RIZ-ANIMA (AURA-v1)

This version of RIZ-ANIMA has been trained on the Anima architecture using a custom-curated dataset to enhance overall quality and consistency. The model features improved prompt understanding, allowing for more accurate interpretation of user inputs and better alignment with intended outputs.

Special focus has been given to refining artist style reproduction. Based on internal testing, several styles now perform closely to expected results, offering more reliable and visually appealing generations. However, not all styles have been extensively tested yet.

In addition, multiple issues present in the base Anima model have been addressed and optimized, resulting in smoother outputs and reduced inconsistencies.

Note: This version has primarily been tested using simple test cases and not through full-scale evaluation. If you encounter any issues or unexpected behavior, your feedback is highly appreciated.

The training process was accelerated using a larger AI computing cluster, enabling faster development while maintaining quality improvements.