<h1 id="krea2-storyteller-lora">Krea2 Storyteller LoRA</h1>A Krea2 LoRA trained on 56 images from my personal image-generation dataset.This LoRA is designed to bring my preferred visual storytelling and image-making tendencies into new generations.Rather than simply applying a fixed art style, it focuses on the things that shape the overall feel of an image:<ul><li>lighting and light-shadow balance</li><li>atmosphere, depth, and mood</li><li>material and surface detail</li><li>soft highlights, reflections, and bloom</li><li>cinematic and nostalgic atmosphere</li><li>the relationship between characters and their environment</li></ul>The goal is to add a more carefully composed, atmospheric feel while leaving the original prompt in control of the subject and composition.This LoRA is not intended to reproduce the subjects or compositions of the training images. It is intended to transfer the recurring visual tendencies found across them.<h3 id="recommended-strength">Recommended Strength</h3>0.2–0.5 is a good starting range, with 0.3 as a default.The effect can already be noticeable around 0.2–0.3.At strengths of 0.7 and above, the LoRA becomes more dominant. Scenes may become darker, hair colors may shift lighter, and objects or backgrounds may take on a more realistic material appearance.Higher strengths can also introduce unintended details learned from the training images, such as accessories, clothing details, glossier lips, or heavier makeup.<h3 id="generation-settings">Generation Settings</h3>Samplers used: <code>euler</code>, <code>euler_a</code>, <code>er_sde</code> Scheduler: <code>Simple</code><h3 id="training">Training</h3><ul><li>Images: 56</li><li>Network Rank: 32</li><li>Training Steps: 1,000</li></ul>

Files with SHA256: ca1afe1b6bc322c020de6d9358025a7215220238993ab792ba17f7ccd12f16c1