| There are a lot of good image models you can use. Flux is the best open weights model. Ideogram, Recraft, Midjourney, Leonardo are all very capable hosted image generators. DALL-E3 was way ahead of its time and is still very good. RunwayML Gen3 Alpha, Lumina, Hailuo, Kling, Minimax and others do video well. Sora is probably the best visual media generator but is not widely available to use. Only people at Meta have used Meta’s Chameleon, which is maybe the most capable visual media generator today. None are particularly good at particular styles or not. All the content on CivitAI is reflective of the quality of the foundational models. Flux and SD3 community fine tunes are very capable. CivitAI isn’t representative of the best in the community, the state of the art, or even what people are using this stuff for. |