Update remesh parameters in generate3d function for enhanced mesh quality

Update remesh iterations in generate3d function for optimized performance

Update remesh parameters in generate3d function for improved mesh generation

Update remesh parameters in generate3d function for improved mesh quality

Update remesh iterations and device handling in generate3d function

Optimize generate3d function for performance and clarity

Refactor mesh generation in generate3d function to streamline processing

Update remesh parameter in generate3d function to disable remeshing

Fix typo in remesh parameter in generate3d function

Refine remesh size in generate3d function for improved mesh detail

Adjust remesh size in generate3d function for improved mesh quality

Enhance color processing in generate3d function by applying clipping and gamma correction

Update vertex color assignment in generate3d function to utilize original colors

Comment out KDTree functionality in generate3d function to simplify color assignment process. Updated mesh creation to set vertex colors to None, removing reliance on nearest neighbor color mapping.

Update remesh iterations in generate3d function to improve mesh quality

Refactor color scaling and assignment in generate3d function

Refine color extraction in generate3d function for original mesh

Enhance color assignment in generate3d function by utilizing original vertex data

Refactor vertex handling in generate3d function for improved performance

Add color assignment to new vertices in generate3d function

Refactor mesh generation in app.py and update remeshing parameters in inference.py

Enhance mesh generation in app.py and inference.py with hole filling and color assignment

Enhance image preprocessing in app.py with improved background handling and color conversion

Refactor background removal and image preprocessing in app.py

Enhance image preprocessing in app.py with background removal and color conversion

Add DIS background remover integration and model download functionality

Refactor generate3d function in inference.py to streamline mesh export process. Removed manual MTL file handling and updated export calls to enhance code clarity and maintainability.

Refactor generate3d function in inference.py to improve mesh export process. Introduced temporary file handling for OBJ, MTL, and PNG outputs, ensuring proper texture mapping. Enhanced code readability by restructuring renderer initialization and file writing logic.

Refactor device handling in generate3d function to use CPU instead of CUDA for mesh processing, ensuring compatibility with non-GPU environments. This change improves flexibility in data handling and maintains the integrity of the mesh generation process.

Update device handling in TwoStagePipeline to default to 'cuda' and adjust model loading to use 'cpu' for compatibility. Ensure consistent device usage across stages and improve code clarity.

Enhance Mesh class documentation by adding missing line breaks in docstrings for improved readability. Update device handling in FlexiCubes and FlexiCubesGeometry classes to default to 'cuda', ensuring consistent device usage across the application. Refactor ImageDreamDiffusion class to assert mode validity and streamline camera matrix pre-computation.

Refactor generate3d function in inference.py to enhance 3D mesh generation. Introduced new image processing steps for color and XYZ data, improved triplane feature extraction, and updated mesh export logic to utilize temporary file handling. This streamlines the rendering process and maintains compatibility with GPU devices.

Update generate3d function to initialize nvdiffrast context using CUDA, enhancing rendering performance and compatibility with GPU devices.

Enhance generate3d function to initialize nvdiffrast context for mesh export, improving rendering capabilities. Updated export call to include rendering context and adjusted parameters for UVs and texture image, ensuring better integration with the model.

Update generate3d function to use a data dictionary for mesh export, improving compatibility with CRM model. Adjusted export call to include output directory and device parameters, while maintaining the conversion from .obj to .glb format.

Remove unused parameters from mesh export call in generate3d function, simplifying the method signature and improving code clarity.

Fix parameter naming in generate3d function for clarity and consistency. Updated rgb and xyz parameters to rgb_image and xyz_image respectively in the mesh export call.

Refactor 3D mesh generation in inference.py to streamline the process. Introduced ensure_dir function for output directory management, updated generate3d function to accept model and image inputs, and improved mesh export and conversion to GLB format. This enhances code clarity and maintains compatibility with Gradio display.

Update requirements.txt to include rembg library for background removal functionality, enhancing the application's capabilities in image processing.

Refactor background removal process in app.py to utilize rembg library, enhancing performance and simplifying the code. Update device handling to allow dynamic selection between CPU and CUDA, improving compatibility across different hardware configurations. Modify output format from OBJ to GLB for better integration with Gradio display.

Refactor texture baking in vertex_color_to_uv_textured_glb function to use a hybrid approach for barycentric calculations. This change enhances accuracy while maintaining performance, replacing vectorized operations with a loop-based method for improved precision in texture generation.

Refactor texture baking in vertex_color_to_uv_textured_glb function to use vectorized operations. This change improves performance by precomputing face data and applying vectorized barycentric tests, enhancing the efficiency of texture generation.

512

Ensure input tensor is on the same device as model weights in FrozenOpenCLIPEmbedder. This change improves compatibility and performance by preventing device mismatch during image encoding.

Enhance token handling in FrozenOpenCLIPEmbedder by adding device checks and conversions. This update ensures that input tokens are correctly processed as torch tensors on the appropriate device, improving compatibility and performance across different hardware configurations.

Refactor device handling across multiple modules to enforce CUDA usage explicitly. This change ensures consistent device allocation for model operations, enhancing performance on systems with GPU support while maintaining compatibility.

Update device handling across multiple modules to support automatic selection of CUDA or CPU based on availability. This change enhances compatibility and performance on systems with or without GPU support, ensuring consistent behavior in model loading and data processing.

Refactor attention module to improve xformers integration. Renamed availability flag to HAS_XFORMERS and added safe_memory_efficient_attention function for better handling of attention operations across devices. Updated related assertions and calls to ensure compatibility with systems lacking GPU support.

Update default device setting to CPU in encoder modules for improved compatibility across systems without GPU support.

Update default device setting to CPU in FlexiCubesGeometry and FlexiCubes classes for improved compatibility across systems without GPU support.