The paper presents a method for generating 3D staging—human poses, lighting, and camera setup—directly from affective textual descriptions. It builds a dataset of 11,911 text–staging pairs derived from 2,328 figurative paintings, reconstructing SMPL bodies, estimating illumination, and recovering camera parameters. A flow‑matching transformer is trained to produce variable‑size scenes and multiple staging alternatives, achieving a 32.2% retrieval R@1 on held‑out prompts, outperforming a CLIP‑based baseline.
arXiv:2608.28823v1 Announce Type: cross Abstract: Artists coordinate human pose, illumination, and camera placement to convey narrative and emotion, but existing generative methods typically model th...
arXiv:2608.22329v1 Announce Type: cross Abstract: Emotion-aware artistic image generation requires a model to satisfy semantic content, artistic style, and target emotion simultaneously. The key chal...
The paper introduces a new sequential text‑to‑scene generation framework that jointly produces scene layouts and objects. It proposes a 3D Autoregressive Diffusion model (3D‑ARD+) that first generates coarse 3D latents conditioned on text and the current scene, then refines them into fine‑grained object geometry and appearance. A large dataset of 230K indoor scenes with paired text instructions is curated for training, and the model demonstrates the ability to place objects according to complex spatial and semantic text prompts.
arXiv:2609.24215v1 Announce Type: new Abstract: Although text-to-image models can accurately depict subjects and scenes, creators still struggle to specify the fine-grained emotions an image should c...
arXiv:2606. 09846v1 Announce Type: cross Abstract: Visual art remains largely inaccessible to blind and low-vision (BLV) audiences due to brief or absent alt-text, which rarely conveys the sensory, spatial, or emotional qualities of an artwork.
Despite remarkable progress in text-guided image editing, generative models frequently fail to preserve visual object consistency, defined as the preservation of a subject's key attributes throughout the editing process. We address this limitation through three contributions.
CogCanvas is a new benchmark for multi-subject reference-based image generation, featuring 1,952 curated reference images of 100 celebrities, 115 objects/fashion items, and 29 real-world backgrounds. It generates 1,361 compositional prompts with 2–5 people, using a pipeline that includes DINOv2 deduplication, aesthetic filtering, and automated graph derivation for interaction and positioning. The benchmark evaluates three tasks—reference-based multi-human-object generation, text-to-image compositional generation, and reference retrieval—under a six-axis protocol, and introduces BG‑Sim and Attr‑VQA metrics to assess background fidelity and attribute binding.
VTR-Bench is a new benchmark designed to evaluate how well video generation models render text within scenes. It includes 300 prompts across five real-world scenarios such as advertisements and scientific videos, and uses an automated pipeline with human alignment to assess text fidelity and scene/motion requirements. Experiments on 11 state‑of‑the‑art models show that even the best performer has a word error rate of 0.250, underscoring widespread challenges in visual text rendering.
The paper introduces Chat-Edit-3D++ (CE3D++), an interactive 3D and 4D scene editing system that uses a Hash-Atlas network to separate 2D editing from 3D reconstruction. CE3D++ employs a large language model to accept arbitrary textual input, interpret user intent, and autonomously invoke appropriate visual models, enabling multi‑round dialogue and diverse editing effects. The approach is extended to monocular 4D scenes by adding motion constraints and a trajectory dataset, allowing a smaller LLM to schedule up to 30 visual tools accurately.
ShotFinder introduces a new benchmark for open‑domain video shot retrieval, formalizing editing requirements as keyframe‑oriented shot descriptions and adding five controllable constraints—temporal order, color, visual style, audio, and resolution. The benchmark comprises 1,210 high‑quality YouTube samples across 20 themes, generated with large models and verified by humans. A three‑stage retrieval pipeline—query expansion via video imagination, candidate video retrieval, and description‑guided shot localization—shows a notable performance gap to humans, especially for color and visual style constraints.
Prior work on aesthetic composition typically produces a single aesthetically pleasing crop, overlooking the narrative value of composing multiple shots from one scene. In practice, multi-shot composition is critical for downstream creative workflows: commercial posters often require multiple crops with different emphases (e.
CompArt introduces a new approach to aesthetic alignment in text-to-image generation by using the Principles of Art (PoA) such as Balance, Rhythm, and Emphasis to define explicit compositional constraints. The authors create a large dataset of 80,032 WikiArt images, each annotated with PoA analyses generated by a multimodal LLM, and present ArtDapter, a lightweight adapter that steers a pretrained diffusion model along ten PoA dimensions while preserving semantic fidelity. Experiments demonstrate that CompArt outperforms strong baselines in adhering to PoA controls under a dual evaluation protocol.