Artificial Analysis 的模型能力拆解显示,MAI-Image-2.6-Preview 在 Materials 和 Knowledge 两项表现最佳。
In our model capability breakdown, MAI-Image-2.6-Preview does best in Materials and Knowledge.
➤ Lighting tests how light behaves: direction, falloff, catchlights, and shadow that matches the source.
➤ Knowledge tests whether real-world specifics come out right.
Example Text to Image Prompt: Virtually staged listing photo of a loft living room in a converted textile mill. The shell stays as found: sandblasted brick walls, original maple floors with patched board sections and old machine-oil staining, a cast iron column wearing layers of chipped paint. The staged furniture should sit naturally in the room's light: a cognac leather recliner with a seat cushion creased and softened from years of use, the grain lightening across the arms, a loveseat in looped cream boucle wool, a slubby linen throw, and a flat-woven kilim in faded madder red under the group. Soft contact shadows grounding every piece on the maple. Steel-framed warehouse windows, overcast daylight, no people, no watermark.
来源:@ArtificialAnlys · x.com