(flyx3) (2026)

Image description: A young woman standing with her arms outstretched to both sides while looking to the right, surrounded by floating water droplets and multi-colored light flares against a cloudy sky above an urban cityscape. She has black hair tied into twin low pigtails, blue eyes, a slight smile, and a thin black choker with a blue jewel pendant. Her outfit consists of a sleeveless white zip-up hoodie over a pink top, denim shorts with rolled cuffs, and a dark red backpack worn on her back.

Full Generation Parameters:

Shinkai Makoto Style, Weathering With You Style, Amano Hina, 1girl, shorts, solo, blue eyes, outdoors, black hair, twintails, hood, sleeveless, choker, sky, cloud, hood down, short shorts, day, low twintails, bag, sleeveless hoodie, blue sky, sleeveless jacket, looking at viewer, cowboy shot, cloudy sky, water drop, smile, hooded jacket, long hair, low-cut armhole, bare arms, black choker, closed mouth, jacket, building, hoodie, breasts, blue shorts, bra, small breasts, denim shorts, white jacket, denim, bare shoulders, water, 这张图片展示了新海诚动画电影《天气之子》中天野阳菜(Amano Hina)充满活力与动感的动漫风格插画, 画面采用极具张力的低角度仰拍视角, 捕捉了她站在户外, 双臂舒展, 面对壮丽天空的瞬间, 阳菜呈半身像(大腿以上), 身体微微后仰, 胸膛敞开迎向空气, 她的右臂向下伸向画面前景下方, 手指大大张开, 左臂则向侧面伸展, 姿态自由而充满欢迎之意, 她肤色白皙透亮, 脸颊泛着淡淡红晕, 鼻梁小巧, 唇边挂着一抹既坚定又温柔的微笑, 她那双明亮的大眼睛呈青蓝色, 透着自信, 目光投向右上方, 深色睫毛轮廓分明, 眉毛细长而专注, 她那深蓝黑色的头发在颈后扎成两个低马尾, 发尾随风飘扬, 呈现出动感的弧度, 而侧刘海和几缕垂落的发丝则修饰着她的脸庞与鬓角, 颈间戴着标志性的黑色颈圈, 中央垂挂着一颗小巧的水滴状蓝色宝石吊坠, 她身穿一件无袖白色拉链连帽外套, 袖口和下摆处隐约露出粉色内衬, 白色兜帽垂在肩后, 前襟拉链清晰可见, 背上背着一个小巧的棕色背包, 深色肩带横跨双肩, 下身穿着一条修身的高腰深蓝色牛仔短裤, 裤脚卷起, 紧贴着她的大腿线条, 周围的空气中弥漫着漂浮的虹彩水滴和发光的粒子, 在她的身周盘旋飞舞, 背景是一片宏大如电影画面的天空:左侧厚重的深色风暴云层向两边散开, 露出右侧阳光灿烂, 呈绿松石蓝色的晴空, 远处的低处隐约可见模糊的城市建筑, 屋顶脚手架以及一座红色的鸟居, 营造出一种光芒穿透雨幕的强烈视觉冲击力, <lora:Shinkai_Makoto_Krea2:1>,

Steps: 8, CFG scale: 1, Sampler: Euler, Seed: 693567466, RNG: CPU, Model: krea2TurboFp8, width: 960, height: 1440, Version: neo-2.26, Module 1: qwen3vl_4b_fp8_scaled, Module 2: qwen_image_vae, Model hash: eb4dd8c612, Hires steps: 1, Hires upscale: 1.6, Schedule type: Simple, Hires Module 1: Use same choices, Hires upscaler: 4x-UltraSharp, Hires CFG Scale: 1, Denoising strength: 0.05

A young woman standing with her arms outstretched to both sides while looking to the right, surrounded by floating water droplets and multi-colored light flares against a cloudy sky above an urban cityscape. She has black hair tied into twin low pigtails, blue eyes, a slight smile, and a thin black choker with a blue jewel pendant. Her outfit consists of a sleeveless white zip-up hoodie over a pink top, denim shorts with rolled cuffs, and a dark red backpack worn on her back.
you are viewing a single comment's thread
view the rest of the comments
[–] 1 point 1 month ago (1 child)

Watched 天気の子 three or four times and never knew her family name. Always just knew her as Hina.

Not a big AI guy, but it's interesting how you go from English to Japanese. My understanding of LLMs is that they natively understand all languages (or at the very least can parse them), so if you were to think a command/parameter would be clearer in another language... my understanding is that if you just go ahead and give it, the LLM will know what to do with it.

For clarification to others: you say "Weathering With You" and I think "the international name for 天気の子". If you'd just said 天気の子, I'd just think "天気の子". Both get me to the same place. I'm not saying "Weathering With You" isn't a valid name for the film (it literally in fact, is), but, it says something different. LLMs can pick up that nuance. Which is what I assume happened and why over half the parameters are in Japanese.

It's interesting to me. I've thanked a chatbot in Japanese (the kanji for "Arigatou Gozamashita") and it just completely switched over to Japanese. That's fine, but I'd have been a lot more interested if it realised I'm not a native Japanese speaker, and encouraged me to learn more of the language (or offered to help, something I know is in its skill set). I also asked it to make a cohesive playlist of songs in English, Thai, and Japanese (not at all related to 天気の子), and because it could understand all the lyrics without needing to parse a translation, it was able to form a narrative from songs in languages I could not understand. I will have to take it at its word that there is a semblance of sense there, but hey, it sounds cool to me. Just like... there could be mistakes in this generated image. I'm not seeing any. We're well past the "six fingers" stuff. I was looking at windows, the scaffolding, until I remembered in the film, no buildings were really higher than that torii gate. Maybe they were. I don't recall us getting that angle. I wouldn't be surprised if it was right and my memory was faulty. Point is, the mistakes are getting less and less obvious.

  • source
  • hideshow 1 child comment