<h1 id="eye-concept-v1-z-image-turbo">Eye Concept v1 — Z-Image Turbo</h1><h3 id="hyper-realistic-iris-detail-or-natural-eye-integration">Hyper-Realistic Iris Detail | Natural Eye Integration</h3>Eye Concept v1 is a dedicated detail LoRA for Z-Image Turbo that enhances eye realism by improving iris structure, natural moisture behavior, and skin integration around the eyes.This LoRA focuses on believable realism — not style replacement. It strengthens the eye as a physical element within the face while preserving identity, lighting, and composition.Eye Concept is designed to remain stable in complex scenes, multiple subjects, and long prompts.<hr /><h2 id="key-features">Key Features</h2>Optimized for Z-Image Turbo Built specifically for Z-Image Turbo behavior and prompt handling.Hyper-Realistic Iris Structure Enhances internal fiber detail, color variation, and natural limbal ring definition.Natural Moisture &amp; Catchlight Behavior Produces believable tear-line reflections and eye moisture without artificial shine.Non-Plastic Skin Integration Improves eyelid folds, lash roots, and micro-skin transitions to avoid the “plastic” look.Stable Under Heavy Prompts Tested with long prompts (600+ tokens), multiple subjects, movement, and outdoor scenes.Non-Style LoRA Does not impose a look — it enhances realism while respecting the base model.<hr /><h2 id="trigger-word">Trigger Word</h2>Use the trigger word in your prompt:<pre><code></code></pre><pre><code>eye_concept</code></pre>Use once per subject.Example:<pre><code></code></pre><pre><code>photorealistic portrait, natural lighting, eye_concept, realistic skin, sharp iris detail, natural catchlight</code></pre><hr /><h2 id="recommended-strength">Recommended Strength</h2>Start with:<pre><code></code></pre><pre><code>0.60 – 0.70</code></pre>Typical working value:<pre><code></code></pre><pre><code>0.65</code></pre>Increase if:<ul><li> eyes look soft </li><li> iris detail is weak </li><li> close-up or macro shot </li></ul>Decrease if:<ul><li> result looks too sharp </li><li> stacking multiple detail LoRAs </li></ul><hr /><h2 id="practical-usage-examples">Practical Usage Examples</h2><h3 id="general-photoreal">General Photoreal</h3><pre><code></code></pre><pre><code>photorealistic portrait, eye_concept, natural lighting, realistic skin, detailed iris, natural catchlight</code></pre><hr /><h3 id="multi-subject-scene">Multi-Subject Scene</h3><pre><code></code></pre><pre><code>two women walking in light rain, eye_concept, cinematic lighting, realistic skin, natural eye detail</code></pre><hr /><h3 id="close-up-beauty-shot">Close-Up / Beauty Shot</h3><pre><code></code></pre><pre><code>close-up portrait, eye_concept, shallow depth of field, ultra-detailed iris, soft lighting</code></pre><hr /><h2 id="how-this-lora-behaves">How This LoRA Behaves</h2>Eye Concept enhances:<ul><li> iris fiber detail </li><li> limbal ring clarity </li><li> eye moisture realism </li><li> eyelid folds </li><li> lash-to-skin integration </li><li> natural skin micro-variation </li></ul>It does not:<ul><li> change identity </li><li> alter body proportions </li><li> override lighting </li><li> impose a visual style </li></ul><hr /><h2 id="compatibility">Compatibility</h2>Designed for:<ul><li> Z-Image Turbo </li><li> photoreal workflows </li><li> character LoRAs </li><li> fashion / portrait rendering </li><li> cinematic scenes </li><li> multi-character prompts </li></ul>Works well with:<ul><li> character LoRAs </li><li> clothing LoRAs </li><li> realism prompts </li><li> long prompt structures </li></ul><hr /><h2 id="performance-characteristics">Performance Characteristics</h2>Eye Concept is tuned to remain:<ul><li> stable </li><li> predictable </li><li> non-destructive </li><li> visually natural </li></ul>Tested scenarios include:<ul><li> multiple subjects </li><li> walking poses </li><li> rain / wet surfaces </li><li> complex clothing </li><li> full-body compositions </li><li> high token prompts </li></ul><hr /><h2 id="important-notes">Important Notes</h2>Eye Concept is a detail LoRA, not a style LoRA.You do not need a separate realism LoRA for Z-Image Turbo.If stacking with other high-detail LoRAs:Lower one of the strengths slightly to avoid excessive sharpness.<hr /><h2 id="version">Version</h2>eye concept v1 — Z-Image TurboInitial release focused on:<ul><li> realistic iris detail </li><li> natural eye moisture </li><li> skin integration </li><li> stability under heavy prompts </li></ul>Future versions may expand into:<ul><li> lighting-adaptive reflections </li><li> age-dependent eye behavior </li><li> cinematic eye response</li></ul>

Files with SHA256: 6da990cdf0f3062442ba942ac811a1134b77b29ce385ea0e3344ef501c242fbf