Diffusion models create data from noise by inverting the forward paths of data towards noise and have emerged as a powerful generative modeling technique for high-dimensional, perceptual data such as images and videos. Rectified flow is a recent generative model formulation that connects data and noise in a straight line. Despite its better theoretical properties and conceptual simplicity, it is not yet decisively established as standard practice. In this work, we improve existing noise sampling techniques for training rectified flow models by biasing them towards perceptually relevant scales. Through a large-scale study, we demonstrate the superior performance of this approach compared to established diffusion formulations for high-resolution text-to-image synthesis. Additionally, we present a novel transformer-based architecture for text-to-image generation that uses separate weights for the two modalities and enables a bidirectional flow of information between image and text tokens, improving text comprehension, typography, and human preference ratings. We demonstrate that this architecture follows predictable scaling trends and correlates lower validation loss to improved text-to-image synthesis as measured by various metrics and human evaluations. Our largest models outperform state-of-the-art models. Stability AI is considering making experimental data, code, and model weights publicly available.

We study the problem of inferring scene affordances by presenting a method for realistically inserting people into scenes. Given a scene image with a marked region and an image of a person, we insert the person into the scene while respecting the scene affordances. Our model can infer the set of realistic poses given the scene context, re-pose the reference person, and harmonize the composition. We set up the task in a self-supervised fashion by learning to repose humans in video clips. We train a large-scale diffusion model on a dataset of 2.4M video clips that produces diverse plausible poses while respecting the scene context. Given the learned human-scene composition, our model can also hallucinate realistic people and scenes when prompted without conditioning and also enables interactive editing. A quantitative evaluation shows that our method synthesizes more realistic human appearance and more natural human-scene interactions than prior work.

We introduce Programmatic Motion Concepts, a hierarchical motion representation for human actions that captures both low-level motion and high-level description as motion concepts. This representation enables human motion description, interactive editing, and controlled synthesis of novel video sequences within a single framework. We present an architecture that learns this concept representation from paired video and action sequences in a semi-supervised manner. The compactness of our representation also allows us to present a low-resource training recipe for data-efficient learning. By outperforming established baselines, especially in the small data regime, we demonstrate the efficiency and effectiveness of our framework for multiple applications.

Shape programs encode shape structures by representing object parts as subroutines and constructing the overall shape by composing these subroutines. This usually involves the reuse of subroutines for repeatable parts, enabling the modeling of correlations among shape elements such as geometric similarity. However, existing learning-based shape programs suffer from limited representation capacity, because they use coarse geometry representations such as geometric primitives and low-resolution voxel grids. Further, their training requires manually annotated ground-truth programs, which are expensive to attain. We address these limitations by proposing Shape Programs with Repeatable Implicit Parts (ProGRIP). Using implicit functions to represent parts, ProGRIP greatly boosts the representation capacity of shape programs while preserving the higher-level structure of repetitions and symmetry. Meanwhile, we free ProGRIP from any inaccessible supervised training via devising a matching-based unsupervised training objective. Our empirical studies show that ProGRIP outperforms existing structured representations in both shape reconstruction fidelity and segmentation accuracy of semantic parts.

Current approaches to video analysis of human motion focus on raw pixels or keypoints as the basic units of reasoning. We posit that adding higher-level motion primitives, which can capture natural coarser units of motion such as "backswing" or "follow-through", can be used to improve downstream analysis tasks. This higher level of abstraction can also capture key features, such as loops of repeated primitives, that are currently inaccessible at lower levels of representation. We therefore introduce Motion Programs, a neuro-symbolic, program-like representation that expresses motions as a composition of high-level primitives. We also present a system for automatically inducing motion programs from videos of human motion and for leveraging motion programs in video synthesis. Experiments show that motion programs can accurately describe a diverse set of human motions and the inferred programs contain semantically meaningful motion primitives, such as arm swings and jumping jacks. Our representation also benefits downstream tasks such as video interpolation and video prediction and outperforms off-the-shelf models. We further demonstrate how these programs can detect diverse kinds of repetitive motion and facilitate interactive video editing.

We consider the task of mapping pseudocode to executable code, assuming a one-to-one correspondence between lines of pseudocode and lines of code. Given test cases as a mechanism to validate programs, we search over the space of possible translations of the pseudocode to find a program that compiles and passes the test cases. While performing a best-first search, compilation errors constitute 88.7% of program failures. To better guide this search, we learn to predict the line of the program responsible for the failure and focus search over alternative translations of the pseudocode for that line. For evaluation, we collected the SPoC dataset (Search-based Pseudocode to Code) containing 18,356 C++ programs with human-authored pseudocode and test cases. Under a budget of 100 program compilations, performing search improves the synthesis success rate over using the top-one translation of the pseudocode from 25.6% to 44.7%.

Given a relational specification between Boolean inputs and outputs, the goal of Boolean functional synthesis is to synthesize each output as a function of the inputs such that the specification is met. In this paper, we first show that unless some hard conjectures in complexity theory are falsified, Boolean functional synthesis must generate large Skolem functions in the worst-case. Given this inherent hardness, what does one do to solve the problem? We present a two-phase algorithm, where the first phase is efficient both in terms of time and size of synthesized functions, and solves a large fraction of benchmarks. To explain this surprisingly good performance, we provide a sufficient condition under which the first phase must produce correct answers. When this condition fails, the second phase builds upon the result of the first phase, possibly requiring exponential time and generating exponential-sized functions in the worst-case. Detailed experimental evaluation shows our algorithm to perform better than other techniques for a large number of benchmarks.

We present a new approach for specifying and verifying resource utilization of higher-order functional programs that use lazy evaluation and memoization. In our approach, users can specify the desired resource bound as templates with numerical holes e.g. as steps ≤ ? * size(l) + ? in the contracts of functions. They can also express invariants necessary for establishing the bounds that may depend on the state of memoization. Our approach operates in two phases: first generating an instrumented first-order program that accurately models the higher-order control flow and the effects of memoization on resources using sets, algebraic datatypes and mutual recursion, and then verifying the contracts of the first-order program by producing verification conditions of the form ∃ ∀ using an extended assume/guarantee reasoning. We use our approach to verify precise bounds on resources such as evaluation steps and number of heap-allocated objects on 17 challenging data structures and algorithms. Our benchmarks, comprising of 5K lines of functional Scala code, include lazy mergesort, Okasaki's real-time queue and deque data structures that rely on aliasing of references to first-class functions; lazy data structures based on numerical representations such as the conqueue data structure of Scala's data-parallel library, cyclic streams, as well as dynamic programming algorithms such as knapsack and Viterbi. Our evaluations show that when averaged over all benchmarks the actual runtime resource consumption is 80% of the value inferred by our tool when estimating the number of evaluation steps, and is 88% for the number of heap-allocated objects.