Daily Papers

Video generation has recently made striking visual progress, but maintaining coherent object motion and interactions remains difficult. We trace two practical bottlenecks: (i) human-provided motion hints (e.g., small 2D maps) often collapse to too few effective tokens after encoding, weakening guidance; and (ii) optimizing for appearance and motion in a single head can favor texture over temporal consistency. We present STANCE, an image-to-video framework that addresses both issues with two simple components. First, we introduce Instance Cues -- a pixel-aligned control signal that turns sparse, user-editable hints into a dense 2.5D (camera-relative) motion field by averaging per-instance flow and augmenting with monocular depth over the instance mask. This reduces depth ambiguity compared to 2D arrow inputs while remaining easy to use. Second, we preserve the salience of these cues in token space with Dense RoPE, which tags a small set of motion tokens (anchored on the first frame) with spatial-addressable rotary embeddings. Paired with joint RGB \(+\) auxiliary-map prediction (segmentation or depth), our model anchors structure while RGB handles appearance, stabilizing optimization and improving temporal coherence without requiring per-frame trajectory scripts.

We introduce JointNet, a novel neural network architecture for modeling the joint distribution of images and an additional dense modality (e.g., depth maps). JointNet is extended from a pre-trained text-to-image diffusion model, where a copy of the original network is created for the new dense modality branch and is densely connected with the RGB branch. The RGB branch is locked during network fine-tuning, which enables efficient learning of the new modality distribution while maintaining the strong generalization ability of the large-scale pre-trained diffusion model. We demonstrate the effectiveness of JointNet by using RGBD diffusion as an example and through extensive experiments, showcasing its applicability in a variety of applications, including joint RGBD generation, dense depth prediction, depth-conditioned image generation, and coherent tile-based 3D panorama generation.

We present JointDiT, a diffusion transformer that models the joint distribution of RGB and depth. By leveraging the architectural benefit and outstanding image prior of the state-of-the-art diffusion transformer, JointDiT not only generates high-fidelity images but also produces geometrically plausible and accurate depth maps. This solid joint distribution modeling is achieved through two simple yet effective techniques that we propose, namely, adaptive scheduling weights, which depend on the noise levels of each modality, and the unbalanced timestep sampling strategy. With these techniques, we train our model across all noise levels for each modality, enabling JointDiT to naturally handle various combinatorial generation tasks, including joint generation, depth estimation, and depth-conditioned image generation by simply controlling the timesteps of each branch. JointDiT demonstrates outstanding joint generation performance. Furthermore, it achieves comparable results in depth estimation and depth-conditioned image generation, suggesting that joint distribution modeling can serve as a viable alternative to conditional generation. The project page is available at https://byungki-k.github.io/JointDiT/.

Recent advancements in diffusion models have set new benchmarks in image and video generation, enabling realistic visual synthesis across single- and multi-frame contexts. However, these models still struggle with efficiently and explicitly generating 3D-consistent content. To address this, we propose World-consistent Video Diffusion (WVD), a novel framework that incorporates explicit 3D supervision using XYZ images, which encode global 3D coordinates for each image pixel. More specifically, we train a diffusion transformer to learn the joint distribution of RGB and XYZ frames. This approach supports multi-task adaptability via a flexible inpainting strategy. For example, WVD can estimate XYZ frames from ground-truth RGB or generate novel RGB frames using XYZ projections along a specified camera trajectory. In doing so, WVD unifies tasks like single-image-to-3D generation, multi-view stereo, and camera-controlled video generation. Our approach demonstrates competitive performance across multiple benchmarks, providing a scalable solution for 3D-consistent video and image generation with a single pretrained model.

Beyond the superiority of the text-to-image diffusion model in generating high-quality images, recent studies have attempted to uncover its potential for adapting the learned semantic knowledge to visual perception tasks. In this work, instead of translating a generative diffusion model into a visual perception model, we explore to retain the generative ability with the perceptive adaptation. To accomplish this, we present Zippo, a unified framework for zipping the color and transparency distributions into a single diffusion model by expanding the diffusion latent into a joint representation of RGB images and alpha mattes. By alternatively selecting one modality as the condition and then applying the diffusion process to the counterpart modality, Zippo is capable of generating RGB images from alpha mattes and predicting transparency from input images. In addition to single-modality prediction, we propose a modality-aware noise reassignment strategy to further empower Zippo with jointly generating RGB images and its corresponding alpha mattes under the text guidance. Our experiments showcase Zippo's ability of efficient text-conditioned transparent image generation and present plausible results of Matte-to-RGB and RGB-to-Matte translation.

Recently, the RGB images and point clouds fusion methods have been proposed to jointly estimate 2D optical flow and 3D scene flow. However, as both conventional RGB cameras and LiDAR sensors adopt a frame-based data acquisition mechanism, their performance is limited by the fixed low sampling rates, especially in highly-dynamic scenes. By contrast, the event camera can asynchronously capture the intensity changes with a very high temporal resolution, providing complementary dynamic information of the observed scenes. In this paper, we incorporate RGB images, Point clouds and Events for joint optical flow and scene flow estimation with our proposed multi-stage multimodal fusion model, RPEFlow. First, we present an attention fusion module with a cross-attention mechanism to implicitly explore the internal cross-modal correlation for 2D and 3D branches, respectively. Second, we introduce a mutual information regularization term to explicitly model the complementary information of three modalities for effective multimodal feature learning. We also contribute a new synthetic dataset to advocate further research. Experiments on both synthetic and real datasets show that our model outperforms the existing state-of-the-art by a wide margin. Code and dataset is available at https://npucvr.github.io/RPEFlow.

Although COLMAP has long remained the predominant method for camera parameter optimization in static scenes, it is constrained by its lengthy runtime and reliance on ground truth (GT) motion masks for application to dynamic scenes. Many efforts attempted to improve it by incorporating more priors as supervision such as GT focal length, motion masks, 3D point clouds, camera poses, and metric depth, which, however, are typically unavailable in casually captured RGB videos. In this paper, we propose a novel method for more accurate and efficient camera parameter optimization in dynamic scenes solely supervised by a single RGB video. Our method consists of three key components: (1) Patch-wise Tracking Filters, to establish robust and maximally sparse hinge-like relations across the RGB video. (2) Outlier-aware Joint Optimization, for efficient camera parameter optimization by adaptive down-weighting of moving outliers, without reliance on motion priors. (3) A Two-stage Optimization Strategy, to enhance stability and optimization speed by a trade-off between the Softplus limits and convex minima in losses. We visually and numerically evaluate our camera estimates. To further validate accuracy, we feed the camera estimates into a 4D reconstruction method and assess the resulting 3D scenes, and rendered 2D RGB and depth maps. We perform experiments on 4 real-world datasets (NeRF-DS, DAVIS, iPhone, and TUM-dynamics) and 1 synthetic dataset (MPI-Sintel), demonstrating that our method estimates camera parameters more efficiently and accurately with a single RGB video as the only supervision.

We present a dataset of large-scale indoor spaces that provides a variety of mutually registered modalities from 2D, 2.5D and 3D domains, with instance-level semantic and geometric annotations. The dataset covers over 6,000m2 and contains over 70,000 RGB images, along with the corresponding depths, surface normals, semantic annotations, global XYZ images (all in forms of both regular and 360{\deg} equirectangular images) as well as camera information. It also includes registered raw and semantically annotated 3D meshes and point clouds. The dataset enables development of joint and cross-modal learning models and potentially unsupervised approaches utilizing the regularities present in large-scale indoor spaces. The dataset is available here: http://3Dsemantics.stanford.edu/

We present IDC-Net (Image-Depth Consistency Network), a novel framework designed to generate RGB-D video sequences under explicit camera trajectory control. Unlike approaches that treat RGB and depth generation separately, IDC-Net jointly synthesizes both RGB images and corresponding depth maps within a unified geometry-aware diffusion model. The joint learning framework strengthens spatial and geometric alignment across frames, enabling more precise camera control in the generated sequences. To support the training of this camera-conditioned model and ensure high geometric fidelity, we construct a camera-image-depth consistent dataset with metric-aligned RGB videos, depth maps, and accurate camera poses, which provides precise geometric supervision with notably improved inter-frame geometric consistency. Moreover, we introduce a geometry-aware transformer block that enables fine-grained camera control, enhancing control over the generated sequences. Extensive experiments show that IDC-Net achieves improvements over state-of-the-art approaches in both visual quality and geometric consistency of generated scene sequences. Notably, the generated RGB-D sequences can be directly feed for downstream 3D Scene reconstruction tasks without extra post-processing steps, showcasing the practical benefits of our joint learning framework. See more at https://idcnet-scene.github.io.

2D portrait animation has experienced significant advancements in recent years. Much research has utilized the prior knowledge embedded in large generative diffusion models to enhance high-quality image manipulation. However, most methods only focus on generating RGB images as output, and the co-generation of consistent visual plus 3D output remains largely under-explored. In our work, we propose to jointly learn the visual appearance and depth simultaneously in a diffusion-based portrait image generator. Our method embraces the end-to-end diffusion paradigm and introduces a new architecture suitable for learning this conditional joint distribution, consisting of a reference network and a channel-expanded diffusion backbone. Once trained, our framework can be efficiently adapted to various downstream applications, such as facial depth-to-image and image-to-depth generation, portrait relighting, and audio-driven talking head animation with consistent 3D output.

Recent approaches to jointly reconstruct 3D humans and objects from a single RGB image represent 3D shapes with template-based or coarse models, which fail to capture details of loose clothing on human bodies. In this paper, we introduce a novel implicit approach for jointly reconstructing realistic 3D clothed humans and objects from a monocular view. For the first time, we model both the human and the object with an implicit representation, allowing to capture more realistic details such as clothing. This task is extremely challenging due to human-object occlusions and the lack of 3D information in 2D images, often leading to poor detail reconstruction and depth ambiguity. To address these problems, we propose a novel attention-based neural implicit model that leverages image pixel alignment from both the input human-object image for a global understanding of the human-object scene and from local separate views of the human and object images to improve realism with, for example, clothing details. Additionally, the network is conditioned on semantic features derived from an estimated human-object pose prior, which provides 3D spatial information about the shared space of humans and objects. To handle human occlusion caused by objects, we use a generative diffusion model that inpaints the occluded regions, recovering otherwise lost details. For training and evaluation, we introduce a synthetic dataset featuring rendered scenes of inter-occluded 3D human scans and diverse objects. Extensive evaluation on both synthetic and real-world datasets demonstrates the superior quality of the proposed human-object reconstructions over competitive methods.

Despite great progress in multimodal tracking, these trackers remain too heavy and expensive for resource-constrained devices. To alleviate this problem, we propose LightFC-X, a family of lightweight convolutional RGB-X trackers that explores a unified convolutional architecture for lightweight multimodal tracking. Our core idea is to achieve lightweight cross-modal modeling and joint refinement of the multimodal features and the spatiotemporal appearance features of the target. Specifically, we propose a novel efficient cross-attention module (ECAM) and a novel spatiotemporal template aggregation module (STAM). The ECAM achieves lightweight cross-modal interaction of template-search area integrated feature with only 0.08M parameters. The STAM enhances the model's utilization of temporal information through module fine-tuning paradigm. Comprehensive experiments show that our LightFC-X achieves state-of-the-art performance and the optimal balance between parameters, performance, and speed. For example, LightFC-T-ST outperforms CMD by 4.3% and 5.7% in SR and PR on the LasHeR benchmark, which it achieves 2.6x reduction in parameters and 2.7x speedup. It runs in real-time on the CPU at a speed of 22 fps. The code is available at https://github.com/LiYunfengLYF/LightFC-X.

Recent advancements in vision-language pre-training (e.g. CLIP) have shown that vision models can benefit from language supervision. While many models using language modality have achieved great success on 2D vision tasks, the joint representation learning of 3D point cloud with text remains under-explored due to the difficulty of 3D-Text data pair acquisition and the irregularity of 3D data structure. In this paper, we propose a novel Text4Point framework to construct language-guided 3D point cloud models. The key idea is utilizing 2D images as a bridge to connect the point cloud and the language modalities. The proposed Text4Point follows the pre-training and fine-tuning paradigm. During the pre-training stage, we establish the correspondence of images and point clouds based on the readily available RGB-D data and use contrastive learning to align the image and point cloud representations. Together with the well-aligned image and text features achieved by CLIP, the point cloud features are implicitly aligned with the text embeddings. Further, we propose a Text Querying Module to integrate language information into 3D representation learning by querying text embeddings with point cloud features. For fine-tuning, the model learns task-specific 3D representations under informative language guidance from the label set without 2D images. Extensive experiments demonstrate that our model shows consistent improvement on various downstream tasks, such as point cloud semantic segmentation, instance segmentation, and object detection. The code will be available here: https://github.com/LeapLabTHU/Text4Point

We present PVSeRF, a learning framework that reconstructs neural radiance fields from single-view RGB images, for novel view synthesis. Previous solutions, such as pixelNeRF, rely only on pixel-aligned features and suffer from feature ambiguity issues. As a result, they struggle with the disentanglement of geometry and appearance, leading to implausible geometries and blurry results. To address this challenge, we propose to incorporate explicit geometry reasoning and combine it with pixel-aligned features for radiance field prediction. Specifically, in addition to pixel-aligned features, we further constrain the radiance field learning to be conditioned on i) voxel-aligned features learned from a coarse volumetric grid and ii) fine surface-aligned features extracted from a regressed point cloud. We show that the introduction of such geometry-aware features helps to achieve a better disentanglement between appearance and geometry, i.e. recovering more accurate geometries and synthesizing higher quality images of novel views. Extensive experiments against state-of-the-art methods on ShapeNet benchmarks demonstrate the superiority of our approach for single-image novel view synthesis.

We present CARTO, a novel approach for reconstructing multiple articulated objects from a single stereo RGB observation. We use implicit object-centric representations and learn a single geometry and articulation decoder for multiple object categories. Despite training on multiple categories, our decoder achieves a comparable reconstruction accuracy to methods that train bespoke decoders separately for each category. Combined with our stereo image encoder we infer the 3D shape, 6D pose, size, joint type, and the joint state of multiple unknown objects in a single forward pass. Our method achieves a 20.4% absolute improvement in mAP 3D IOU50 for novel instances when compared to a two-stage pipeline. Inference time is fast and can run on a NVIDIA TITAN XP GPU at 1 HZ for eight or less objects present. While only trained on simulated data, CARTO transfers to real-world object instances. Code and evaluation data is available at: http://carto.cs.uni-freiburg.de

Reconstructing interacting hands from a single RGB image is a very challenging task. On the one hand, severe mutual occlusion and similar local appearance between two hands confuse the extraction of visual features, resulting in the misalignment of estimated hand meshes and the image. On the other hand, there are complex spatial relationship between interacting hands, which significantly increases the solution space of hand poses and increases the difficulty of network learning. In this paper, we propose a decoupled iterative refinement framework to achieve pixel-alignment hand reconstruction while efficiently modeling the spatial relationship between hands. Specifically, we define two feature spaces with different characteristics, namely 2D visual feature space and 3D joint feature space. First, we obtain joint-wise features from the visual feature map and utilize a graph convolution network and a transformer to perform intra- and inter-hand information interaction in the 3D joint feature space, respectively. Then, we project the joint features with global information back into the 2D visual feature space in an obfuscation-free manner and utilize the 2D convolution for pixel-wise enhancement. By performing multiple alternate enhancements in the two feature spaces, our method can achieve an accurate and robust reconstruction of interacting hands. Our method outperforms all existing two-hand reconstruction methods by a large margin on the InterHand2.6M dataset.

We present a real-time approach for multi-person 3D motion capture at over 30 fps using a single RGB camera. It operates successfully in generic scenes which may contain occlusions by objects and by other people. Our method operates in subsequent stages. The first stage is a convolutional neural network (CNN) that estimates 2D and 3D pose features along with identity assignments for all visible joints of all individuals.We contribute a new architecture for this CNN, called SelecSLS Net, that uses novel selective long and short range skip connections to improve the information flow allowing for a drastically faster network without compromising accuracy. In the second stage, a fully connected neural network turns the possibly partial (on account of occlusion) 2Dpose and 3Dpose features for each subject into a complete 3Dpose estimate per individual. The third stage applies space-time skeletal model fitting to the predicted 2D and 3D pose per subject to further reconcile the 2D and 3D pose, and enforce temporal coherence. Our method returns the full skeletal pose in joint angles for each subject. This is a further key distinction from previous work that do not produce joint angle results of a coherent skeleton in real time for multi-person scenes. The proposed system runs on consumer hardware at a previously unseen speed of more than 30 fps given 512x320 images as input while achieving state-of-the-art accuracy, which we will demonstrate on a range of challenging real-world scenes.

Implicit neural SLAM has achieved remarkable progress recently. Nevertheless, existing methods face significant challenges in non-ideal scenarios, such as motion blur or lighting variation, which often leads to issues like convergence failures, localization drifts, and distorted mapping. To address these challenges, we propose EN-SLAM, the first event-RGBD implicit neural SLAM framework, which effectively leverages the high rate and high dynamic range advantages of event data for tracking and mapping. Specifically, EN-SLAM proposes a differentiable CRF (Camera Response Function) rendering technique to generate distinct RGB and event camera data via a shared radiance field, which is optimized by learning a unified implicit representation with the captured event and RGBD supervision. Moreover, based on the temporal difference property of events, we propose a temporal aggregating optimization strategy for the event joint tracking and global bundle adjustment, capitalizing on the consecutive difference constraints of events, significantly enhancing tracking accuracy and robustness. Finally, we construct the simulated dataset DEV-Indoors and real captured dataset DEV-Reals containing 6 scenes, 17 sequences with practical motion blur and lighting changes for evaluations. Experimental results show that our method outperforms the SOTA methods in both tracking ATE and mapping ACC with a real-time 17 FPS in various challenging environments. Project page: https://delinqu.github.io/EN-SLAM.

Unlike color photography images, which are consistently encoded into RGB channels, biological images encompass various modalities, where the type of microscopy and the meaning of each channel varies with each experiment. Importantly, the number of channels can range from one to a dozen and their correlation is often comparatively much lower than RGB, as each of them brings specific information content. This aspect is largely overlooked by methods designed out of the bioimage field, and current solutions mostly focus on intra-channel spatial attention, often ignoring the relationship between channels, yet crucial in most biological applications. Importantly, the variable channel type and count prevent the projection of several experiments to a unified representation for large scale pre-training. In this study, we propose ChAda-ViT, a novel Channel Adaptive Vision Transformer architecture employing an Inter-Channel Attention mechanism on images with an arbitrary number, order and type of channels. We also introduce IDRCell100k, a bioimage dataset with a rich set of 79 experiments covering 7 microscope modalities, with a multitude of channel types, and channel counts varying from 1 to 10 per experiment. Our proposed architecture, trained in a self-supervised manner, outperforms existing approaches in several biologically relevant downstream tasks. Additionally, it can be used to bridge the gap for the first time between assays with different microscopes, channel numbers or types by embedding various image and experimental modalities into a unified biological image representation. The latter should facilitate interdisciplinary studies and pave the way for better adoption of deep learning in biological image-based analyses. Code and Data to be released soon.

We present a pipeline for generating defurnished replicas of indoor spaces represented as textured meshes and corresponding multi-view panoramic images. To achieve this, we first segment and remove furniture from the mesh representation, extend planes, and fill holes, obtaining a simplified defurnished mesh (SDM). This SDM acts as an ``X-ray'' of the scene's underlying structure, guiding the defurnishing process. We extract Canny edges from depth and normal images rendered from the SDM. We then use these as a guide to remove the furniture from panorama images via ControlNet inpainting. This control signal ensures the availability of global geometric information that may be hidden from a particular panoramic view by the furniture being removed. The inpainted panoramas are used to texture the mesh. We show that our approach produces higher quality assets than methods that rely on neural radiance fields, which tend to produce blurry low-resolution images, or RGB-D inpainting, which is highly susceptible to hallucinations.

Articulated objects are prevalent in daily life. Understanding their kinematic structure and reconstructing them have numerous applications in embodied AI and robotics. However, current methods require carefully captured data for training or inference, preventing practical, scalable, and generalizable reconstruction of articulated objects. We focus on reconstruction of an articulated object from a casually captured RGBD video shot with a hand-held camera. A casually captured video of an interaction with an articulated object is easy to acquire at scale using smartphones. However, this setting is quite challenging, as the object and camera move simultaneously and there are significant occlusions as the person interacts with the object. To tackle these challenges, we introduce a coarse-to-fine framework that infers joint parameters and segments movable parts of the object from a dynamic RGBD video. To evaluate our method under this new setting, we build a 20times larger synthetic dataset of 784 videos containing 284 objects across 11 categories. We compare our approach with existing methods that also take video as input. Experiments show that our method can reconstruct synthetic and real articulated objects across different categories from dynamic RGBD videos, outperforming existing methods significantly.

Image inpainting is an important task in computer vision. As admirable methods are presented, the inpainted image is getting closer to reality. However, the result is still not good enough in the reconstructed texture and structure based on human vision. Although recent advances in computer hardware have enabled the development of larger and more complex models, there is still a need for lightweight models that can be used by individuals and small-sized institutions. Therefore, we propose a lightweight model that combines a specialized transformer with a traditional convolutional neural network (CNN). Furthermore, we have noticed most researchers only consider three primary colors (RGB) in inpainted images, but we think this is not enough. So we propose a new loss function to intensify color details. Extensive experiments on commonly seen datasets (Places2 and CelebA) validate the efficacy of our proposed model compared with other state-of-the-art methods. Index Terms: HSV color space, image inpainting, joint attention, stripe window, transformer

While methods for monocular depth estimation have made significant strides on standard benchmarks, zero-shot metric depth estimation remains unsolved. Challenges include the joint modeling of indoor and outdoor scenes, which often exhibit significantly different distributions of RGB and depth, and the depth-scale ambiguity due to unknown camera intrinsics. Recent work has proposed specialized multi-head architectures for jointly modeling indoor and outdoor scenes. In contrast, we advocate a generic, task-agnostic diffusion model, with several advancements such as log-scale depth parameterization to enable joint modeling of indoor and outdoor scenes, conditioning on the field-of-view (FOV) to handle scale ambiguity and synthetically augmenting FOV during training to generalize beyond the limited camera intrinsics in training datasets. Furthermore, by employing a more diverse training mixture than is common, and an efficient diffusion parameterization, our method, DMD (Diffusion for Metric Depth) achieves a 25\% reduction in relative error (REL) on zero-shot indoor and 33\% reduction on zero-shot outdoor datasets over the current SOTA using only a small number of denoising steps. For an overview see https://diffusion-vision.github.io/dmd

Remote sensing change detection, identifying changes between scenes of the same location, is an active area of research with a broad range of applications. Recent advances in multimodal self-supervised pretraining have resulted in state-of-the-art methods which surpass vision models trained solely on optical imagery. In the remote sensing field, there is a wealth of overlapping 2D and 3D modalities which can be exploited to supervise representation learning in vision models. In this paper we propose Contrastive Surface-Image Pretraining (CSIP) for joint learning using optical RGB and above ground level (AGL) map pairs. We then evaluate these pretrained models on several building segmentation and change detection datasets to show that our method does, in fact, extract features relevant to downstream applications where natural and artificial surface information is relevant.

Estimating robot pose from RGB images is a crucial problem in computer vision and robotics. While previous methods have achieved promising performance, most of them presume full knowledge of robot internal states, e.g. ground-truth robot joint angles. However, this assumption is not always valid in practical situations. In real-world applications such as multi-robot collaboration or human-robot interaction, the robot joint states might not be shared or could be unreliable. On the other hand, existing approaches that estimate robot pose without joint state priors suffer from heavy computation burdens and thus cannot support real-time applications. This work introduces an efficient framework for real-time robot pose estimation from RGB images without requiring known robot states. Our method estimates camera-to-robot rotation, robot state parameters, keypoint locations, and root depth, employing a neural network module for each task to facilitate learning and sim-to-real transfer. Notably, it achieves inference in a single feed-forward pass without iterative optimization. Our approach offers a 12-time speed increase with state-of-the-art accuracy, enabling real-time holistic robot pose estimation for the first time. Code and models are available at https://github.com/Oliverbansk/Holistic-Robot-Pose-Estimation.

We introduce ScanBot, a novel dataset designed for instruction-conditioned, high-precision surface scanning in robotic systems. In contrast to existing robot learning datasets that focus on coarse tasks such as grasping, navigation, or dialogue, ScanBot targets the high-precision demands of industrial laser scanning, where sub-millimeter path continuity and parameter stability are critical. The dataset covers laser scanning trajectories executed by a robot across 12 diverse objects and 6 task types, including full-surface scans, geometry-focused regions, spatially referenced parts, functionally relevant structures, defect inspection, and comparative analysis. Each scan is guided by natural language instructions and paired with synchronized RGB, depth, and laser profiles, as well as robot pose and joint states. Despite recent progress, existing vision-language action (VLA) models still fail to generate stable scanning trajectories under fine-grained instructions and real-world precision demands. To investigate this limitation, we benchmark a range of multimodal large language models (MLLMs) across the full perception-planning-execution loop, revealing persistent challenges in instruction-following under realistic constraints.

Image-generation diffusion models have been fine-tuned to unlock new capabilities such as image-editing and novel view synthesis. Can we similarly unlock image-generation models for visuomotor control? We present GENIMA, a behavior-cloning agent that fine-tunes Stable Diffusion to 'draw joint-actions' as targets on RGB images. These images are fed into a controller that maps the visual targets into a sequence of joint-positions. We study GENIMA on 25 RLBench and 9 real-world manipulation tasks. We find that, by lifting actions into image-space, internet pre-trained diffusion models can generate policies that outperform state-of-the-art visuomotor approaches, especially in robustness to scene perturbations and generalizing to novel objects. Our method is also competitive with 3D agents, despite lacking priors such as depth, keypoints, or motion-planners.