An Interactive Drawing Assistant Harmonizing User Control and Creative Freedom in Image Generation
Abstract
Recent advances in text-to-image models have brought about impressive visual quality and improved language understanding. However, a persistent trade-off remains: while control techniques such as ControlNet help better align with user intent, they usually constrain the model’s creative freedom and place a greater burden on the user for creating control images. To bridge this gap, we introduce an Interactive Drawing Assistant that aims at harmonizing user control and creative freedom in image generation. Our assistant follows a three-stage, human-in-the-loop workflow: 1) Intent-to-Layout – A large language model interprets the user’s prompt and suggests editable composition layouts, giving users high-level control with minimal effort. 2) Co-Sketching – A collaborative canvas allows users to draw sketches themselves or automatically generate sketches using our sketch generation module. 3) Interactive sketch-to-image – A diffusion-based module converts sketches into high-resolution images while allowing iterative, natural-language edits to fine-tune style or content without restarting the pipeline. Comprehensive user studies and baseline comparisons show that our system is widely acknowledged for striking an effective balance between controllability and creativity. The consistently positive feedback also highlights the promise of tight, step-by-step human-AI cooperation for creative yet controlled image creation.
Read the paper
Related studies
- Learning from First-Generation Qualitative Approaches in the IS Discipline: An Evolutionary View and Some Implications for Authors and Evaluators (PART 1/2)
- Commentary on the clinical and preclinical dosage limits of interstitially administered magnetic fluids for therapeutic hyperthermia based on current practice and efficacy models
- Prosthesis: New Technological Opportunities and Innovative Biomedical Devices
- The history and future of neural modeling for cochlear implants
- Ryanodine receptors: waking up from refractoriness
- Novel Transgenic Rabbit Model Sheds Light on the Puzzling Role of Matrix Metalloproteinase-12 in Atherosclerosis
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.