<h1 id="furrydalle">FurryDalle</h1><h2 id="descripcion-description">Descripción / Description</h2><h3 id="espanol">Español</h3><p>La evolución espiritual de FurryCore, ahora expandida y adaptada para múltiples ecosistemas base, incluyendo Illustrious, Anima[base], YiffyMix SD1.5 y YiffyMix SDXL.</p><p>FurryDalle es un LoRA experimental centrado principalmente en el estilo visual más que en la generación específica de personajes anthro. Su objetivo es recrear una estética inspirada en ilustración tradicional, renderizado pictórico y acabados estilizados, manteniendo al mismo tiempo una buena flexibilidad con prompts y compatibilidad entre distintos modelos base.</p><p>Esta nueva iteración fue entrenada utilizando un conjunto cuidadosamente curado de imágenes previamente generadas con DALL·E 3, empleadas como referencia estilística para explorar cómo datasets sintéticos pueden transformarse en una identidad visual consistente, adaptable y multiplataforma.</p><p>El proyecto continúa en desarrollo activo, por lo que el feedback sobre consistencia, anatomía, composición, comportamiento de prompts y calidad visual es muy valioso para futuras mejoras.</p><hr /><h3 id="english">English</h3><p>The spiritual evolution of FurryCore, now expanded and adapted for multiple base ecosystems including Illustrious, Anima[base], YiffyMix SD1.5, and YiffyMix SDXL.</p><p>FurryDalle is an experimental LoRA primarily focused on visual style rather than strict anthro character generation. Its goal is to emulate a traditional illustration aesthetic with painterly rendering and stylized finishes while maintaining strong prompt flexibility and cross-model compatibility.</p><p>This new iteration was trained using a carefully curated dataset composed primarily of images previously generated with DALL·E 3, used as stylistic references to explore how synthetic datasets can evolve into a cohesive, adaptable, and cross-model visual identity.</p><p>The project remains under active development, so feedback regarding consistency, anatomy, composition, prompt behavior, and visual quality is highly appreciated for future improvements.</p><hr /><h1 id="modelos-base-base-models">Modelos Base / Base Models</h1><ul><li><p>YiffyMix v3.5 — SD1.5</p></li><li><p>YiffyMix SDXL v51</p></li><li><p>Illustrious</p></li><li><p>Anima[base]</p></li></ul><hr /><h1 id="entrenamiento-training">Entrenamiento / Training</h1><h3 id="espanol">Español</h3><p>Entrenado utilizando un dataset curado de imágenes generadas previamente con DALL·E 3, con énfasis en ilustración estilizada, estética tradicional y consistencia visual entre diferentes modelos base.</p><h3 id="english">English</h3><p>Trained using a curated dataset of previously generated DALL·E 3 images, focusing on stylized illustration, traditional-inspired aesthetics, and visual consistency across different base models.</p><hr /><h1 id="licencia-y-creditos-license-and-credits">Licencia y Créditos / License &amp; Credits</h1><h3 id="espanol">Español</h3><p>Este proyecto puede incluir entrenamiento, referencias estilísticas o compatibilidad derivada de modelos base externos como Anima[base].</p><p>Las imágenes generadas pueden utilizarse comercialmente; sin embargo, algunos modelos base y derivados mantienen restricciones sobre redistribución comercial, checkpoints derivados, LoRAs derivados o servicios basados en dichos modelos.</p><p>Consulta la licencia oficial de Anima aquí:<br /><a target="_blank" rel="ugc" href="https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md">https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md</a></p><p>Puntos relevantes:</p><ul><li><p>Uso comercial de outputs/imágenes:<br /><a target="_blank" rel="ugc" href="https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#outputs">https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#outputs</a></p></li><li><p>Restricciones sobre modelos derivados y redistribución:<br /><a target="_blank" rel="ugc" href="https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#restrictions">https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#restrictions</a></p></li></ul><h3 id="english">English</h3><p>This project may include training influence, stylistic references, or compatibility derived from external base models such as Anima[base].</p><p>Generated images may be used commercially; however, some base models and derivatives maintain restrictions regarding commercial redistribution, derivative checkpoints, derivative LoRAs, or hosted services based on those models.</p><p>Official Anima license:<br /><a target="_blank" rel="ugc" href="https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md">https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md</a></p><p>Relevant sections:</p><ul><li><p>Commercial usage of outputs/images:<br /><a target="_blank" rel="ugc" href="https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#outputs">https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#outputs</a></p></li><li><p>Restrictions regarding derivative models and redistribution:<br /><a target="_blank" rel="ugc" href="https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#restrictions">https://huggingface.co/circlestone-labs/Anima/blob/main/LICENSE.md#restrictions</a></p></li></ul><hr /><h1 id="tags-activadores-trigger-tags">Tags Activadores / Trigger Tags</h1><p>FurryDalle</p><hr /><h1 id="redes-sociales-social-networks">Redes Sociales / Social Networks</h1><p>Encuéntrame aquí / Find me here:<br />Linktree</p>

FurryDalle

<p>This is a new experimental version for the Anima base model.<br />Rather than being a LoRA specifically focused on generating anthro characters, this LoRA is mainly intended as a style-focused addition for this version of the model, aiming to emulate a more traditional illustrated aesthetic while preserving the flexibility of the base model.</p><p>The training dataset used for this experiment was primarily composed of images previously generated with OpenAI’s DALL·E 3, which were then curated and adapted as stylistic references during training. The goal was to explore how well these generated references could translate into a cohesive visual identity inside the LoRA.</p><p>Since this is still an experimental release, feedback is highly appreciated. Any comments regarding consistency, anatomy, rendering quality, prompt behavior, strengths, weaknesses, or unexpected results will help improve future iterations and refinements of the model.</p>

Files with SHA256: 538a3bf8f7810d5eb14cdf93bdc5e11a3b53c61aa7304862099ee8251fc5a0e3