分享几张我用 nano banana 2 生成的运动图像。我一直想让自己的作品更真实，更像是比赛中真正被捕捉到的瞬间，到目前为止，一些…

sharing a few sports images I generated with nano banana 2. I’ve been trying to make mine feel more realistic, more like an actual moment being captured during the game, and a few prompt details have been helping a lot so far: - prompt the exact split second, not just the sport. things like “fully extended mid-air, fingertips reaching toward the ball” or “one leg pushing off the floor, the other extended in stride” gave me much better results than just saying goalkeeper making a save or basketball player in action. - be specific about what needs to stay readable and what can fall into blur. lines like “the goalkeeper remains the hero with face and hands sharply readable” or “keeping the player’s face, torso, and ball sharply readable” made a huge difference in these images. - small physical details change everything. things like “jersey stretch,” “sweat,” “fabric tension,” “stadium mist” and “subtle airborne particles catching light” made the image feel way less generic. - camera language is essential. “dynamic low-angle action framing,” “telephoto sports lens feel around 200mm,” and “fast shutter look with selective motion blur” gave the images a much clearer direction. - negatives matter more than I expected. “no fake anatomy, no clutter,” “no stock sports poster look,” and “no superhero exaggeration” were genuinely useful. still exploring sports AI imagery, but these are the details getting me closer to the result I want for now. full prompts are in the ALT, in case anyone wants the exact wording.

分享几张我用 nano banana 2 生成的运动图像。我一直想让自己的作品更真实，更像是比赛中真正被捕捉到的瞬间，到目前为止，一些提示词的细节帮了大忙： - 提示词要写出确切的瞬间，而不仅仅是运动本身。像“fully extended mid-air, fingertips reaching toward the ball”（完全伸展腾空，指尖触向球）或“one leg pushing off the floor, the other extended in stride”（一条腿蹬地起跳，另一条腿大步伸展）这样的描述，比只写守门员扑救或篮球运动员在行动，效果要好得多。 - 要明确指出哪些部分必须保持清晰可读，哪些可以进入模糊。像“the goalkeeper remains the hero with face and hands sharply readable”（守门员依然是主角，脸部和双手保持锐利清晰）或“keeping the player’s face, torso, and ball sharply readable”（保持球员的脸部、躯干和球都清晰可读）这样的句子，对这些图像有巨大的影响。 - 微小的物理细节会改变一切。像“jersey stretch”（球衣拉扯）、“sweat”（汗水）、“fabric tension”（面料张力）、“stadium mist”（体育场雾气）以及“subtle airborne particles catching light”（细微的空气颗粒捕捉光线），都让图像少了很多千篇一律的感觉。 - 镜头语言至关重要。“dynamic low-angle action framing”（动感低角度动作构图）、“telephoto sports lens feel around 200mm”（约 200mm 长焦运动镜头感）以及“fast shutter look with selective motion blur”（高速快门效果搭配选择性动态模糊），让图像有了更明确的方向。 - 负面提示比想象中更重要。“no fake anatomy, no clutter”（不要错误的解剖结构，不要杂乱）、“no stock sports poster look”（不要图库式运动海报的感觉）以及“no superhero exaggeration”（不要超级英雄式的夸张），真的非常有用。 我仍在探索 AI 运动图像的创作，但这些细节正让我越来越接近自己想要的效果。 完整的提示词写在 ALT 里，方便大家查阅原文。