Text-to-image Diffusion Models in Generative AI: A Survey

cs.HC 2026-07 accept novelty 7.0 of 10

A novel 6M-image Pixiv dataset shows open-source image generation has long-tail model usage, slow life cycles with version inertia, and surging multi-LoRA customization linked to higher engagement.

cs.LG 2025-06 conditional novelty 7.0 of 10

BadReward uses clean-label feature-collision images to poison CLIP-based reward models so that a text-to-image model produces target attributes (e.g., glasses, skin tone, blood) when the trigger phrase is present.

cs.MM 2026-08 conditional novelty 6.0 of 10

In emergency scenes, text-to-image models skew who appears and who helps, favoring men, middle-aged, and lighter-skinned people, and a soft-token tweak reduces the gap.

cs.SD 2025-09 conditional novelty 6.0 of 10

All chatbot-designed audio encoders failed to align with CLIP text embeddings and produced incoherent images, while showing a surprising architectural similarity across chatbots.

cs.CV 2025-06 conditional novelty 6.0 of 10

DiffRIS combines frozen Stable Diffusion and CLIP encoders with a new text adapter and decoder to set a new state-of-the-art mean IoU on three referring remote sensing image segmentation benchmarks.

cs.AI 2025-06 conditional novelty 6.0 of 10

A new taxonomy and multi-aspect evaluation framework for image generation unlearning, with a curated dataset, shows that ten existing unlearning methods perform poorly on preservation and robustness.

cs.CV 2025-05 conditional novelty 6.0 of 10

LGDM uses perceptual guidance during Stable Diffusion sampling and aggregates multi-scale, multi-timestep U-Net features to predict human-rated image quality, achieving state-of-the-art correlations on ten NR-IQA datasets.

eess.IV 2025-05 conditional novelty 6.0 of 10

Text2CT is a unified 3D diffusion model that generates 512x512x192 chest CT volumes from free-text clinical descriptions, outperforming prior text-to-CT baselines on FID, CLIP score, and data augmentation.

cs.CV 2025-02 conditional novelty 6.0 of 10

A training-free pipeline makes diffusion text-to-video generation run on an iPhone 15 Pro with quality close to GPU output, at the cost of slower generation.

cs.CV 2024-12 conditional novelty 6.0 of 10

ConceptCoSOD extracts a shared text embedding from an image group and guides co-salient object segmentation with it, outperforming five baselines on three clean and five corrupted datasets.

cs.CV 2024-12 conditional novelty 6.0 of 10

AsymRnR selectively reduces query and key/value tokens in video DiT attention to cut FLOPs and latency by 10 to 30 percent with minor or no VBench score change.

cs.CV 2024-12 conditional novelty 6.0 of 10

A new benchmark and fine-tuned models show that multi-task training improves average perceptual-similarity accuracy on known tasks but does not generalize to out-of-distribution perceptual tasks.

cs.IR 2024-12 conditional novelty 6.0 of 10

Needle generates AI-made query images from text, embeds them with an ensemble of visual models, and uses nearest-neighbor search to retrieve matching real images, beating zero-shot text-image baselines on complex queries.

cs.CV 2024-11 conditional novelty 6.0 of 10

SPAI uses spectral reconstruction similarity from a frozen masked-frequency ViT plus attention pooling to reach 91.0 average AUC for AI-generated image detection across 13 unseen generators.

cs.CV 2026-06 unverdicted novelty 5.0 of 10

JuZhou 1.0 is a 0.387B-parameter T2I diffusion model with 4-step inference achieving 0.69 GenEval, trained on 9M Chinese pairs using Sugon K100 accelerators and deployable on Android/iOS devices.

cs.LG 2025-12 conditional novelty 5.0 of 10

A concept trained on visually uniform images tends to dominate and suppress a second concept in multi-concept text-to-image generation, a failure the authors name DvD.

cs.CV 2025-12 conditional novelty 5.0 of 10

Natural-language geometry descriptions can be autoformalized into a custom geometry language and converted to coordinates by Monte Carlo optimization, yielding usable diagrams in seconds for about 81-85% of test problems.

cs.LG 2025-08 reject novelty 5.0 of 10

VIPER-R1 fine-tunes a vision-language model to read kinematic plots and propose symbolic equations, then refines them with symbolic regression, but its final metric is computed on the same data used for the refinement fit.

cs.CV 2025-08 unverdicted novelty 5.0 of 10

VideoGuard adds joint, motion-aware perturbations to videos to block unauthorized diffusion-model editing.

cs.CY 2025-05 conditional novelty 5.0 of 10

Ethics experts' definitions of five ethical theories, rendered as DALL-E 3 images and re-evaluated by the same experts, yield eight themes showing how morality, society, and learned associations shape and bias the mod...

cs.LG 2025-05 reject novelty 5.0 of 10

LGR-AD claims to improve text-to-image generation by coordinating multiple diffusion models via a graph convolutional network, but the loss function's KL term contradicts the stated goal of maximizing diversity.

cs.CV 2025-02 conditional novelty 5.0 of 10

Masking the image-feature dimensions most correlated with the style reference's content text reduces content leakage and improves text fidelity in text-to-image style transfer diffusion models.

cs.CL 2025-02 conditional novelty 5.0 of 10

Applying pairwise direct preference optimization to score distillation makes text-to-3D outputs better aligned with human preferences and more controllable.

cs.CV 2025-02 reject novelty 5.0 of 10

The authors report that singular value decomposition of diffusion latent codes reveals stable, order-mobile attribute directions and propose Attribute Vector Integration, a per-pair MLP-based editor that transfers tex...

cs.MM 2024-12 conditional novelty 5.0 of 10

An automatic music-visualizer pipeline that uses genre-guided image generation and audio-energy-controlled frame interpolation beats linear interpolation on a new synchrony metric.

cs.CV 2024-11 conditional novelty 5.0 of 10

Blending a safe-caption denoising branch with the original-prompt branch during early diffusion steps generates safer images without editing or erasing concepts.

cs.CV 2026-06 conditional novelty 4.5 of 10

A DAAM-based visual analytics workflow links step-resolved token attention trajectories, phase summaries, and spatial competition maps for Stable Diffusion-class models on a 60-prompt benchmark.

cs.MM 2025-09 conditional novelty 4.0 of 10

A video-processing pipeline created a 2.24 million-sample audio-image-text dataset, with text captions generated by BLIP from video frames.

cs.CV 2025-07 reject novelty 4.0 of 10

A 1,200-prompt benchmark across six world-knowledge domains reports that ten state-of-the-art text-to-video models average below 0.70 on a 0 to 1 scale for producing videos consistent with real-world knowledge.

cs.CV 2025-06 conditional novelty 4.0 of 10

A spatial-frequency feature fusion network with attention achieves improved accuracy on remote sensing image forgery detection and introduces a stable-diffusion-based benchmark.

cs.CV 2025-06 reject novelty 4.0 of 10

A new 150K-video 2D skeleton dataset with text captions and a diffusion model for single- and double-character motion generation, though the claimed FID-rewarded RL training is misrepresented.

cs.SD 2025-06 reject novelty 4.0 of 10

A proposed audio-native translation framework called MAST with fractional diffusion is described, but no evidence is given that it produces working translations.

cs.CV 2025-05 reject novelty 4.0 of 10

Rhet2Pix combines staged LLM prompt decomposition with a discounted PPO fine-tuning scheme for Stable Diffusion, claiming strong rhetorical text-to-image generation, but the quantitative evidence is circular and undefined.

eess.IV 2025-05 conditional novelty 4.0 of 10

Fine-tuning Stable Diffusion with MRI-text pairs yields plausible brain MRI images by field strength and modality, and synthetic images appear to improve a small MRI contrast classification task.

stat.ML 2025-02 conditional novelty 4.0 of 10

GDP predicts by sampling candidate outputs from a conditional generative model and selecting the one with the lowest average loss; its excess risk is bounded by generation error plus a vanishing sampling term.

cs.CV 2025-01 reject novelty 4.0 of 10

FlexMotion reports a latent diffusion model with a physics-aware multimodal autoencoder and a ControlNet-style module for controlling joint locations, contact forces, joint actuations, and muscle activations in genera...

cs.CV 2025-01 reject novelty 4.0 of 10

A new 10.5M-pair remote sensing dataset and a 1.3B diffusion model generate resolution-controlled satellite imagery from text, with large reported gains on the RSICD benchmark.

cs.CV 2024-12 reject novelty 4.0 of 10

CA-DPS approximates the covariance of the reverse diffusion process via a finite-difference Hessian estimate and uses it to improve posterior sampling for linear inverse problems.

cs.LG 2024-12 conditional novelty 4.0 of 10

A review of non-centralized continual learning that taxonomizes data-, model-, and device-level methods and benchmarks twelve federated continual learning methods on six datasets.

cs.GR 2025-07 reject novelty 3.0 of 10

A softmax and spatial attention fusion of SDXL base and refiner latents yields an ImageNet FID drop of about 1 point, but the result is not statistically supported.

cs.CV 2025-05 conditional novelty 3.0 of 10

A broad survey of text-to-image generation and editing research from 2021 to 2024, organized by architecture and comparison tables.

eess.IV 2025-05 conditional novelty 2.0 of 10

A literature review that organizes diffusion-model work for hyperspectral imaging into eight task categories and compiles comparative performance tables from prior papers.

cs.IR 2025-01 unverdicted novelty 1.0 of 10

A survey chapter proposing that generative AI reshapes information access through two paradigms, information generation and information synthesis.

cs.CV 2024-11 conditional novelty 1.0 of 10

A decade-spanning survey categorizes over 440 text-to-image papers by architecture, research problem, dataset, and evaluation metric.