Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
GuideFlow3D: Optimization-Guided Rectified Flow For Appearance Transfer
Authors: Sayan Deb Sarkar, Sinisa Stekovic, Vincent Lepetit, Iro Armeni
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our experiments show that our approach successfully transfers texture and geometric details to the input 3D asset, outperforming baselines both qualitatively and quantitatively. We also show that traditional metrics are not suitable for evaluating the task due to their inability of focusing on local details and comparing dissimilar inputs, in absence of ground truth data. We thus evaluate appearance transfer quality with a GPT-based system objectively ranking outputs, ensuring robust and human-like assessment, as further confirmed by our user study. |
| Researcher Affiliation | Academia | Sayan Deb Sarkar Stanford University Sinisa Stekovic ENPC, IP Paris Vincent Lepetit ENPC, IP Paris Iro Armeni Stanford University |
| Pseudocode | No | The paper describes methods and processes using mathematical formulations and descriptive text, but it does not include a distinct block labeled as "Pseudocode" or "Algorithm" with structured steps. |
| Open Source Code | Yes | We make our code and benchmark publicly available on our project website. |
| Open Datasets | Yes | Since there are no publicly available datasets for our task of transferring appearance across different shapes, we create a benchmark for evaluation. First, for input mesh, we generate synthetic objects using procedural models from [62]. Second, for the appearance mesh and images, we leverage the ABO dataset [16], with the text captions provided by [72]. |
| Dataset Splits | Yes | Using simple and complex meshes for the different categories, we create 250 input-appearance object pairs for each of our 4 experimental setups: (i) simple-complex intra-category, (ii) simple-complex inter-category, (iii) complex-complex intra-category, and, (iv) complex-complex inter-category. |
| Hardware Specification | Yes | All experiments are conducted on a single NVIDIA RTX 4090 GPU. |
| Software Dependencies | No | Our implementation is based on the pre-trained models and configurations from Trellis [72] and Part Field [41]. While these tools are mentioned, specific version numbers for software dependencies like Python, PyTorch, or CUDA are not provided. |
| Experiment Setup | Yes | We run Guide Flow3D for single-instance optimization interleaved with rectified flow sampling over 300 steps. Optimization is performed using Adam W with a learning rate of 5e-4. All experiments are conducted on a single NVIDIA RTX 4090 GPU. We use identical optimization settings across all conditioning types to ensure fairness and consistency. |