support of 3D printing using shadow projection}

The present invention relates to a method for predicting a support for 3D printing using a shadow projection method, and more particularly, to a method for predicting a support for 3D printing using a shadow projection method for predicting a support in order to minimize 3D printing filament consumption.

A 3D printer is a device for producing three-dimensional shapes, and was developed by an American company in the 1980s, starting with a printer that hardens plastic liquid to create three-dimensional objects. In the early days, due to the high production cost and intellectual property rights associated with 3D printers, it was limitedly used for making prototypes in the aerospace and automobile industries. As it is over, it is being developed and distributed in earnest.

Types of 3D printers typically include a Fused Deposition Modeling (FDM) method, a Selective Laser Sintering (SLS) method, a Stereo Lithography Apparatus (SLA) method, and a Digital Light Processing (DLP) method. The FDM-type 3D printer melts the filament at a high temperature, squeezes the filament through the nozzle hole, and creates a shape by stacking it. It has the disadvantages of poor finish quality, low strength in the lamination direction, and slow production speed. The SLS method selectively heats a specific part of a powder-type material to harden it, and the pedestal in the powder moves to build up the output layer by layer. As the powder is used as a material, no support structure is required, which has the advantage of free design. However, it has the disadvantage that it takes a relatively long time and the equipment is expensive. The SLA type 3D printer uses photocurable liquid resin to move the pedestal within the liquid resin and builds up layers of the cured resin. It has the advantage of efficient lamination speed and high precision, but adds supporter creation time. The downside is that it is expensive and the cost of materials, equipment and maintenance is high.

The DLP method is a method of curing the resin by shining a UV light source on a tank containing the resin to form an output object. Since it is molded on a surface-by-surface basis, it has the advantage of relatively fast molding speed and high precision, but the size of the molding that can be produced is limited and The downside is that it is expensive. Korean Patent Publication No. 2016-0128963 ("lamination method of stereolithography 3D printer", 2016.11.08) discloses 3D printer technology of SLA method and DLP method.

Among the various methods described above, the 3D printer of the DLP method is not suitable for mass production due to the advantages and disadvantages described above, but it is very suitable when mainly producing relatively small and complicated products such as medical implants. do. Accordingly, the use of DLP-type 3D printers is gradually increasing in practical fields such as the medical field.

On the other hand, the DLP-type 3D printer also needs to additionally form a supporter according to the shape of the modeling to be printed. There has been a problem such as printing failure of .

In detail, when the upper end of the modeling is formed of a plurality of ends having different heights rather than a single shape, the end with the highest height may be supported by being adhered to the stage and printed, but the other upper ends are supported during the printing process. Therefore, additional supporters had to be formed according to the height of the highest end, and novice users who did not know this fact led to printing failure of the product. In addition, even if the supporter is appropriately formed, a lot of material is consumed for forming the supporter.

On the other hand, Wang et al.'s (Wang et al. 2017) study is representative of the support structure minimization. They succeeded in reducing filament consumption by 66.56% for the famous dragon model by using a technique of dividing large objects into genus zero basic shapes such as cone, pyramid, ellipse, and sphere. However, this method does not take into account the optimal orientation of the pieces.

In order to find the optimal orientation of the output object, the most accurate method is to measure the g-code filament consumption for each rotation direction of the object. Some commercial slicing SWs provide an analysis function for predicting filament consumption as shown in FIG. 1 . However, these functions require calculation time because g-code generation is required for each direction, and thus only limited information such as x, y, and z-axis directions is provided.

Therefore, alternatives for calculating filament consumption per output direction instead of g-code generation have been proposed. Representatively, Ezair et al. assumed that the area obtained by subtracting the actual volume ( V _o ) of the object from the top cover volume ( V _tc ) of the object for a specific output direction (yaw, pitch, roll) is the amount of the support structure ( V _ss ). , the filament consumption was predicted more simply.

Here, V _o : The volume of an object, calculated geometrically from the coordinates of the triangular mesh.

V _ss : The volume of the support structure when the initial direction is (y,p,r).

Vtc _: Top cover volume. The volume formed by the top most triangle elements of an object with the bottom surface

(y,p,r): The rotation angle based on the yaw, pitch, and roll axes to indicate the output direction.

Calculating V _tc is a crucial step in determining the optimal output direction, but in fact this value is rather tricky to calculate. Ezair calculated Vo exactly and algebraically from the coordinates of the triangular mesh, and V _tc was calculated using the depth test of the graphic card (GPU). The advantage of using the GPU is that it can graphically find the top cover triangle at a very high speed. Wang et al. (Wang et al. 2010) and Das et al. (Das et al. 2017) calculated not only V _tc but also the entire object, such as Vo, at very high speed using voxelization within the GPU. did However, these methods require additional hardware such as a GPU, and in particular, the voxel-based method has a disadvantage in that memory usage increases as the size of the voxel decreases.

KR 10-1978996 KR 10-10-1756897 KR 10-2109812 KR 10-2019-0004593 KR 10-2017-0138618

The present invention for solving the above problems is to calculate the volume of the object, the amount of support structure generated, and the optimal orientation without the help of special hardware, and as a result, to output the output object while reducing the volume of the support during 3D printing, the amount of support The purpose is to provide a prediction method.

The present invention for achieving the above object is a method for predicting the amount of a support structure (support) for supporting an output object when printing the output object using 3D printing, obtaining the output object with a camera Calculate the volume of the surface and the volume of the invisible surface, and integrate the pixel values of the tomography of the output object obtained by the camera to predict the consumption of the support structure, and through this, output the object with the minimum amount of the support structure outputs

In addition, the volume of the surface of the output object and the volume of the invisible surface obtained by the camera converts the object data of the output object from the +/- sign of the normal vector z component value of the mesh triangle pixel to the alpha pixel, beta pixel, TC pixel, After acquiring the NV pixels, the consumption of the support structure is predicted through their integration.

라고 할 때, 베타 픽셀 중에서 법선 벡터의 z좌표인 n_z값이 아래 수학식 1을 만족하는 것들을 복사하여 NV픽셀로 분류하는 단계와, 상기 NV픽셀로부터도 단층사진을 형성하는 단계를 포함하여 구성된다.In addition, the critical angle of the overhang at which the filament used begins to create the support structure.

, among the beta pixels, the steps of copying and classifying beta pixels whose n _z value, which is the z coordinate of the normal vector, satisfies Equation 1 below, into NV pixels, and forming a tomogram from the NV pixels are also included. do.

[Equation 1]

는 지지구조 발생이 시작되는 오버행의 임계각으로써, +Z축과 이루는 각도값).(here,

is the critical angle of the overhang at which the support structure starts to occur, and is the angle value formed with the +Z axis).

In addition, the pixel information of the tomogram of the alpha, beta, TC, and NV pixels is respectively Vα,i, Vβ,j, Vtc,k, Vnv,l, and i,j,k,l and I, J,K , where L is a subscript for the number of each type of pixel and the total number, calculating the final support structure volume Vss from the four tomographic images by Equation 2 below.

[Equation 2]

(Where i, j, k, l and I, J, K, L are the number of alpha, beta, TC, and NV pixels, respectively).

In addition, the pixel information integration is based on subtracting the z-direction height value of the 3D printing floor from the z-direction height value of the pixel.

In addition, the calculation of the support structure volume V _ss is configured to include the step of repeating the rotation transformation for an arbitrary interval and displaying the values in a 4-dimensional box-shaped graph.

In addition, it is configured to form a continuous graph by filling empty spaces in the 4-dimensional box graph through 3-dimensional interpolation.

In addition, it is configured to include a step of finding the coordinate value of the pixel having the smallest value in the continuous 4-dimensional box graph and selecting it as an optimal alignment value.

In addition, in order to minimize a decimal point error, it is configured to include a step of removing duplicate pixels located at the same (x, y, z) integer coordinates so that only one remains.

Also, to reduce the volume calculation error, a step of removing pixels whose normal vectors are parallel to the XY plane is included.

In the present invention configured and operated as described above, the current state of the support structure (support) can be visually confirmed through tomography, which is a 2D picture file, and integrating the pixel values of this tomography can quantitatively determine the amount of support structure. , and the darker color of the final support structure tomography (Vss tomography) means an area where a lot of support structures occur, so if necessary, the user can cut and print this area to minimize the support structure. there is.

In addition, since the calculation process of the present invention is simple because it consists of only pixelation and four arithmetic operations, it can be easily implemented in various languages, and the process of obtaining a TC pixel from an alpha pixel is a kind of depth test of a GPU emulated by a CPU. That is, by using pixelation, GPU is not required, and there is an advantage that it can be applied to small PCs such as various embedded systems.

In addition, even if the volume (V _o ) of an object is not geometrically obtained through pixelation, it can be obtained from alpha and beta pixels. In particular, by using the NV pixel, the volume (V _ss ) of the support structure obtained with the critical angle can be easily obtained. There are advantages.

1 is a diagram showing a function of predicting filament consumption of 3D printing according to the prior art;
Figure 2 is a diagram showing the process of calculating the normal vector of a triangle in the support prediction method of 3D printing using a shadow projection method according to the present invention;
Figure 3 is an embodiment showing the orientation of the measurement object in the support prediction method of 3D printing using the shadow projection method according to the present invention,
4 is an example of a 4D box graph in the prediction method according to the present invention;
5 is a process diagram for analyzing the amount of support according to pixels in the prediction method according to the present invention;
6 is a diagram showing a process in which a boundary edge of a triangle is changed into a pixel in a prediction method according to the present invention;
7 is a diagram showing the range of alpha or beta pixels generated according to decimal point errors in the prediction method according to the present invention;
8 is a view for explaining the calculation process of the cone shape in the prediction method according to the present invention;
9 is a diagram showing the results of a Stanford bunny whose symmetric code is Gz in the prediction method according to the present invention;
10 is a diagram analyzing the influence of the critical angle condition in the prediction method according to the present invention;
11 is an example of box graph implementation for each measurement object in the prediction method according to the present invention;
12 is an exemplary view showing the initial orientation of a measurement object in a 4D box graph in the prediction method according to the present invention;
13 is a diagram showing pixels of support structure amount according to angular transfer in the prediction method according to the present invention;
14 is a diagram showing an embodiment of a prediction method according to the present invention;
15 illustrates an embodiment of a pixelation technique in accordance with the present invention;

Hereinafter, a support prediction method for 3D printing using a shadow projection method according to the present invention will be described in detail with reference to the accompanying drawings.

A method for predicting a support body for 3D printing using a shadow projection method according to the present invention is a method for predicting the amount of a support structure (support) for supporting an output object when the output object is printed using 3D printing, wherein the output object Calculate the volume of the surface acquired by the camera and the volume of the invisible surface, and estimate the consumption of the support structure by integrating the pixel values of the tomography of the output object obtained by the camera, and through this, Print the object in the minimum amount.

In addition, the volume of the surface of the output object and the volume of the invisible surface obtained by the camera converts the object data of the output object from the +/- sign of the normal vector z component value of the mesh triangle pixel to the alpha pixel, beta pixel, TC pixel, After acquiring the NV pixels, the consumption of the support structure is predicted through their integration.

The support prediction method of 3D printing using the shadow projection method according to the present invention calculates the volume (V _ss ) of the support structure (support), but instead of directly geometrically obtaining the object volume (V _o ), it is visible from the light source (or camera) The main technical point is to calculate by dividing the volume of the face and the volume of the invisible face. In the field of computer graphics, this invisible surface is referred to as a hidden surface. When rendering a complex 3D shape, the graphics card (GPU) removes the hidden surface through a depth-test in advance to improve rendering speed.

Four types of pixel groups, such as alpha and beta, are generated from the +/- sign of the normal vector z component value of the object mesh triangle pixel, and the filament consumption is predicted through their integration. In the prior art, only the total amount of filament consumption of the entire object could be known, but the present invention is characterized in that it is possible to accurately calculate the amount of filament (support material) consumed for each part of the object.

2 is a diagram illustrating a process of calculating a normal vector of a triangle in the method of predicting a support body for 3D printing using a shadow projection method according to the present invention.

Calculation of the V _TC value is the key to calculating the generation amount of the support structure. At this time, in the present invention, so that no special equipment such as a GPU is required, each triangle of the input object mesh is first converted into pixels as shown in (e), and then these pixels are divided into four groups according to the direction of the normal vector to calculate the volume. The reason for converting to pixels is to accurately calculate V _tc . The pixelation conversion of a triangle can be easily obtained through a 3D polygon scan-converting algorithm (Kaufman and Shimony 1987), a triangle rasterization algorithm (Kugler 1996), or a 3D Bresenham algorithm (Andres et al. 1997).

Converted pixels receive and store the normal vector of the parent triangle as it is. After that, as the first grouping, if the normal vector is in the +z direction, the pixel is copied into the alpha group and classified (a). The alpha group is characterized by not forming a supporting structure. Second, the opposite case is the beta group, which occurs in the overhang structure and causes the formation of the support structure (b). Thirdly, among the pixels of the alpha group, those located at the top in the +z direction are called the TC pixel group (c). These are key pieces of information needed to calculate V _tc . Finally, among the beta groups, as shown in (g), those whose angle of the normal vector is "below the critical angle for supporting structure generation" are named "NV (NonoVerhang) pixels". Although these are overhang structures, they do not generate support structures, so their value must be subtracted from the calculation of V _ss .

The next step is to calculate the volume each group forms with the bottom surface. For example, in the case of the alpha group, the total volume ( V _α,i ) of the alpha group can be calculated by integrating the height difference between the alpha pixels and the bottom surface of the 3D printer in the vertical direction, as shown in (f). The sum of these, V _α , means the total volume formed by the alpha triangles with the bottom surface, which has a value equal to or greater than the actual volume ( V _o ) of the object.

The volume of _the beta, TC pixel group ( Vβ , V′tc ) _is similarly calculated.

When [Equation 2,3,4] is applied in combination with (f) of FIG. 2, the actual volume ( V ₀ ) of the object can be calculated from the alpha and beta volumes as follows.

In addition, V _tc in Equation 1 has the same value as V _tc 'in Equation 4. Therefore, applying Equation 1 again gives Equation 6 below.

Here, the physical properties of each filament are not considered. That is, in Equation 6, the critical angle θ is 0°, and all beta pixels form a support structure. However, in actual 3D printing, the polymer filament has a critical angle of around 45 degrees depending on the type of material. Then, as shown in (d) of FIG. 2, some of the beta pixels do not generate a support structure according to the critical angle θ _c . This will be referred to as a Non-oVerhang (NV) pixel. When the normal vector of a specific beta pixel is n, the NV pixel can be determined under the following condition.

Here, θc is the critical angle of the overhang at which the support structure starts to occur, and is an angle value formed with the +Z axis. Considering this, Equation 6 is modified as follows.

As described above, support structure information is provided in the form of a 2D bitmap picture file only for a specific orientation. It is necessary to repeat these calculations for all angles to find the optimal orientation. The volume information obtained as a result is shown as a graph in the form of a 4D box in the present invention.

Figure 3 is an embodiment diagram showing the orientation of the measurement object in the support prediction method of 3D printing using a shadow projection method according to the present invention.

In the present invention, the notation "(y, p, r)" of orientation information is first defined. In the present invention, as shown in Equation 9, the orientation information of the object ( M ₁ ) heading in the (y, p, r,) direction is divided into three rotation matrices from the object's original position ( M ₀ ) and one translation matrix. It is intended to be expressed as a combination of , that is, R _X (y) , R _Y (p) , R _Z (r), T _Z . For example, R _X (y) means a transformation matrix that rotates counterclockwise by y degrees around the X-axis of the global coordinate system.

In FIG. 3, the orientation of an object is shown using the y, p, r notation. It is assumed that the input mesh data always exists on the XY plane as in (a) and is placed with the +Z direction as an up vector. Then, after rotating the object about the x-axis, the matrix that rotates about the y-axis and z-axis is expressed as "(y,p,r)". In addition, the axis Z axis is defined as the yaw axis, and the y axis is defined as the pitch axis and yaw axis. For example, if the input object is not rotated, the default orientation is represented as (0,0,0). In addition, “ V _ss (y, p, r)” means the volume of the support structure in the direction given (y, p, r). In Equation 9 below, the Tz matrix is a translation matrix that adjusts the vertical position of the rotated object so that it is placed on the floor. This matrix is necessary because the volume integral of the present invention is based on the bottom surface of the printer.

4 is an example of a 4D box graph for finding an optimal orientation. The three sides of the box mean orientation angles in three directions, such as yaw, pitch, and roll, respectively, and each has a range of 0° to 360°. Since it takes a lot of time to create tomographs for all angles, in the present invention, tomographs are created at specific intervals, the integral values are displayed as point data on a 4D box graph, and these are used after 3D interpolation. The bright white area inside the box indicates that there are many support structures. On the contrary, dark blue is the area with the minimum support structure, and the coordinate value of this point becomes the optimal output orientation. Later, in some graphs, red dots were manually marked to emphasize the position of the optimal alignment point.

The present invention is largely composed of two processes. First, for a given orientation angle, the amount of support structure generated is predicted in the form of a tomograph and the V _ss value is calculated. Second, after displaying the V _ss values obtained by varying the orientation angle on the 4D box graph, the optimal orientation angle is found at the point having the minimum value among them.

In the present invention, in order to form one final tomograph picture, four pictures, such as alpha, beta, TC, and NV, are required in the intermediate calculation process. Combining these pictures according to Equation 8, the final support structure amount distribution can be expressed as one tomography picture, and the sum of the tomography pixel values becomes the total amount of support structure. To verify the present invention, volume calculation was first compared by measuring tomography for simple figures such as spheres. However, the following assumptions were used.

Assumption 1: Filament material characteristics other than the critical angle ( θ _c ) condition for support structure generation are not considered.

Assumption 2: Internal structure, surface, and support structure are printed with the same filament.

Assumption 3: The interior of the object and its support structure are completely filled.

5 is a process diagram for analyzing the amount of support according to pixels in the prediction method according to the present invention. A sphere mesh whose symmetry code belongs to Gd is used as an example. From the input triangle mesh (a), triangles are divided into alpha/beta/TC groups according to the normal vector direction, and then displayed with different colors (b). When the height values of these pixels are integrated and displayed as colors on the floor, it is expressed as (c). Combining these three tomography pictures according to Equation 6 results in a tomography picture representing the volume of the final support structure as shown in (d). In the tomograph of (d), the dark colored pixels in common mean the places where the amount of support structure is large. This is also true for (c). Therefore, according to the present invention, the local support structure generation information of an object can also be intuitively known through a picture, and the entire volume can be accurately and numerically predicted through these pixel information, like the existing method of Ezair et al.

Here, as a result of measuring the volume by integrating the pixel values from the tomographs in (c) and (d) of FIG. 5, some errors were found. To find out the cause, the volume of the sphere mesh was measured by varying the radius without changing the geometry. As a result, it was confirmed that the smaller the average edge length, the larger the error. This is to use real data as pixels, i.e. integers, like the Bresenham algorithm used when the triangular mesh pixelation of the present invention is used to express straight lines on a monitor. As shown in FIG. It appears to be the cause of In order to prevent this, there is a method of greatly enlarging and inputting the size of the object, but in this case, there is a disadvantage in that the calculation time increases.

Another source of computational error is incorrect integration due to the pixels of vertical triangular elements. Triangles whose Z-axis angle of the normal vector is close to 0 degrees should theoretically have a volume value of 0 formed with the floor, but as shown in FIG. 7, some may be included in alpha or beta pixels due to decimal point error, in which case they are error cause of As a solution, there is a method of excluding those whose angles in the +Z-axis direction of the normal vector are close to 0 degrees in the pixelization process of alpha and beta pixels.

8 is a diagram for explaining a process of calculating a cone shape in the prediction method according to the present invention.

The preceding sphere is a case where orientation and support structure generation are not related. The next example uses the more general symmetric Gc type, cone, as an example. First, as a result of measuring tomography in a normal direction as shown in FIG. 8, it was confirmed that the support structure was 0 as shown in b. The error due to the size of the object occurred up to 25% in the case of R/H = 10/20, but as the _object became larger, the error tended to decrease to the 2% level, as shown in Table 2 above. Conversely, as a result of creating tomography in an inverted direction as in (c), it can be seen that the amount of support structure has increased significantly as in (d). In both cases, the calculation time could be executed in a short time of less than 1 second. Therefore, if the tomography is created while changing the orientation of the object in this way, the optimal printing direction that minimizes the generation of support structures can be easily predicted.

Accordingly, the pixelation technique of the present invention is similar to the "depth test" in which the GPU is used to remove the hidden surface of an object, but the difference is that the beta triangle group is not considered in the present invention. Using the present invention, it is possible to analyze the support structure in a CPU or embedded system without the need for a GPU.

In the present invention, the object volume (Vo) can also be obtained through simple arithmetic operations of alpha and beta pixel group height values. This is a part with originality that has not been used in previous studies.

In addition, in the present invention, NV pixels, that is, the amount of support structure (nv) considering the critical angle can be known through pixelation, and since all calculations are simple integer-based linear algebra techniques, the amount of calculation is small. In the case of the existing g-code generation technique, it consists of the process of slicing the object and the process of generating the tool path within the cut slice. In particular, the latter case belongs to the Traveling saleman problem, which is an NP-hard problem that requires a lot of computation time if there is no approximation (Zambito 2006). On the other hand, the pixelation of the present invention has an operation amount of O(N*(ηlogη+νμ)), which is simply proportional to the total surface area of the object. ,μ = the two longest sides of the three axes of the AABB enclosing the polygon)

9 is a diagram showing the results of a Stanford bunny whose symmetric code is Gz in the prediction method according to the present invention. To validate for a more general object, the results are presented for a Stanford bunny with symmetry code Gz. The output orientation was set to (0°, 0°, 0°) once, and the critical angle was assumed to be 0° once, that is, the support structure must occur in all overhang structures. (a) is the input mesh data, and (b) is the pixelation of these triangles and then separated into alpha, beta, and TC groups and displayed in different colors. By integrating the pixel volume of (b), 2D tomograph images for each group can be obtained as shown in (c). As a result, from the final support structure tomograph in (d), it can be visually confirmed that the most support structure occurs in the ear part of Stanford Bunny. As such, the present invention can be applied not only to simple geometric figures but also to objects of general and complicated shapes.

10 is an analysis of the influence of the critical angle condition. As a result of measurement while varying the critical angle, it can be seen that NV pixels are formed in different amounts as shown in purple in the figure. It can be seen that it has the same tendency when compared with the volume value measured by G-code conversion.

11 is an exemplary view of box graph implementation for each measurement object in the prediction method according to the present invention.

Since the support structure information of an object is represented by a total of four variables, such as orientation information (y, p, r) and the amount of support (support structure) Vss, the amount of support structure generated in each case must be expressed in 4D form. As shown in FIG. 10, integration was performed at regular intervals for three angles, such as yaw, pitch, and roll, and the result was plotted as point data on the graph, and then interpolated to implement a box graph. In order to find the optimal 3D printing output orientation angle, find the "(y,p,r)" coordinates of the point with the color next to it in this box graph. Of course, a plurality of optimal orientations may exist.

12 is an exemplary view showing the initial orientation of a measurement object in a 4D box graph in the prediction method according to the present invention. In the graph, the optimal orientation is indicated by a red dot. As shown in FIG. 11, the optimal orientation can be easily seen visually, and it can be seen that the calculation result of our method is also consistent with this. The calculation time is almost the same as the tomograph creation time in the previous section * (2π/ yaw angular interval) * (2π/ pitch angular interval) * (2π/ roll angular interval). For example, it took 41.92 seconds to create a 4d box graph of "Cone".

Regardless of the object's initial orientation, it should always exhibit the same optimal orientation. To verify this, as shown in FIG. 12, a 4D box graph was created in a state where the initial orientation of the object was different. As a result, it was demonstrated that the same results were produced regardless of the initial orientation.

13 is a diagram showing the pixels of the amount of support structure according to the angular transfer in the prediction method according to the present invention. This is the result of 4D graph creation when the angular interval of " Stanford Bunny " is changed. Regardless of the angular spacing, it can be seen that the pixel colors representing the amount of support structure in the box show an almost uniform distribution. As such, most normal objects have a closed-volume, so there is no need to calculate all angles. It can be seen that an angular spacing of about 30° is sufficient.

14 is a diagram showing an embodiment of a prediction method according to the present invention according to the present invention.

For most simple objects, the user can intuitively predict the optimal direction. Next, FIG. 14 is a drawing of a virtual Serpentine type heat exchanger. Using the present invention, it is easy to predict the optimal orientation at (90 °, 360 °, 360 °) as shown in (d) quantitatively even for a complex structure in which several shells are stacked.

15 is a diagram illustrating an embodiment of the pixelation technique of the present invention. 14A is a prior art, and an example using a voxelization technique of an object. In this case, as one or more voxels are generated in the Z-axis direction, an error in volume value calculation occurs. For pixelation, the present invention once creates pixels on the XY plane, as shown in FIG. 14b, and then translates them in the Z direction so that their center of gravity exists on the face of the object triangle. proceed with In this case, since only one pixel exists in the Z direction, there is an advantage in reducing the error in volume calculation.

The present invention configured as described above provides a technology capable of predicting the generation amount of a support structure in advance during 3D printing. Unlike the prior art, the present invention can calculate the volume of the object, the amount of support structure generated, and the optimal orientation without the help of special hardware such as a GPU.

The output target object is input in the form of a triangular mesh through a file format, and then each triangle is converted into pixels, and is divided into three groups: alpha, beta, and top-cover (TC) according to the direction of the normal vector. Among them, those whose direction of the beta pixel normal vector is smaller than the support structure critical angle are additionally classified as non-overhang (NV) and stored.

The final tomography picture can be obtained by integrating the volume of these pixel groups with the floor surface according to the given formula. Each pixel color of the tomography represents the amount of support structure, and integrating them all gives the total amount of support structure. If this process is repeated for three rotation directions, such as yaw, pitch, and roll, a support structure graph in the form of a 4D box is obtained. Likewise, each color means the amount of the support structure, and the coordinates of the point with the minimum value mean the optimal orientation in which the amount of support structure generated is the smallest during 3D printing.

Although the above has been described and illustrated in relation to preferred embodiments for illustrating the principles of the present invention, the present invention is not limited to the configuration and operation as shown and described. Rather, it will be appreciated by those skilled in the art that many changes and modifications may be made to the present invention without departing from the spirit and scope of the appended claims. Accordingly, all such appropriate changes and modifications and equivalents should be regarded as falling within the scope of the present invention.

Claims (11)

In the method of predicting the amount of the support structure (support) for supporting the output object when printing the output object using 3D printing,
Calculate the volume of the surface of the output object obtained by the camera and the volume of the invisible surface,
3D printing support prediction method using a shadow projection method for predicting the consumption of the support structure by integrating the pixel values of the tomography of the output object obtained by the camera and outputting the output object with the minimum amount of the support structure through this .
According to claim 1,
The volume of the surface of the output object and the volume of the invisible surface obtained by the camera are obtained by converting the object data of the output object to the +/- sign of the normal vector z component value of the mesh triangle pixel to the alpha pixel, beta pixel, TC pixel, and NV pixel. A support prediction method for 3D printing using a shadow projection method to predict the consumption of a support structure through their integration after acquisition.
The method of claim 1, wherein the consumption amount prediction,
Preparing object data of the output object to be 3D printed in the form of a triangular mesh and calculating each of the triangular normal vectors;
converting the triangle into pixels with integer coordinates and giving the triangle normal vector information to the pixels;
classifying the pixels as alpha pixels if the z-coordinates of the normal vectors assigned to the pixels are greater than 0, and as beta pixels if they are smaller;
copying and classifying only those having the largest z coordinate value among the alpha pixels having the same x and y coordinate values as TC pixels; and
Accumulating the alpha pixels, beta pixels, and TC pixels on the XY plane where the object is located, and representing the number of pixels in the form of a tomograph;
A support prediction method for 3D printing using a shadow projection method consisting of predicting each of the alpha, beta, and TC pixel values to both sides of the support structure.
According to claim 2,
The critical angle of the overhang at which the filaments used begin to create the support structure.

, copying beta pixels whose n _z value, which is the z coordinate of the normal vector, satisfies Equation 1 below, among beta pixels, and classifying them as NV pixels;
A method for predicting a support body for 3D printing using a shadow projection method, comprising: forming a tomogram from the NV pixels.
[Equation 1]

(here,

is the critical angle of the overhang at which the support structure starts to occur, and is the angle value formed with the +Z axis.)
According to claim 4,
The pixel information of the tomogram of the alpha, beta, TC, and NV pixels is respectively referred to as Vα,i, Vβ,j, Vtc,k, Vnv,l, and i,j,k,l and I, J, K,L 3D printing using a shadow projection method comprising: Support prediction method.
[Equation 2]

(Where i, j, k, l and I, J, K, L are the number of alpha, beta, TC, and NV pixels, respectively)
The method of claim 5, wherein the integration of pixel information comprises:
A support prediction method for 3D printing using a shadow projection method that targets the subtraction of the z-direction height value of a 3D printed floor from the z-direction height value of a pixel.
According to claim 6,
A method for predicting a support body for 3D printing using a shadow projection method comprising the steps of repeating rotation conversion for the calculation of the support structure volume V _ss at an arbitrary interval and displaying the values on a 4-dimensional box-shaped graph.
According to claim 7,
Forming a continuous graph by filling empty spaces in the 4-dimensional box graph through 3-dimensional interpolation; A support prediction method for 3D printing using a shadow projection method, comprising:
According to claim 8,
A method for predicting a support for 3D printing using a shadow projection method comprising: finding the coordinate value of the pixel having the smallest value in the continuous 4-dimensional box graph and selecting it as an optimal orientation value.
According to claim 3,
A support prediction method for 3D printing using a shadow projection method comprising the steps of removing duplicate pixels located at the same (x, y, z) integer coordinates so that only one remains in order to minimize decimal point error.
According to claim 3,
A support prediction method for 3D printing using a shadow projection method comprising the steps of removing pixels whose normal vector is parallel to the XY plane in order to reduce volume calculation error.

Images