A method for low-face-number 3D modeling of an object with high precision and fidelity

Abstract

The application discloses a kind of low face number 3D modeling method of high-precision realistic to object, belong to three-dimensional modeling technical field.The application is first to the three-dimensional model obtained by three-dimensional scanning is reduced face, to overcome the defects that the face number of model obtained by 3D scanning mode is more, will be replaced into temporary three-dimensional model based on first three-dimensional model, and utilize color feature based on photo obtains and carries out texture baking, overcome the defects that model obtained by 3D scanning mode does not have the color (texture) map of original object or color is not clear, color difference is larger, etc., normal map and ambient light shading map are generated based on second three-dimensional model before reducing face, guarantee the high definition of realistic 3D model obtained, the application combines 3D scanning modeling mode and photo modeling mode, can realize the 1:1 reduction of object, make model can faithfully reflect the concave-convex, luster, roughness etc. Characteristics of original object, while reducing the face number of model, improve the realistic effect of model obtained by modeling.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional modeling, in particular to a method for low-face-number 3D modeling of high-precision realistic objects. BACKGROUND

[0002] The current digital development is in full swing, which has brought about some needs for 3D digital modeling of real objects. For example, in the planning and design industry, RTK unmanned aerial vehicles are used to take pictures of the designed terrain, and then software is used to form a planar topographic map with terrain and elevation and a 3D digital model reflecting the terrain, providing clear design conditions for planning and design. For example, in the game industry, when it is necessary to reflect real objects, it is necessary to scan and take pictures of the objects, and then use 3D modeling software to manually perform reverse modeling to form a model with high similarity to the original object for display in the scene. In the field of digital protection and digital collection of cultural relics, the relatively mature 3D scanning method is used to restore the original object to a digital model in a 1:1 ratio for display and application on the WEB or in a program.

[0003] Currently, there are two ways to achieve realistic 3D modeling of physical objects. One is to use a 3D scanner to scan and model to obtain a 3D model, and the other is to take photos and use photo modeling software to obtain a 3D model.

[0004] The 3D model formed by 3D scanning can restore the geometric features and actual size of the original object in a 1:1 ratio, but has the following disadvantages: the model has a large number of faces, with tens of thousands of faces in the least and millions of faces in the most, resulting in a large amount of model data, which consumes high computing resources for display on the WEB or in a program. At the same time, existing 3D scanners generally use colorless scanners for scanning, and the model formed by colorless scanning does not have the color (texture) map of the original object, which cannot achieve realistic results.

[0005] The 3D model established by the existing photo modeling method has high similarity in shape compared to the original object, and can generate a model with a low number of faces, but has the following disadvantages: the size of the 3D model differs greatly from that of the original object, and the size of the original object cannot be accurately restored. Moreover, the UV layout of the color (texture) map generated by the software is not full-scale, which results in insufficient pixels and low clarity of the model map. The model only has a color (texture) map, but no normal, ambient light shading map, metal degree, and roughness map, which makes the model details insufficient and cannot accurately reflect the concave-convex, gloss, and roughness of the original object. SUMMARY

[0006] The application aims to provide a low-face-number 3D modeling method with high-precision and high-fidelity for objects, which can realize 1:1 reduction of objects, reflect the concave-convex, gloss, roughness and other characteristics of the original objects, reduce the face number of the obtained model and improve the fidelity of the obtained model.

[0007] To achieve the above-mentioned purpose, the application provides the following solutions.

[0008] The application provides a low-face-number 3D modeling method with high-precision and high-fidelity for objects, which comprises the following steps.

[0009] The target object is three-dimensionally modeled and color features are extracted by using a photo modeling method to obtain a first three-dimensional model and color features.

[0010] The target object is three-dimensionally modeled by using a three-dimensional scanning method to obtain a second three-dimensional model.

[0011] The second three-dimensional model is subjected to face reduction processing to obtain a second three-dimensional model after face reduction, and the face number of the second three-dimensional model after face reduction is consistent with that of the first three-dimensional model.

[0012] The second three-dimensional model after face reduction is split and flattened UV to obtain a second three-dimensional model after UV addition, and the second three-dimensional model after UV addition is copied to obtain a temporary three-dimensional model.

[0013] The temporary three-dimensional model is subjected to size and position adjustment so that the size and position of the temporary three-dimensional model are consistent with those of the first three-dimensional model to obtain an aligned temporary three-dimensional model.

[0014] The aligned temporary three-dimensional model is baked by using the color features to generate a color map.

[0015] According to the external texture of the second three-dimensional model, a normal map and an ambient occlusion map of the second three-dimensional model after UV addition are obtained.

[0016] A metalness map and a roughness map corresponding to the target object are obtained.

[0017] According to the second three-dimensional model after UV addition, the color map, the normal map, the ambient occlusion map, the metalness map and the roughness map, a high-fidelity 3D model of the target object is obtained.

[0018] Optionally, the target object is three-dimensionally modeled by using a photo modeling method to obtain a first three-dimensional model, which specifically comprises the following steps.

[0019] photographing M target real object photos by using preset photographing tracks, and photographing at least one color correction card photo; M is greater than or equal to 360, and the preset photographing track is a track formed by wrapping the target real object in a spherical, hemispherical or cylindrical shape;

[0020] color calibrating the M target real object photos based on the color correction card photo respectively;

[0021] respectively subjecting the M target real object photos after color calibration to subject matting processing to obtain M target real object foreground pictures;

[0022] importing the M target real object foreground pictures into 3D modeling software to perform three-dimensional modeling to obtain a first three-dimensional model.

[0023] Optionally, importing the M target real object foreground pictures into 3D modeling software to perform three-dimensional modeling to obtain a first three-dimensional model, specifically comprising:

[0024] importing the M target real object foreground pictures into 3D modeling software and checking whether the pictures are recognized completely;

[0025] if yes, generating a first three-dimensional model based on the M target real object foreground pictures;

[0026] if no, not generating the first three-dimensional model.

[0027] Optionally, performing three-dimensional modeling on the target real object by using a photo modeling method to obtain a first three-dimensional model, and then comprising:

[0028] performing surface reduction on the first three-dimensional model.

[0029] Optionally, splitting and flattening UV of the second three-dimensional model after surface reduction to obtain a second three-dimensional model after UV addition, specifically comprising:

[0030] performing UV splitting on the second three-dimensional model after surface reduction to obtain a UV model covering the entire frame;

[0031] adding color materials to the UV model to obtain a second three-dimensional model after UV addition.

[0032] Optionally, performing baking on the aligned temporary three-dimensional model by using the color features to generate a color map, specifically comprising:

[0033] replacing the first three-dimensional model with the aligned temporary three-dimensional model in the 3D software and performing texture baking by using the color features to generate a color map.

[0034] Optionally, obtaining a metalness map and a roughness map corresponding to the target real object, and then comprising:

[0035] According to the second three-dimensional model after UV addition, the color map and the environment light shading map, a 3D map model is obtained;

[0036] Based on the 3D map model, the color map is detected for defects, and the defect part in the color map is repaired.

[0037] Optionally, the color map, the normal map, the environment light shading map, the metal degree map and the roughness map are of the same preset size.

[0038] The preset size is 8K (8192x8192) or 1K (1024x1024).

[0039] According to the specific embodiments of the present application, the following technical effects are provided:

[0040] In the embodiments of the present application, the first three-dimensional model and the second three-dimensional model are obtained based on the photo and the three-dimensional scanning respectively, and then the second three-dimensional model obtained by the three-dimensional scanning is reduced to be consistent with the first three-dimensional model, and the subsequent reduced second three-dimensional model is used to generate a realistic three-dimensional (3D) model of the target object, so as to overcome the defect that the model obtained by the 3D scanning has more surfaces, and at the same time, the advantages of 1:1 reduction of the geometric characteristics and actual size of the target object by the 3D scanning method are combined.

[0041] In the process of generating the above-mentioned realistic 3D model, the second three-dimensional model after reduction is subjected to UV splitting and flattening to obtain the second three-dimensional model after splitting and flattening UV. UV splitting can obtain a series of UV coordinates. The UV coordinates can be regarded as a bridge connecting the 3D model and the 2D texture, and describe where the 2D texture should be placed on the surface of the 3D model. In the process of generating the above-mentioned realistic 3D model, the first three-dimensional model is replaced by the aligned temporary three-dimensional model to generate a color map, so as to overcome the defect that the model obtained by the 3D scanning does not have the color (texture) map of the original object. In addition, the normal map and the environment light shading map are generated based on the second three-dimensional model before reduction, and the metal degree map and the roughness map are obtained. The above-mentioned each map and the split second three-dimensional model together generate the above-mentioned realistic 3D model. The realistic 3D model obtained in this way has the color (texture) of the original target object, and due to the effect of the normal map and the environment light shading map, the realistic 3D model can faithfully reflect the concave-convex details of the surface of the target object, and the use of the metal degree map and the roughness map can make the realistic 3D model more realistically reflect the gloss and roughness characteristics of the object surface.

[0042] In conclusion, the embodiment of the present application combines the modeling method of 3D scanning and the modeling method of photos, can realize 1:1 reduction of the object, and can make the 3D model obtained by modeling reflect the concave-convex, gloss, roughness, texture and other characteristics of the original object, while reducing the face number of the model obtained by modeling and improving the realistic effect of the model obtained by modeling. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0044] Figure 1 The flow chart of the method for low-face-number 3D modeling of high-precision realistic object provided by the embodiment of the present application;

[0045] Figure 2 The principle diagram of the method for low-face-number 3D modeling of high-precision realistic object provided by the embodiment of the present application;

[0046] Figure 3 The schematic diagram of the photo shooting in the constant light of the studio provided by the embodiment of the present application;

[0047] Figure 4 The schematic diagram of the angle position of the tripod provided by the embodiment of the present application;

[0048] Figure 5 The schematic diagram of the spherical trajectory provided by the embodiment of the present application;

[0049] Figure 6 The schematic diagram of the semi-spherical trajectory provided by the embodiment of the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0051] The purpose of the present application is to provide a method for low-face-number 3D modeling of high-precision realistic object. By combining the modeling method of 3D scanning and the modeling method of photos, 1:1 reduction of the object can be realized, the model obtained by modeling can reflect the concave-convex, gloss, roughness and other characteristics of the original object, while the face number of the model obtained by modeling is reduced and the realistic effect of the model obtained by modeling is improved.

[0052] In order to make the above objectives, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0053] The embodiment of the present application provides a method for 1:1 high-precision 3D realistic modeling of an entity object with low surface number, which solves the problems that the size error of the result of the 3D modeling of the entity object with low surface number is too large, the color (texture) map is stretched, blurred and low in precision, the number of maps is insufficient, and the concave-convex, gloss, roughness and other characteristics of the original entity object cannot be faithfully restored, and solves the problems that the standard is not unified and the quality is uncontrollable when a large number of realistic 3D models are made.

[0054] It should be noted that the embodiment of the present application improves the existing method for realistic 3D modeling of an object, so that the problems and shortcomings of the existing realistic 3D modeling can be effectively solved under the condition of using the existing software, and the present application is only suitable for objects that can be moved and turned by a single person or objects that cannot be moved but have enough operating space around them, and is not suitable for objects with transparent surfaces, uniform colors and high reflectivity, such as glassware, pure color and high reflectivity porcelain and the like.

[0055] As shown in Figure 1 and Figure 2 The embodiment of the present application provides a method for low-surface-number 3D modeling of an object with high-precision realism, which comprises the following steps:

[0056] Step 101, using a photo modeling method to perform three-dimensional modeling and color feature extraction on the target entity object to obtain a first three-dimensional model and color features.

[0057] Step 101 in the embodiment of the present application specifically comprises:

[0058] Step 11: Taking a photo. The embodiment of the present application takes M photos of the target entity object using a preset shooting track, and takes at least one color calibration card photo; M is greater than or equal to 360, and the preset shooting track is a track formed by wrapping the target entity object in the shape of a sphere, a hemisphere or a cylinder.

[0059] Exemplary, the embodiment of the present application adopts a high-definition digital single-lens reflex camera or a mirrorless full-frame camera with effective pixels not less than 20 million pixels. In a constant light environment, the camera is placed at 0 degrees, 20 degrees, 40 degrees, 60 degrees, and 80 degrees (or 0 degrees, 15 degrees, 30 degrees, 45 degrees, 60 degrees, and 75 degrees) of the object, respectively, and a photo is taken every 5-15 degrees around the object in the horizontal direction. The object is inverted (if the object can be inverted), and a photo is taken according to the above method. A total of not less than 360 photos are taken, and a color calibration card is also taken. The shooting track should be spherical (or hemispherical) or cylindrical wrapping around the object. The technical means and measures in this step enable all the photos to be aligned and recognized by 3D modeling software, and ensure the high pixel and high definition of the color (texture) map of the generated 3D model, greatly reducing and avoiding the stretching and blurring of the color (texture) map. The key technology of this step is the photo shooting method. If the photo is not taken well, it will directly lead to the failure of model establishment in the later process.

[0060] Taking a stone as the target object, the embodiment of the present application provides a specific implementation of step 11 as follows:

[0061] 11-1, preparation stage

[0062] Observe the shape, color, luster, etc. of the stone to determine the scheme for taking photos and scanning the model. Since the stone is small and movable, a studio constant light is adopted, such as Figure 3 In this example, a multi-face inversion all-around shooting scheme is adopted. Since the stone has a reflection, a polarizer is needed to reduce the surface reflection when shooting, so as to reduce the texture stretching and blurring of the generated 3D model in the later step. A light reduction agent is needed to be applied when scanning to eliminate reflection and ensure scanning quality. The preparation work is to find the scheme for taking photos and scanning. For example, if the surface color of the object is relatively uniform, it is necessary to mark points on the surface that are different from the surface color (adhesive can be used, etc.). The marking points should be marked on different surfaces of the object, and there should be not less than 5 marking points. The purpose of this is to enable all the photos to be recognized by the 3D modeling software (RealityCapture) in the later step. If there are many objects to be made, they need to be numbered and a registration table needs to be established to prevent confusion in the later production.

[0063] 11-2, taking high-definition photos of the stone

[0064] The high-definition digital single-lens reflex camera or reflexless full-frame camera with effective pixels not less than 20 million pixels is used in the constant light environment in the studio. The stone is placed in the center of the turntable in the studio. The camera is set in front of the stone using a tripod, at the 0-degree angle position of the turntable. The exposure parameters of the camera are adjusted. The aperture of the camera is locked at F22 or above. The ISO is set at 100. The focal length of the camera cannot use the wide-angle section. The focal length of 50mm-80mm is used. After the focal length is determined, the focal length should not be changed at will until the required photos of the stone are completely taken. The storage format of the photos is RAW format. The stone image is in the center of the camera screen. The size ratio must exceed half of the screen to ensure the clarity and resolution of the photos. A photo is taken every 10 degrees of rotation of the turntable. The height of the tripod is adjusted vertically to make the camera at the 0-degree, 20-degree, 40-degree, and 60-degree angles of the stone in turn. Then the stone is reversed in the opposite direction (the bottom of the last time faces up). The camera is lowered from the high position to take photos at the 60-degree, 40-degree, 20-degree, and 0-degree angles of the stone in turn. Then the side of the stone is turned up. The camera is taken at the 0-degree, 20-degree, 40-degree, and 60-degree angles of the stone in turn. As shown in Figure 4 , the number of photos of the stone is not less than 432. In this example, the number of photos of the stone is 792. One photo of the 24-color calibration card and the gray card is taken. The shooting track must be spherical wrapping for the stone, as shown in Figure 5 . The shooting method of this step can be determined according to the shape of the object. If the object shape is quadrangular, it only needs to be turned up and down once to take photos. If the object shape is triangular, it needs to be turned up once to take four circles. If the object cannot be turned over, it needs to be taken around the object with the object as the center. A photo is taken every 10 degrees horizontally. A photo is taken every 10 degrees at the 0-degree, 20-degree, 40-degree, 60-degree, and 80-degree positions vertically. Finally, a kind of hemispherical wrapping shooting track is formed, as shown in Figure 6 . This kind of shooting can solve the problem that the photos can be fully recognized when generating point cloud data models in the software, enhance the details of the generated models, and solve the problems of stretching and blurring of the texture map. This step of taking photos is the key to whether the photos can be successfully aligned and fully recognized in the 3D modeling software. Therefore, it is very important to take photos according to the above steps and methods.

[0065] Step 12: processing photos. The M target object photos (i.e. 792 in this example) are color calibrated based on the calibration card photos according to the embodiment of the application. The M target object photos after color calibration are subjected to main body cutout processing to obtain M target object foreground pictures, as follows:

[0066] The photos are imported into the Lightroom software for photo color calibration, and after the calibration is completed, the photos are exported and then imported into the Photoshop software for subject cutout processing, and other information except the real object is deleted, and the photos with transparent background are exported. The technical means and measures of this step are used for picture processing, which ensures that the next 3D modeling software (RealityCapture) can completely identify the position and shooting track of the photos when aligning the photos, improves the completeness of the generated point cloud model, and the color (texture) of the generated 3D model is more accurate, and the speed and accuracy of the generated model are improved. The technical means of photo color calibration makes the color (texture) of the model closer to the color of the original real object.

[0067] The specific implementation of step 12 provided by the embodiment of the present application is as follows when the stone is taken as the target object:

[0068] The taken photos are imported into the Lightroom software for photo color calibration, and after the calibration is completed, the photos are exported and then imported into the Photoshop software for stone subject cutout processing, and other information except the stone is deleted, and the photos with transparent background in tif format are exported. Here, the photo color calibration is performed to make the color of the generated model closer to the color of the stone and the fidelity higher. The photo cutout processing is performed to solve the problem that all the photos can be identified when aligning the photos in the 3D modeling software (RealityCapture) in the next step, and to ensure the effect and clarity of the generated 3D model.

[0069] Step 13: generating a model. The embodiment of the present application imports the M target object foreground pictures into the 3D modeling software for three-dimensional modeling to obtain a first three-dimensional model, and the specific implementation is as follows:

[0070] The processed photos are imported into the 3D modeling software (RealityCapture) for photo alignment to generate a preview model, the model is reduced to not more than 20,000 faces according to the complexity, and then the model is exported, that is, a first three-dimensional model, the model is named as RC.obj, and the current project file is saved as RC.rcpoj. The technical means and measures of this step are to replace the current 3D model baked color (texture) map with the light data of the scanned model in the next step.

[0071] The specific implementation of step 13 provided by the embodiment of the present application is as follows when the stone is taken as the target object:

[0072] The processed tif format photo is imported into 3D modeling software (RealityCapture) to perform photo alignment, and a point cloud 3D model of the stone is generated. Check whether the photo recognition is complete, if the photo is not all recognized or the recognized photo is divided into multiple groups, identify how many pictures in each group are recognized, if the photo recognition rate in each group is less than 90%, a new shooting plan is formulated, and all photos are shot again, including color correction and matting processing of the photos. If the recognition rate of a group of pictures reaches more than 90%, the group is selected for the following procedure, and the other groups are deleted. If all are recognized, a preview level 3D model of the stone is generated, the generated stone 3D model is reduced to 13000 faces according to its complexity, and then an obj format 3D model, that is, a first three-dimensional model, is exported, which is named as RC.obj, and the current project file is saved as RC.rcpoj.

[0073] In step 102, a three-dimensional scanning method is used to perform three-dimensional modeling on the target real object to obtain a second three-dimensional model.

[0074] In the embodiment of the application, a laser 3D scanner is used to perform high-precision (error ±0.1mm) three-dimensional scanning on the real object to be modeled, and a 1:1 model without color and high-precision consistent with the real object, that is, a second three-dimensional model, is formed. The model is named Scan.stl.

[0075] In step 103, the second three-dimensional model is subjected to face reduction processing to obtain a second three-dimensional model after face reduction, and the number of faces of the second three-dimensional model after face reduction is consistent with the number of faces of the first three-dimensional model.

[0076] In the embodiment of the application, the Scan.stl model is subjected to face reduction in 3D software (blender) to the same number of faces as the RC.obj model, and a second three-dimensional model after face reduction is obtained, which is then exported and named Scan_Low.obj, as follows:

[0077] The Scan.stl model is subjected to face reduction in 3D software (blender) to the same number of faces as the RC.obj model (13000 faces) by using multiple face reduction methods, and then exported and named Scan_Low.obj. The purpose here is to use the high precision of the scanning model to greatly reduce the model and stone size error after face reduction, and the high precision of the scanning model provides conditions for the subsequent normal mapping and ambient occlusion mapping.

[0078] In step 104, the second three-dimensional model after face reduction is split and flattened UV to obtain a second three-dimensional model after UV addition, and the second three-dimensional model after UV addition is copied to obtain a temporary three-dimensional model.

[0079] In the embodiment of the present application, the UV software (RizomUV) is used to perform UV splitting and flattening on the Scan_Low.obj model, the UV requirement is 8K (8192x8192) size covering the entire frame, the model after splitting and flattening is exported to the blender software, an arbitrary color material is made for the model (the purpose is that the UV is not lost when the file is exported in the next step), then the original file of the model is overwritten and exported, and the file name is still "Scan_Low.obj". A copy of the "Scan_Low.obj" file is copied, the copied file is a temporary three-dimensional model, and is named "Scan_Tmp.obj". The purpose of copying the file is to use it to align the "RC.obj" model generated in the above step.

[0080] The "Scan_Low.obj" model will be used to connect all the material files in the final result model. Although the "Scan_Tmp.obj" model and the "Scan_Low.obj" model only change in size and position after the subsequent scaling and alignment operations, the UV of the two models is the same, so the texture will not be affected.

[0081] In step 105, the size and position of the temporary three-dimensional model are adjusted so that the size and position of the temporary three-dimensional model are consistent with the size and position of the first three-dimensional model, and an aligned temporary three-dimensional model is obtained.

[0082] In the embodiment of the present application, the 3D model with the name "Scan_Tmp.obj" and the 3D model with the name "RC.obj" are imported into the alignment software (GeomagicWrap) together, the "Scan_Tmp.obj" model is enlarged or reduced to the same size as the "RC.obj" model and is aligned to the same coordinate position as the "RC.obj" model, the error is controlled within ±0.1mm, then the "Scan_Tmp.obj" model is exported and the original file is overwritten. The key of the operation here is that the "Scan_Tmp.obj" model should be the same size as the "RC.obj" model, and the coordinate position of the "Scan_Tmp.obj" model should also be the same as the "RC.obj" model. The smaller the error is, the less the stretching and blurring of the color (texture) map will be, and even no stretching and blurring will occur.

[0083] In step 106, the aligned temporary three-dimensional model is baked using the color feature to generate a color map, that is, in the 3D software, the first three-dimensional model is replaced by the aligned temporary three-dimensional model, and the color feature is used for texture baking to generate a color map.

[0084] The embodiment of the application opens the previously saved project file named "RC.rcproj" by using 3D modeling software (RealityCapture), imports the "Scan_Tmp.obj" model aligned in the previous step into the 3D modeling software (RealityCapture) by clicking "Import Model" in the menu "Reconstruction Settings", replaces the model in the current file, then clicks "Texture" in the menu to perform texture baking, exports the model together with the color map (export 8K pixels) after completion, and the original file is overwritten, deletes the "Scan_Tmp.obj" model file after completion, and only the color (texture) map file is retained; the purpose here is to replace the RC.obj model in the "RC.rcproj" project file with the "Scan_Tmp.obj" model to bake the color (texture) map.

[0085] In step 107, the normal map and the ambient occlusion map of the second three-dimensional model after UV addition are obtained according to the shape and texture of the second three-dimensional model.

[0086] In the embodiment of the application, the "Scan_Low.obj" model and the "Scan.stl" model are imported into the software (such as MarmosetToolbag) used for baking at the same time, the shape and texture information of the "Scan.stl" model is baked into the normal map and the ambient occlusion map of the "Scan_Low.obj" model in the form of an image, and the image size is 8K (8192x8192) pixels; the purpose here is to use the high-precision model scanned before, that is, the second three-dimensional model, to bake the normal and ambient occlusion map of the reduced model (the second model after reduction), and increase the surface details of the low-face model.

[0087] In step 108, the metalness map and the roughness map corresponding to the target real object are obtained.

[0088] The embodiment of the present application imports the "Scan_Low.obj" model using 3D software, connects the color (texture) map and the ambient occlusion map, obtains a 3D map model, checks whether there are stretched, blurred and other parts in the 3D map model, and if so, determines that the color map needs to be repaired. The software (Substance 3D Painter) is used to repair the parts with flaws in the color map, and after completion, the 8K (8192x8192) pixel metal degree map and roughness map are baked according to the original appearance of the stone. All the maps are exported and sequentially named as Scan_Low_Color.png (color / texture), Scan_Low_AO.png (ambient occlusion), Scan_Low_Normal.png (normal), Scan_Low_Metallic.png (metal degree), and Scan_Low_Roughness.png (roughness). The low-face-number 3D model is made to reflect the surface concave-convex, gloss, roughness and other characteristics of the original stone through the baking of the normal, metal degree and roughness, thereby improving the accuracy and the degree of reality of the model.

[0089] In step 109, a realistic 3D model of the target object is obtained according to the second 3D model after UV addition, the color map, the normal map, the ambient occlusion map, the metal degree map and the roughness map.

[0090] All the maps are saved as a 1K (1024x1024) size map using image software, the 1K map is added with "1K" at the end of the file name, and the 8K map is added with "8K" at the end of the file name. The "Scan_Low.obj" model is opened using 3D software (blender), the color (texture) map, the ambient occlusion map, the metal degree map, the roughness map and the normal map are connected, the related parameters are adjusted according to the actual situation of the stone, such as reflection and glossiness, the size of the connected map can be connected according to the scene situation required by the object, if used for high definition, 8K can be used, and if used for general scene, 1K can be used, and finally output into a final 3D model format file.

[0091] The embodiment of the present application uses a low-face-number 3D model to maximize the concave-convex and surface details of the 3D model represented by a high-face-number model, so as to achieve a high-precision realistic degree, retains the advantages of 3D scanning models and photo modeling, replaces the model established in the photo modeling software with the 3D scanning model after reducing the surface and splitting UV for color (texture) baking, avoids the respective shortcomings of 3D scanning models and photo modeling, makes the production of low-face-number high-precision realistic 3D models more convenient, solves the problems of size, color (texture) clarity and lack of map in pure photo modeling, and solves the problems of lack or distortion of color (texture) in the model formed by pure 3D scanning.

[0092] The embodiment of the present application adopts a 3D scanning model to replace a low-face-number model generated in a photo modeling software after UV reduction and splitting, the model with color (texture) map formed by baking has high fidelity, the UV layout of the map is full layout, and the clarity of the model color map is improved; the present application adopts the key step of high-low model baking normal, environment light shading map, so that the final model can faithfully reflect the concave-convex details of the object surface; the baking of metal degree and roughness map in the process of the present application can make the final model more realistically reflect the gloss and rough characteristics of the object surface, so that the model is more realistic and has higher fidelity.

[0093] The fidelity 3D model obtained by the embodiment of the present application adopts a low-face-number model (the face number is determined according to the complexity of the object, generally reduced to not more than 20000 faces), which can easily cope with WEB or program display, reduce loading time and computing resource consumption.

[0094] The fidelity 3D model obtained by the embodiment of the present application has a size consistent with the original object 1:1 (error ±0.1mm), the 3D model UV layout is full frame layout, and the map precision is 8K (8192x8192) and 1K (1024x1024) two kinds of pixels, which can meet the needs of high-definition 8K and general low-precision 1K.

[0095] The map of the fidelity 3D model obtained by the embodiment of the present application conforms to the PBR map specification, has color (texture), environment light shading, metal degree, roughness, normal, and can faithfully reflect whether the surface of the original object has concave-convex, gloss, roughness and the degree of object characteristics such as size.

[0096] The technical features of the above embodiments can be combined arbitrarily, in order to make the description simple, not all possible combinations of the technical features in the above embodiments are described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0097] The principles and implementation modes of the present application are described by specific examples in this paper, the above embodiment is only used to help understand the method and core idea of the present application; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In view of the above, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method of low face count 3D modeling of an object with high precision photorealism, characterized in that, The method includes the following steps: Photo modeling is used to create a 3D model of the target object and extract its color features, thus obtaining the first 3D model and color features. A second three-dimensional model is obtained by using a three-dimensional scanning method to create a three-dimensional model of the target object. The second 3D model is subjected to a face reduction process to obtain a face-reduced second 3D model. The face count of the face-reduced second 3D model is the same as that of the first 3D model. The UVs of the second 3D model after reducing the number of faces are split and flattened to obtain the second 3D model after adding UVs. The second 3D model after adding UVs is then copied to obtain a temporary 3D model. The size and position of the temporary 3D model are adjusted so that the size and position of the temporary 3D model are consistent with the size and position of the first 3D model, thus obtaining an aligned temporary 3D model; The aligned temporary 3D model is baked using the color features to generate a color texture. Based on the shape and texture of the second 3D model before the reduction, obtain the normal map and ambient occlusion map of the second 3D model after UV addition; Obtain the metallic texture and roughness texture corresponding to the target object; Based on the second 3D model after UV addition, the color map, the normal map, the ambient occlusion map, the metallic map, and the roughness map, a realistic 3D model of the target object is obtained; The aligned temporary 3D model is baked using the aforementioned color features to generate a color texture, specifically including: In 3D software, the first 3D model is replaced with an aligned temporary 3D model, and texture baking is performed using the color features to generate a color map.

2. The method of low facet count 3D modeling of high fidelity to an object of claim 1, wherein, The first 3D model is obtained by using photogrammetry to create a 3D model of the target object, specifically including: Take M photos of the target object using a preset shooting trajectory, and take at least one photo of the color calibration card; M is greater than or equal to 360, and the preset shooting trajectory is a trajectory that wraps around the target object in a spherical, hemispherical or cylindrical shape; Color calibration is performed on M target object photos based on the color calibration card photos; Each of the M color-calibrated target object photos is processed to remove the main subject, resulting in M ​​target object foreground images. Import M images of the target object into 3D modeling software to perform 3D modeling and obtain the first 3D model.

3. The method of low facet count 3D modeling of high fidelity to an object of claim 2, wherein, Import M images of the target object's foreground into 3D modeling software for 3D modeling to obtain the first 3D model, which includes: Import M images of the target object into the 3D modeling software and check whether the images are fully recognized. If complete, then generate the first 3D model based on M images of the target object foreground; If the model is incomplete, the first 3D model will not be generated.

4. The method of low facet count 3D modeling of high fidelity to an object of claim 1, wherein, The first 3D model is obtained by creating a 3D model of the target object using photogrammetry. Subsequent steps include: Reduce the number of faces in the first 3D model.

5. The method of low facet count 3D modeling of high fidelity to an object of claim 1, wherein, The UVs of the reduced-face second 3D model are split and flattened to obtain the UV-added second 3D model, specifically including: UV decomposition is performed on the second 3D model after the reduction of the polygons to obtain a UV model that fills the entire frame; Add color materials to the UV model to obtain a second 3D model with added UVs.

6. The method of low facet count 3D modeling of high fidelity to an object of claim 1, wherein, The metal degree map and the roughness map corresponding to the target real object are acquired, and the previous process further includes: obtaining a 3D map model according to the second three-dimensional model after UV addition, the color map and the ambient light shading map; performing defect detection on the color map based on the 3D map model, and repairing the defect part in the color map.

7. The method of low facet count 3D modeling of high fidelity to an object of claim 1, wherein, The color map, the normal map, the ambient light shading map, the metal degree map and the roughness map all have the same preset size.

Images