Today, we introduce HunyuanImage 3.0-Instruct, a native multimodal model focusing on image-editing by integrating visual understanding with precise image synthesis! 🚀 It understands input images and reasons before generating images. Built on an 80B-parameter MoE architecture (13B activated), it natively unifies deep multimodal comprehension and high-fidelity generation. 🧠 A "Thinking" Model with Native CoT & MixGRPO: The model doesn’t just execute commands, it processes them through a Native Chain-of-Thought (CoT) schema. Enhanced by our self-developed MixGRPO algorithm, it reasons through complex instructions to achieve flawless intent alignment and human-preference consistency. 🎨 Precise Editing & Multi-Image Fusion: The model enables accurate image editing by adding, removing, or modifying elements while keeping non-target areas perfectly intact. It also excels at seamless multi-image fusion, synthesizing complex scenes by extracting and blending elements from multiple sources into a unified, consistent output. 🏆 SOTA Performance: HunyuanImage 3.0-Instruct sets a new benchmark in visual quality and alignment, delivering performance that matches leading proprietary models. We aim to enable the community to explore new ideas with a state-of-the-art foundation model, fostering a dynamic and vibrant image generation ecosystem. 🛠️🎨 💻Try it at (PC only): hunyuan.tencent.com/chat/Hun…

A winter night straight out of a fairytale ❄️✨ Fairytale or reality? Tell me 🤍 • Gemini Nano Banana 🍌 - How it works? Input Photo + Prompt Note: This prompt is flexible all attributes (hair, eyes, makeup, outfit, etc.) are fully customizable without affecting the result. Prompt: { "shot_type": "A vertical 4:5 flash photography shot", "subject_action_location": "of a young woman kneeling on the snowy ground at night.", "clothing": "She is wearing a tight black ribbed long-sleeve mini dress with a tiny pink satin ribbon bow on the bottom hem, sheer black pantyhose,", "accessories": "and accessories including a large fluffy black faux fur hat and matching oversized black fur wrist cuffs.", "hair": "She has long wavy copper-strawberry blonde hair", "makeup": "and heavy pink 'cold girl' makeup.", "background_location": "The background is the majestic GUM department store facade in Moscow,", "background_lighting": "completely illuminated with thousands of warm yellow Christmas string lights outlining the arches and towers.", "background_elements": "There are decorated Christmas trees and a blurred crowd of people in the background.", "weather_and_flash": "Snowflakes are falling and sparkling in the harsh direct camera flash.", "aesthetic_and_vibe": "2000s Y2K aesthetic, candid vibe, high contrast, festive winter atmosphere.", "parameter": "--ar 4:5" }

すごいインプ、、、 みんなもやりたかったんですね！ プロンプト置いておきます。お好きなキャラクターでどうぞ(少し実写になりがちかも) ---------------------- Based on the uploaded reference character, generate a live-action scene inside a wide Tokyo girl’s apartment — a bright, lived-in one-room that feels almost 1LDK in scale. White walls, warm wooden floor, beige curtains, a low bed with soft bedding, a desk with cosmetics, bookshelves, plants, a standing mirror, a rug, scattered personal items, and a compact kitchen area in the back. The room must have strong depth with clear foreground, mid-ground, and deep background layers. Place around thirty identical versions of the same character (same face, hairstyle, outfit as the reference) throughout the room, each in a different action or interaction. Vary distance, scale, height, and visibility so the density feels natural. Foreground (very close to camera / partial occlusion): - characters walking past the lens, slightly out of focus - a hand or shoulder entering frame - one leaning close toward the camera - one half-visible behind a large plant - one sitting directly in front, tying her hair - one kneeling by the desk adjusting objects Mid-ground (main room area): - one stretching beside the bed - one sitting on the bed checking her phone - one lying belly-down across the bed - one reaching under the bed - one organizing cosmetics on the desk - one flipping through a book on the shelf - one standing in front of the mirror - one crouching on the rug - one leaning against the wall - one looking out the window - one adjusting the curtain - one carrying laundry - one drinking from a cup - one tidying pillows - one sitting on the floor eating snacks - one doing a small jump or motion blur gesture - one moving a small chair Background (deep perspective / near kitchen and hallway): - one standing near the stove drinking water - one opening a cabinet - one sitting on a stool - one leaning in a doorway - one walking toward the hall - one silhouette partly hidden behind the fridge - one reaching up to a high shelf - one standing far by the entrance area - one barely visible through the hallway frame - one sitting on the floor near the kitchen rug Ensure strong layered occlusion: foreground characters partially block mid-ground ones, and background characters appear smaller with natural perspective falloff. Scatter the thirty characters organically, avoiding symmetry or grid alignment. Lighting is soft natural daylight from the window, consistent across all characters for full integration. Placed within a live-action background that matches the illustration’s posture and composition — while faithfully preserving the illustrated texture and style. Realistic lighting, depth of field, and subtle filming effects are applied to blend the illustration seamlessly with the real environment.

一句话用gemini3做了个贪吃蛇小游戏，一起来玩玩吧