# Carve3D: Improving Multi-view Reconstruction Consistency for Diffusion Models with RL Finetuning

arXiv: https://arxiv.org/abs/2312.13980 · HTML: https://mlsanity.com/paper/2312.13980/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Placed in 3D, point cloud and scene generation, one of 14 directions in Diffusion models.
- Builds on 37 of its 54 references that are in the map.
- First posted on arXiv 21 Dec 2023.

## Placed in

- Diffusion models / [3D, point cloud and scene generation](https://mlsanity.com/diffusion/3d-point-cloud-and-scene-generation/)

## Builds on (most cited first, 30 of 37)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [Classifier-Free Diffusion Guidance](https://mlsanity.com/paper/2207.12598/) (2022)
- [SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis](https://mlsanity.com/paper/2307.01952/) (2023)
- [LAION-5B: An open large-scale dataset for training next generation image-text models](https://mlsanity.com/paper/2210.08402/) (2022)
- [Instant Neural Graphics Primitives with a Multiresolution Hash Encoding](https://mlsanity.com/paper/2201.05989/) (2022)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [DreamFusion: Text-to-3D using 2D Diffusion](https://mlsanity.com/paper/2209.14988/) (2022)
- [Scaling Instruction-Finetuned Language Models](https://mlsanity.com/paper/2210.11416/) (2022)
- [Zero-1-to-3: Zero-shot One Image to 3D Object](https://mlsanity.com/paper/2303.11328/) (2023)
- [TensoRF: Tensorial Radiance Fields](https://mlsanity.com/paper/2203.09517/) (2022)
- [Training Diffusion Models with Reinforcement Learning](https://mlsanity.com/paper/2305.13301/) (2023)
- [MVDream: Multi-view Diffusion for 3D Generation](https://mlsanity.com/paper/2308.16512/) (2023)
- [LRM: Large Reconstruction Model for Single Image to 3D](https://mlsanity.com/paper/2311.04400/) (2023)
- [DreamGaussian: Generative Gaussian Splatting for Efficient 3D Content Creation](https://mlsanity.com/paper/2309.16653/) (2023)
- [Point-E: A System for Generating 3D Point Clouds from Complex Prompts](https://mlsanity.com/paper/2212.08751/) (2022)
- [RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback](https://mlsanity.com/paper/2309.00267/) (2023)
- [Score Jacobian Chaining: Lifting Pretrained 2D Diffusion Models for 3D Generation](https://mlsanity.com/paper/2212.00774/) (2022)
- [One-2-3-45: Any Single Image to 3D Mesh in 45 Seconds without Per-Shape Optimization](https://mlsanity.com/paper/2306.16928/) (2023)
- [Zero123++: a Single Image to Consistent Multi-view Diffusion Base Model](https://mlsanity.com/paper/2310.15110/) (2023)
- [DPOK: Reinforcement Learning for Fine-tuning Text-to-Image Diffusion Models](https://mlsanity.com/paper/2305.16381/) (2023)
- [Directly Fine-Tuning Diffusion Models on Differentiable Rewards](https://mlsanity.com/paper/2309.17400/) (2023)
- [Aligning Text-to-Image Models using Human Feedback](https://mlsanity.com/paper/2302.12192/) (2023)
- [Instant3D: Fast Text-to-3D with Sparse-View Generation and Large Reconstruction Model](https://mlsanity.com/paper/2311.06214/) (2023)

## Built on by (most cited first, 22 of 22)

- [Progressive Autoregressive Video Diffusion Models](https://mlsanity.com/paper/2410.08151/) (2024)
- [Amodal3R: Amodal 3D Reconstruction from Occluded 2D Images](https://mlsanity.com/paper/2503.13439/) (2025)
- [LRM-Zero: Training Large Reconstruction Models with Synthesized Data](https://mlsanity.com/paper/2406.09371/) (2024)
- [Ouroboros3D: Image-to-3D Generation via 3D-aware Recursive Diffusion](https://mlsanity.com/paper/2406.03184/) (2024)
- [3D-Adapter: Geometry-Consistent Multi-View Diffusion for High-Quality 3D Generation](https://mlsanity.com/paper/2410.18974/) (2024)
- [Flex3D: Feed-Forward 3D Generation with Flexible Reconstruction Model and Input View Curation](https://mlsanity.com/paper/2410.00890/) (2024)
- [Integrating Reinforcement Learning with Visual Generative Models: Foundations and Advances](https://mlsanity.com/paper/2508.10316/) (2025)
- [Nabla-R2D3: Effective and Efficient 3D Diffusion Alignment with 2D Rewards](https://mlsanity.com/paper/2506.15684/) (2025)
- [DreamCS: Geometry-Aware Text-to-3D Generation with Unpaired 3D Reward Supervision](https://mlsanity.com/paper/2506.09814/) (2025)
- [GECO: Generative Image-to-3D within a SECOnd](https://mlsanity.com/paper/2405.20327/) (2024)
- [GaMO: Geometry-aware Multi-view Diffusion Outpainting for Sparse-View 3D Reconstruction](https://mlsanity.com/paper/2512.25073/) (2025)
- [Voxify3D: Pixel Art Meets Volumetric Rendering](https://mlsanity.com/paper/2512.07834/) (2025)
- [MD-ProjTex: Texturing 3D Shapes with Multi-Diffusion Projection](https://mlsanity.com/paper/2504.02762/) (2025)
- [Refining Few-Step Text-to-Multiview Diffusion via Reinforcement Learning](https://mlsanity.com/paper/2505.20107/) (2025)
- [Text2Stereo: Repurposing Stable Diffusion for Stereo Generation with Consistency Rewards](https://mlsanity.com/paper/2506.05367/) (2025)
- [AlignCVC: Aligning Cross-View Consistency for Single-Image-to-3D Generation](https://mlsanity.com/paper/2506.23150/) (2025)
- [Track, Inpaint, Resplat: Subject-driven 3D and 4D Generation with Progressive Texture Infilling](https://mlsanity.com/paper/2510.23605/) (2025)
- [Stroke3D: Lifting 2D strokes into rigged 3D model via latent diffusion models](https://mlsanity.com/paper/2602.09713/) (2026)
- [GeoFlow: Enforcing Implicit Geometric Consistency in Video Generation](https://mlsanity.com/paper/2605.18365/) (2026)
- [Kaleidoscopic Background Attack: Disrupting Pose Estimation with Multi-Fold Radial Symmetry Textures](https://mlsanity.com/paper/2507.10265/) (2025)
- [RealMat: Realistic Materials with Diffusion and Reinforcement Learning](https://mlsanity.com/paper/2509.01134/) (2025)
- [Debiasing Diffusion Priors via 3D Attention for Consistent Gaussian Splatting](https://mlsanity.com/paper/2512.07345/) (2025)

## Abstract (arXiv)

Multi-view diffusion models, obtained by applying Supervised Finetuning (SFT) to text-to-image diffusion models, have driven recent breakthroughs in text-to-3D research. However, due to the limited size and quality of existing 3D datasets, they still suffer from multi-view inconsistencies and Neural Radiance Field (NeRF) reconstruction artifacts. We argue that multi-view diffusion models can benefit from further Reinforcement Learning Finetuning (RLFT), which allows models to learn from the data generated by themselves and improve beyond their dataset limitations during SFT. To this end, we introduce Carve3D, an improved RLFT algorithm coupled with a novel Multi-view Reconstruction Consistency (MRC) metric, to enhance the consistency of multi-view diffusion models. To measure the MRC metric on a set of multi-view images, we compare them with their corresponding NeRF renderings at the same camera viewpoints. The resulting model, which we denote as Carve3DM, demonstrates superior multi-view consistency and NeRF reconstruction quality than existing models. Our results suggest that pairing SFT with Carve3D's RLFT is essential for developing multi-view-consistent diffusion models, mirroring the standard Large Language Model (LLM) alignment pipeline. Our code, training and testing data, and video results are available at: https://desaixie.github.io/carve-3d.