After working with the Illustrious model for about two weeks, I've found it excels in certain areas where Pony Diffusion V6 falls short. Illustrious offers impressive style adaptability and concept versatility, while Pony V6 shines in creating images with consistent anatomy, especially in complex areas like limbs and hands.
The Strengths of IllustriousIn my last review, I shared some initial thoughts on Illustrious, but diving deeper has revealed even more strengths. One standout feature is its ability to interpret and replicate intricate style details. Illustrious "understands" how styles work, capturing subtle elements that make each style unique, which is a big plus for anyone looking to build distinct style models.
Additionally, Illustrious handles negative prompts exceptionally well. Unlike Pony, where negative prompts can heavily distort an image, Illustrious allows you to adjust strength without unwanted deformations. This flexibility extends to both positive and negative tags, making it easier to guide the model's output without losing control.
Background generation is another area where Illustrious outperforms Pony. While Pony tends to default to indoor settings (likely due to its training data), Illustrious adapts based on its learned styles. For example, if your training images feature dark backgrounds, Illustrious will often produce a similar setting unless prompted otherwise.
Concept generation also stands out. Illustrious allows for a wider range of dynamic poses—jumping, floating, splits—without losing image cohesion. Even classic concepts from Pony, like a person leaning against a pole, come through more accurately in Illustrious. Instead of disconnected elements, as sometimes happens with Pony, Illustrious usually keeps lines clean and continuous.
Lastly, Illustrious supports more tags without relying on score adjustments. This freedom enables more precise customization of generated images and makes it easier to experiment with various traits.
Where Pony V6 Holds Its GroundDespite Illustrious’s strengths, there are areas where Pony V6 still has an edge. Pony’s accuracy in generating anatomically correct figures—whether human, anime, furry, or even statues—remains impressive. If you prompt it with “1girl, solo, blonde hair,” you’ll get exactly that with high consistency. Illustrious, while accurate, occasionally introduces oddities, like mismatched limb sizes or missing hands. These issues are rare, but they do appear from time to time.
Pony also performs better with certain character types, particularly Western styles and furry characters, making it more versatile in these areas. While these styles may be niche for some, it’s convenient to generate them without needing an additional model.
Final ThoughtsOverall, Illustrious has impressed me with its adaptability and creativity, and I find it currently better suited for style-focused projects than Pony V6. However, I’ll continue testing both and may revisit this comparison as I gather more insights. If you’ve noticed other strengths or quirks with Illustrious, feel free to share—I’d love to learn what else this model can do!