基于简约线条图的三维模型三轴切片视图生成数据

在3D内容创作和3D打印领域，通过单张简约线条图生成三轴多切片视图。该技术可以将设计师绘制的单视图草图或概念图快速转化为三维模型在XYZ三轴上的切片视图，特别适合复杂结构和被遮挡区域的还原，大大提升了三维模型制作的效率和精度。这种技术不仅适用于工业原型设计和艺术创作，还能推动教育和科研中3D内容的便捷生成，为用户提供更高效的创作体验。1. 数据收集：在3D内容创作和3D打印场景中，采集设计师绘制的单视图简约线条图，并生成三维模型的多切片视图。使用内部数据采集方法，根据单张线条图获取对应的三维模型数据和三轴上的切片渲染图。每个三维模型包含了三维渲染图片和简约线条图。
2. 数据处理：处理阶段先对三维模型在XYZ三轴上进行多角度切片，生成相应的三轴切片简约线条图、三维切片渲染图。通过图像处理算法和三维点云转换，提取简约线条图的特征向量F_sketch，采用公式F_sketch = Encoder_sketch(Sketch_img)进行编码，其中Encoder_sketch为预训练的线条图编码器，Sketch_img为简约线条图。然后，切片图像通过另一个编码器Encoder_slice处理生成切片特征向量F_slice。最终将F_sketch和F_slice进行匹配处理，以确保切片图与线条图特征的一致性。
3. 模型构建：构建基于扩散模型，将简约线条图的特征作为输入，并生成三轴切片简约线条图。首先，通过公式Slice_gen = Decoder_slice(F_sketch)，利用线条图特征生成三轴切片的预测结果Slice_gen，其中Decoder_slice为三维切片图解码器。之后计算平均精度（Acc）以评估模型的预测效果。

In the fields of 3D content creation and 3D printing, this technology generates multi-slice views along the three XYZ axes from a single simplified line drawing. It can quickly convert single-view sketches or conceptual drawings created by designers into slice views of the 3D model on the XYZ three axes, which is particularly suitable for the restoration of complex structures and occluded areas, greatly improving the efficiency and accuracy of 3D model production. This technology is not only applicable to industrial prototyping and artistic creation, but also promotes the convenient generation of 3D content in education and scientific research, providing users with a more efficient creative experience.
1. Data Collection: In the scenarios of 3D content creation and 3D printing, collect single simplified line drawings created by designers, and generate multi-slice views of the corresponding 3D model. Use internal data collection methods to obtain the corresponding 3D model data and slice renderings along the XYZ three axes from a single line drawing. Each 3D model includes 3D rendered images and simplified line drawings.
2. Data Processing: The processing stage first performs multi-angle slicing of the 3D model along the XYZ three axes to generate corresponding simplified line drawings of the three-axis slices and 3D slice renderings. Through image processing algorithms and 3D point cloud conversion, extract the feature vector F_sketch of the simplified line drawing, and encode it using the formula F_sketch = Encoder_sketch(Sketch_img), where Encoder_sketch is a pre-trained line drawing encoder, and Sketch_img is the simplified line drawing. Then, the slice images are processed by another encoder Encoder_slice to generate the slice feature vector F_slice. Finally, perform matching processing on F_sketch and F_slice to ensure the consistency between the slice images and the line drawing features.
3. Model Construction: Build a diffusion model-based framework that takes the features of the simplified line drawing as input and generates simplified line drawings of the three-axis slices. First, use the formula Slice_gen = Decoder_slice(F_sketch) to generate the prediction results Slice_gen of the three-axis slices using the line drawing features, where Decoder_slice is a 3D slice map decoder. Afterwards, calculate the average accuracy (Acc) to evaluate the prediction effect of the model.

搜集汇总

数据集介绍

特点

该数据集包含6396条基于简约线条图生成的三维模型三轴切片视图数据，适用于3D内容创作和3D打印领域，能够通过单张简约线条图快速生成三维模型的三轴切片视图，提升模型制作的效率和精度。