Pinar Yanardag

I am a tenure-track assistant professor at Virginia Tech, Department of Computer Science where I lead 💎 GEMLAB. I am also a member of Sanghani Center for AI and Discovery Analytics.

Prior to Virginia Tech, I was a postdoc at MIT. I received my Ph.D. in Computer Science at Purdue. During my graduate studies, I also worked at Amazon (P13N team) and VMware. I am a Fulbright PhD Fellow and Google Anita Borg Memorial Scholar.

My research is published at top computer science conferences such as CVPR, ICCV, NeurIPS, and featured in mainstream media (e.g., The Washington Post, BBC, CNN) and magazines (e.g., Motherboard, Rolling Stone). For more information, see the Publications page.

My main research area is centered on the development of generative AI methods, targeting three key aspects:

Discovering and Leveraging Latent Capabilities for Efficient Control: My research uncovers and exploits internal representations of generative models to enable efficient, training-free control of image and video synthesis. I develop methods for fine-grained, controllable edits (Conform, CVPR’24, Fluxspace, CVPR’25, NoiseCLR, CVPR’24), and extend these controls to the temporal domain through video and motion editing (RAVE, CVPR’24), video diffusion models (Motionflow, AAAI’26) and dynamic view sysnthesis (Inverse DVS, NeurIPS’25).
Integrating Cross-Modal Capabilities for Advanced Generation Tasks: My research bridges vision and language by unifying the reasoning capabilities of VLMs/LLMs with the generative priors of visual models, enabling advanced tasks beyond either modality alone. I develop agentic frameworks (CREA, NeurIPS’25), harness VLMs for multi-modal explainability (DiffEx, CVPR’25), and propose RL-based methods for visual synthesis (C-DPO, NeurIPS’25).
Personalized, Human-Aligned, and Democratized Generative AI: My research moves beyond one-size-fits-all generative AI by developing methods for personalized content generation (LoRACLR, CVPR’25, CloRA, ICCV’25, LoRAShop, NeurIPS’25). In parallel, I propose techniques to democratize generative AI for creators, enabling accessible and intuitive creative control (Plot’n Polish, AAAI’26, Stylebreeder, NeurIPS’24).

Prior to joining Virginia Tech, I was CEO of AI Fiction, a creative design studio specializing in AI. Some of our work includes generative AI for HBO’s Westworld, for which I was a Creative Director nominee at 72nd Primetime Emmy Awards. I also co-founded GLITCH–the world’s first generative AI clothing line.

I’m passionate about helping the general public understand and appreciate generative models. At MIT, I launched the How to Generate (Almost) Anything project, which demystifies generative AI through collaboration with artists and artisans, and encourages broader dialogue about its future and everyday implications. I also taught “AI & Fashion” course at the London College of Fashion. See the Courses section for more information.

Please contact me via pinary at vt.edu.

news

Oct 10, 2025	Four papers are accepted to NeurIPS’25 (main)!
Apr 3, 2025	One paper is accepted to ICML 2025 as Oral!
Feb 3, 2025	Three papers are accepted to CVPR 2025!
Mar 3, 2024	Three papers (1 oral, 1 highlight) are accepted to CVPR’24!

selected publications

NeurIPS 2025

[Spotlight] LoRAShop: Training-Free Multi-Concept Image Generation and Editing with Rectified Flow Transformers

Yusuf Dalva, Hidir Yesiltepe, and Pinar Yanardag

The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS) 2025

Abstract PDF Website

We introduce LoRAShop, the first framework for multi-concept image editing with LoRA models. LoRAShop builds on a key observation about the feature interaction patterns inside Flux-style diffusion transformers: concept-specific transformer features activate spatially coherent regions early in the denoising process. We harness this observation to derive a disentangled latent mask for each concept in a prior forward pass and blend the corresponding LoRA weights only within regions bounding the concepts to be personalized. The resulting edits seamlessly integrate multiple subjects or styles into the original scene while preserving global context, lighting, and fine details. Our experiments demonstrate that LoRAShop delivers better identity preservation compared to baselines. By eliminating retraining and external constraints, LoRAShop turns personalized diffusion models into a practical `photoshop-with-LoRAs' tool and opens new avenues for compositional visual storytelling and rapid creative iteration.
ICCV 2025

[Spotlight] CLoRA: A Contrastive Approach to Compose Multiple LoRA Models

Tuna Han Salih Meral*, Enis Simsar*, Federico Tombari, and Pinar Yanardag

International Conference on Computer Vision (ICCV) 2025

Abstract PDF Website

Low-Rank Adaptations (LoRAs) have emerged as a powerful and popular technique in the field of image generation, offering a highly effective way to adapt and refine pre-trained deep learning models for specific tasks without the need for comprehensive retraining. By employing pre-trained LoRA models, such as those representing a specific cat and a particular dog, the objective is to generate an image that faithfully embodies both animals as defined by the LoRAs. However, the task of seamlessly blending multiple concept LoRAs to capture a variety of concepts in one image proves to be a significant challenge. Common approaches often fall short, primarily because the attention mechanisms within different LoRA models overlap, leading to scenarios where one concept may be completely ignored (e.g., omitting the dog) or where concepts are incorrectly combined (e.g., producing an image of two cats instead of one cat and one dog). To overcome these issues, CLoRA addresses them by updating the attention maps of multiple LoRA models and leveraging them to create semantic masks that facilitate the fusion of latent representations. Our method enables the creation of composite images that truly reflect the characteristics of each LoRA, successfully merging multiple concepts or styles. Our comprehensive evaluations, both qualitative and quantitative, demonstrate that our approach outperforms existing methodologies, marking a significant advancement in the field of image generation with LoRAs. Furthermore, we share our source code, benchmark dataset, and trained LoRA models to promote further research on this topic.
CVPR 2025

FluxSpace: Disentangled Semantic Editing in Rectified Flow Transformers

Yusuf Dalva, Kavana Venkatesh, and Pinar Yanardag

Conference on Computer Vision and Pattern Recognition

Abstract PDF Website

Rectified flow models have emerged as a dominant approach in image generation, showcasing impressive capabilities in high-quality image synthesis. However, despite their effectiveness in visual generation, rectified flow models often struggle with disentangled editing of images. This limitation prevents the ability to perform precise, attribute-specific modifications without affecting unrelated aspects of the image. In this paper, we introduce FluxSpace, a domain-agnostic image editing method leveraging a representation space with the ability to control the semantics of images generated by rectified flow transformers, such as Flux. By leveraging the representations learned by the transformer blocks within the rectified flow models, we propose a set of semantically interpretable representations that enable a wide range of image editing tasks, from fine-grained image editing to artistic creation. This work offers a scalable and effective image editing approach, along with its disentanglement capabilities.
CVPR 2024

[ORAL] NoiseCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directions in Diffusion Models

Yusuf Dalva, and Pinar Yanardag

Conference on Computer Vision and Pattern Recognition

Abstract PDF Website

Generative models have been very popular in the recent years for their image generation capabilities. GAN-based models are highly regarded for their disentangled latent space, which is a key feature contributing to their success in controlled image editing. On the other hand, diffusion models have emerged as powerful tools for generating high-quality images. However, the latent space of diffusion models is not as thoroughly explored or understood. Existing methods that aim to explore the latent space of diffusion models usually relies on text prompts to pinpoint specific semantics. However, this approach may be restrictive in areas such as art, fashion, or specialized fields like medicine, where suitable text prompts might not be available or easy to conceive thus limiting the scope of existing work. In this paper, we propose an unsupervised method to discover latent semantics in text-to-image diffusion models without relying on text prompts. Our method takes a small set of unlabeled images from specific domains, such as faces or cats, and a pre-trained diffusion model, and discovers diverse semantics in unsupervised fashion using a contrastive learning objective. Moreover, the learned directions can be applied simultaneously, either within the same domain (such as various types of facial edits) or across different domains (such as applying cat and face edits within the same image) without interfering with each other. Our extensive experiments show that our method achieves highly disentangled edits, outperforming existing approaches in both diffusion-based and GAN-based latent space editing methods.
CVPR 2024

[Highlight] RAVE: Randomized Noise Shuffling for Fast and Consistent Video Editing with Diffusion Models

Ozgur Kara*, Bariscan Kurtkaya*, Hidir Yesiltepe, James M. Rehg, and Pinar Yanardag

Conference on Computer Vision and Pattern Recognition

Abstract PDF Website

Recent advancements in diffusion-based models have demonstrated significant success in generating images from text. However, video editing models have not yet reached the same level of visual quality and user control. To address this, we introduce RAVE, a zero-shot video editing method that leverages pre-trained text-to-image diffusion models without additional training. RAVE takes an input video and a text prompt to produce high-quality videos while preserving the original motion and semantic structure. It employs a novel noise shuffling strategy, leveraging spatio-temporal interactions between frames, to produce temporally consistent videos faster than existing methods. It is also efficient in terms of memory requirements, allowing it to handle longer videos. RAVE is capable of a wide range of edits, from local attribute modifications to shape transformations. In order to demonstrate the versatility of RAVE, we create a comprehensive video evaluation dataset ranging from object-focused scenes to complex human activities like dancing and typing, and dynamic scenes featuring swimming fish and boats. Our qualitative and quantitative experiments highlight the effectiveness of RAVE in diverse video editing scenarios compared to existing methods.