T2I generation using stable diffusion
Abstract
This research outlines the design and modification of a deep learning-based text-to-image (T2I) generator that utilizes the available Stable Diffusion capabilities. The model was constructed in PyTorch and the Diffusers library, the model packaging that utilizes Gradio to simplify interface accessibility. Technical improvements to the models include CUDA acceleration, prompt-guided dynamic-re-renders, and memory efficiencies that allow for more rapid generation, and enhanced images. All performance tests on the models show that they are functional at transferring descriptive information into images with detail. While models produced better performance characteristics than original implementations with xFormers and customized mid-tier GPU optimization, challenges remain which include vague science prompts, limited computational resources to produce images, and ethical considerations related to bias, detection of AI-generated materials, and continued reliance on proprietary, under-represented datasets. In ongoing work, we expect to build on our accomplishments and increase the representation of existing datasets to increase knowledge equity, and enhance efficiencies in computational performance.
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.