CVPR 2025 Papers

We present LT3SD, a novel latent diffusion model for large-scale 3D scene generation. Recent advances in diffusion models have shown impressive results in 3D object generation, but are limited in spatial extent and quality when extended to 3D scenes. To generate complex and diverse 3D scene structures, we introduce a latent tree representation to effectively encode both lower-frequency geometry and higher-frequency detail in a coarse-to-fine hierarchy. We can then learn a generative diffusion process in this latent 3D scene space, modeling the latent components of a scene at each resolution level. To synthesize large-scale scenes with varying sizes, we train our diffusion model on scene patches and synthesize arbitrary-sized output 3D scenes through shared diffusion generation across multiple scene patches. Through extensive experiments, we demonstrate the efficacy and benefits of LT3SD for large-scale, high-quality unconditional 3D scene generation and for probabilistic completion for partial scene observations.

While consumer displays increasingly support more than 10 stops of dynamic range, most image assets -- such as internet photographs and generative AI content -- remain limited to 8-bit low dynamic range (LDR), constraining their utility across high dynamic range (HDR) applications. Currently, no generative model can produce high-bit, high-dynamic range content in a generalizable way. Existing LDR-to-HDR conversion methods often struggle to produce photorealistic details and physically-plausible dynamic range in the clipped areas. We introduce LEDiff, a method that enables a generative model with HDR content generation through latent space fusion inspired by image-space exposure fusion techniques. It also functions as an LDR-to-HDR converter, expanding the dynamic range of existing low-dynamic range images. Our approach uses a small HDR dataset to enable a pretrained diffusion model to recover detail and dynamic range in clipped highlights and shadows. LEDiff brings HDR capabilities to existing generative models and converts any LDR image to HDR, creating photorealistic HDR outputs for image generation, image-based lighting (HDR environment map generation), and photographic effects such as depth of field simulation, where linear HDR data is essential for realistic quality.

Visual Emotion Recognition (VER) is a critical yet challenging task aimed at inferring emotional states of individuals based on visual cues. However, existing works focus on single domains, e.g., realistic images or stickers, limiting VER models' cross-domain generalizability. To fill this gap, we introduce an Unsupervised Cross-Domain Visual Emotion Recognition (UCDVER) task, which aims to generalize visual emotion recognition from the source domain (e.g., realistic images) to the low-resource target domain (e.g., stickers) in an unsupervised manner. Compared to the conventional unsupervised domain adaptation problems, UCDVER presents two key challenges: a significant emotional expression variability and an affective distribution shift. To mitigate these issues, we propose the Knowledge-aligned Counterfactual-enhancement Diffusion Perception (KCDP) framework. Specifically, KCDP leverages a VLM to align emotional representations in a shared knowledge space and guides diffusion models for improved visual affective perception. Furthermore, a Counterfactual-Enhanced Language-image Emotional Alignment (CLIEA) method generates high-quality pseudo-labels for the target domain. Extensive experiments demonstrate that our model surpasses SOTA models in both perceptibility and generalization, e.g., gaining 12% improvements over SOTA VER model TGCA-PVT.

Diffusion models have achieved great success in generating 2D images. However, the quality and generalizability of 3D content generation remain limited. State-of-the-art methods often require large-scale 3D assets for training, which are challenging to collect. In this work, we introduce Kiss3DGen (Keep It Simple and Straightforward in 3D Generation), an efficient framework for generating, editing, and enhancing 3D objects by repurposing a well-trained 2D image diffusion model for 3D generation. Specifically, we fine-tune a diffusion model to generate "3D Bundle Image", a tiled representation composed of multi-view images and their corresponding normal maps. The normal maps are then used to reconstruct a 3D mesh, and the multi-view images provide texture mapping, resulting in a complete 3D model. This simple method effectively transforms the 3D generation problem into a 2D image generation task, maximizing the utilization of knowledge in pretrained diffusion models. Furthermore, we demonstrate that our Kiss3DGen model is compatible with various diffusion model techniques, enabling advanced features such as 3D editing, mesh and texture enhancement, etc. Through extensive experiments, we demonstrate the effectiveness of our approach, showcasing its ability to produce high-quality 3D models efficiently.

Recent studies have explored combining different LoRAs to jointly generate learned style and content. However, existing methods either fail to effectively preserve both the original subject and style simultaneously or require additional training. In this paper, we argue that the intrinsic properties of LoRA can effectively guide diffusion models in merging learned subject and style. Building on this insight, we propose K-LoRA, a simple yet effective training-free LoRA fusion approach. In each attention layer, K-LoRA compares the Top-K elements in each LoRA to be fused, determining which LoRA to select for optimal fusion. This selection mechanism ensures that the most representative features of both subject and style are retained during the fusion process, effectively balancing their contributions. Experimental results demonstrate that the proposed method effectively integrates the subject and style information learned by the original LoRAs, outperforming state-of-the-art training-based approaches in both qualitative and quantitative results.

Existing state-of-the-art feature matchers capture long-range dependencies with Transformers but are hindered by high spatial complexity,leading to demanding training and high-latency inference.Striking a better balance between performance and efficiency remains a critical challenge in feature matching.Inspired by the linear complexity \mathcal O (N) of Mamba, we propose an ultra-lightweight Mamba-based matcher, named JamMa, which converges on a single GPU and achieves an impressive performance-efficiency balance in inference.To unlock the potential of Mamba for feature matching,we propose Joint Mamba with a scan-merge strategy named JEGO, which enables:(1) Joint scan of two images to achieve high-frequency mutual interaction, (2) Efficient scan with skip steps to reduce sequence length, (3) Global receptive field, and (4) Omnidirectional feature representation.With the above properties, the JEGO strategy significantly outperforms the scan-merge strategies proposed in VMamba and EVMamba in the feature matching task.Compared to attention-based sparse and semi-dense matchers, JamMa demonstrates a notably superior balance between performance and efficiency,delivering better performance with less than 50% of the parameters and FLOPs.

Concept bottleneck models (CBM) aim to produce inherently interpretable models that rely on human-understandable concepts for their predictions. However, existing approaches to design interpretable generative models based on CBMs are not yet efficient and scalable, as they require expensive generative model training from scratch as well as real images with labor-intensive concept supervision. To address these challenges, we present two novel and low-cost methods to build interpretable generative models through post-hoc techniques and we name our approaches: concept-bottleneck autoencoder (CB-AE) and concept controller (CC). Our proposed approaches enable efficient and scalable training without the need of real data and require only minimal to no concept supervision. Additionally, our methods generalize across modern generative model families including generative adversarial networks and diffusion models. We demonstrate the superior interpretability and steerability of our methods on numerous standard datasets like CelebA, CelebA-HQ, and CUB with large improvements (average 25%) over the prior work, while being 4-15x faster to train. Finally, a large-scale user study is performed to validate the interpretability and steerability of our methods.

Recent advances in 3D human-aware generation have made significant progress. However, existing methods still struggle with generating novel Human Object Interaction (HOI) from text, particularly for open-set objects. We identify three main challenges of this task: precise human-object relation reasoning, affordance parsing for any object, and detailed human interaction pose synthesis aligning description and object geometry. In this work, we propose a novel zero-shot 3D HOI generation framework without training on specific datasets, leveraging the knowledge from large-scale pre-trained models. Specifically, the human-object relations are inferred from large language models (LLMs) to initialize object properties and guide the optimization process. Then we utilize a pre-trained 2D image diffusion model to parse unseen objects and extract contact points, avoiding the limitations imposed by existing 3D asset knowledge. The initial human pose is generated by sampling multiple hypotheses through multi-view SDS based on the input text and object geometry. Finally, we introduce a detailed optimization to generate fine-grained, precise, and natural interaction, enforcing realistic 3D contact between the 3D object and the involved body parts, including hands in grasping. This is achieved by distilling human-level feedback from LLMs to capture detailed human-object relations from the text instruction. Extensive experiments validate the effectiveness of our approach compared to prior works, particularly in terms of the fine-grained nature of interactions and the ability to handle open-set 3D objects. Project page: jinluzhang.site/projects/interactanything.

Predicting the dynamics of interacting objects is essential for both humans and intelligent systems. However, existing approaches are limited to simplified, toy settings and lack generalizability to complex, real-world environments. Recent advances in generative models have enabled the prediction of state transitions based on interventions, but focus on generating a single future state which neglects the continuous dynamics resulting from the interaction. To address this gap, we propose InterDyn, a novel framework that generates videos of interactive dynamics given an initial frame and a control signal encoding the motion of a driving object or actor. Our key insight is that large video generation models can act as both neural renderers and implicit physics "simulators", having learned interactive dynamics from large-scale video data. To effectively harness this capability, we introduce an interactive control mechanism that conditions the video generation process on the motion of the driving entity. Qualitative results demonstrate that InterDyn generates plausible, temporally consistent videos of complex object interactions while generalizing to unseen objects. Quantitative evaluations show that InterDyn outperforms baselines that focus on static state transitions. This work highlights the potential of leveraging video generative models as implicit physics engines.

Although natural language instructions offer an intuitive way to guide automated image editing, deep-learning models often struggle to achieve high-quality results, largely due to the difficulty of creating large, high-quality training datasets. To do this, previous approaches have typically relied on text-to-image (T2I) generative models to produce pairs of original and edited images that simulate the input/output of an instruction-guided image- editing model. However, these image pairs often fail to align with the specified edit instructions due to the limitations of T2I models, which negatively impacts models trained on such datasets. To address this, we present Instruct-CLIP (I-CLIP), a selfsupervised method that learns the semantic changes between original and edited images to refine and better align the instructions in existing datasets. Furthermore, we adapt Instruct-CLIP to handle noisy latent images and diffusion timesteps so that it can be used to train latent diffusion models (LDMs) and efficiently enforce alignment between the edit instruction and the image changes in latent space at any step of the diffusion pipeline. We use Instruct-CLIP to correct the InstructPix2Pix dataset and get over 120K refined samples we then use to fine-tune their model, guided by our novel I- CLIP-based loss function. The resulting model can produce edits that are more aligned with the given instructions. Our code and dataset are available at https://github.com/SherryXTChen/Instruct-CLIP.git.

We propose a generative technique to edit 3D shapes, represented as meshes, NeRFs, or Gaussian Splats, in ~3 seconds, without the need for running an SDS type of optimization.Our key insight is to cast 3D editing as a multiview image inpainting problem, as this representation is generic and can be mapped back to any 3D representation using the bank of available Large Reconstruction Models. We explore different fine-tuning strategies to obtain both multiview generation and inpainting capabilities within the same diffusion model. In particular, the design of the inpainting mask is an important factor of training an inpainting model, and we propose several masking strategies to mimic the types of edits a user would perform on a 3D shape. Our approach takes 3D generative editing from hours to seconds and produces higher-quality results compared to previous works.

Neural networks have revolutionized numerous fields with their exceptional performance, yet they remain susceptible to adversarial attacks through subtle perturbations. While diffusion-based purification methods like DiffPure offer promising defense mechanisms, their computational overhead presents a significant practical limitation. In this paper, we introduce One Step Control Purification (OSCP), a novel defense framework that achieves robust adversarial purification in a single Neural Function Evaluation (NFE) within diffusion models. We propose Gaussian Adversarial Noise Distillation (GAND) as the distillation objective and Controlled Adversarial Purification (CAP) as the inference pipeline, which makes OSCP demonstrate remarkable efficiency while maintaining defense efficacy. Our proposed GAND addresses a fundamental tension between consistency distillation and adversarial perturbation, bridging the gap between natural and adversarial manifolds in the latent space, while remaining computationally efficient through Parameter-Efficient Fine-Tuning (PEFT) methods such as LoRA, eliminating the high computational budget request from full parameter fine-tuning. The CAP guides the purification process through the unlearnable edge detection operator calculated by the input image as an extra prompt, effectively preventing the purified images from deviating from their original appearance when using large purification steps. Our experimental results on ImageNet showcase OSCP's superior performance, achieving a 74.19% defense success rate with merely 0.1s per purification --- a 100-fold speedup compared to conventional approaches.

We present Infinity, a Bitwise Visual AutoRegressive Modeling capable of generating high-resolution, photorealistic images following language instruction. Infinity refactors visual autoregressive model under a bitwise token prediction framework with an infinite-vocabulary classifier and bitwise self-correction mechanism. By theoretically expanding the tokenizer vocabulary size to infinity in Transformer, our method significantly unleashes powerful scaling capabilities to infinity compared to vanilla VAR. Extensive experiments indicate Infinity outperforms AutoRegressive Text-to-Image models by large margins, matches or surpasses leading diffusion models. Without extra optimization, Infinity generates a 1024x1024 image in 0.8s, 2.6xfaster than SD3-Medium, making it the fastest Text-to-Image model. Models and codes will be released to promote the further exploration of Infinity for visual generation.

Without using explicit reward, direct preference optimization (DPO) employs paired human preference data to fine-tune generative models, a method that has garnered considerable attention in large language models (LLMs). However, exploration of aligning text-to-image (T2I) diffusion models with human preferences remains limited. In comparison to supervised fine-tuning, existing methods that align diffusion model suffer from low training efficiency and subpar generation quality due to the long Markov chain process and the intractability of the reverse process. To address these limitations, we introduce DDIM-InPO, an efficient method for direct preference alignment of diffusion models. Our approach conceptualizes diffusion model as a single-step generative model, allowing us to fine-tune the outputs of specific latent variables selectively. In order to accomplish this objective, we first assign implicit rewards to any latent variable directly via a reparameterization technique. Then we construct an Inversion technique to estimate appropriate latent variables for preference optimization. This modification process enables the diffusion model to only fine-tune the outputs of latent variables that have a strong correlation with the preference dataset. Experimental results indicate that our DDIM-InPO achieves state-of-the-art performance with just 400 steps of fine-tuning, surpassing all preference aligning baselines for T2I diffusion models in human preference evaluation tasks.

Diffusion models have recently achieved success in solving Bayesian inverse problems with learned data priors. Current methods build on top of the diffusion sampling process, where each denoising step makes small modifications to samples from the previous step. However, this process struggles to correct errors from earlier sampling steps, leading to worse performance in complicated nonlinear inverse problems, such as phase retrieval. To address this challenge, we propose a new method called Decoupled Annealing Posterior Sampling (DAPS) that relies on a novel noise annealing process. Specifically, we decouple consecutive steps in a diffusion sampling trajectory, allowing them to vary considerably from one another while ensuring their time-marginals anneal to the true posterior as we reduce noise levels. This approach enables the exploration of a larger solution space, improving the success rate for accurate reconstructions. We demonstrate that DAPS significantly improves sample quality and stability across multiple image restoration tasks, particularly in complicated nonlinear inverse problems.

Variational Autoencoder (VAE) aims to compress pixel data into low-dimensional latent space, playing an important role in OpenAI's Sora and other latent video diffusion generation models. While most existing video VAEs inflate a pre-trained image VAE into the 3D causal structure for temporal-spatial compression, this paper presents two astonishing findings: (1) The initialization from a well-trained image VAE with the same latent dimensions is not an optimal scheme. (2) The adoption of causal reasoning leads to unequal information interactions and unbalanced performance between frames. To alleviate these problems, we propose a keyframe-based temporal compression (KTC) architecture and a group causal convolution (GCConv) module to further improve video VAE (IV-VAE). Specifically, the KTC architecture divides the latent space into two branches, in which one half completely inherits the compression prior of keyframes from a lower-dimension image VAE while the other half involves temporal compression through 3D group causal convolution, reducing temporal-spatial conflicts and accelerating the convergence speed of video VAE. The GCConv in the above 3D half uses standard convolution within each frame group to ensure inter-frame equivalence, and employs causal logical padding between groups to maintain flexibility in processing variable frame video. Extensive experiments on five benchmarks demonstrate the SOTA video reconstruction and generation abilities of our IV-VAE.

The proliferation of text-to-image diffusion models (T2I DMs) has led to an increased presence of AI-generated images in daily life. However, biased T2I models can generate content with specific tendencies, potentially influencing people's perceptions. Intentional exploitation of these biases risks conveying misleading information to the public. Current research on bias primarily addresses explicit biases with recognizable visual patterns, such as skin color and gender. This paper introduces a novel form of implicit bias that lacks explicit visual features but can manifest in diverse ways across various semantic contexts. This subtle and versatile nature makes this bias challenging to detect, easy to propagate, and adaptable to a wide range of scenarios. We further propose an implicit bias injection attack framework (IBI-Attacks) against T2I diffusion models by precomputing a general bias direction in the prompt embedding space and adaptively adjusting it based on different inputs. Our attack module can be seamlessly integrated into pre-trained diffusion models in a plug-and-play manner without direct manipulation of user input or model retraining. Extensive experiments validate the effectiveness of our scheme in introducing bias through subtle and diverse modifications while preserving the original semantics.The strong concealment and transferability of our attack across various scenarios further underscore the significance of our approach.

In-domain generation aims to perform a variety of tasks within a specific domain, such as unconditional generation, text-to-image, image editing, 3D generation, and more. Early research typically required training specialized generators for each unique task and domain, often relying on fully-labeled data. Motivated by the powerful generative capabilities and broad applications of diffusion models, we are driven to explore leveraging label-free data to empower these models for in-domain generation.Fine-tuning a pre-trained generative model on domain data is an intuitive but challenging way and often requires complex manual hyper-parameter adjustments since the limited diversity of the training data can easily disrupt the model's original generative capabilities.To address this challenge, we propose a guidance-decoupled prior preservation mechanism to achieve high generative quality and controllability by image-only data, inspired by preserving the pre-trained model from a denoising guidance perspective.We decouple domain-related guidance from the conditional guidance used in classifier-free guidance mechanisms to preserve open-world control guidance and unconditional guidance from the pre-trained model. We further propose an efficient domain knowledge learning technique to train an additional text-free UNet copy to predict domain guidance.Besides, we theoretically illustrate a multi-guidance in-domain generation pipeline for a variety of generative tasks, leveraging multiple guidances from distinct diffusion models and conditions. Extensive experiments demonstrate the superiority of our method in domain-specific synthesis and its compatibility with various diffusion-based control methods and applications.

Sketch colorization plays an important role in animation and digital illustration production tasks. However, existing methods still meet problems in that text-guided methods fail to provide accurate color and style reference, hint-guided methods still involve manual operation, and image-referenced methods are prone to cause artifacts. To address these limitations, we propose a diffusion-based framework inspired by real-world animation production workflows. Our approach leverages the sketch as the spatial guidance and an RGB image as the color reference, and separately extracts foreground and background from the reference image with spatial masks. Particularly, we introduce a split cross-attention mechanism with LoRA (Low-Rank Adaptation) modules. They are trained separately with foreground and background regions to control the corresponding embeddings for keys and values in cross-attention. This design allows the diffusion model to integrate information from foreground and background independently, preventing interference and eliminating the spatial artifacts. During inference, we design switchable inference modes for diverse use scenarios by changing modules activated in the framework. Extensive qualitative and quantitative experiments, along with user studies, demonstrate our advantages over existing methods in generating high-qualigy artifact-free results with geometric mismatched references. Ablation studies further confirm the effectiveness of each component. Codes are available at https://github.com/tellurion-kanata/colorizeDiffusion.