prompts.

├── MainActivity.kt # Main activity and UI composition

2. Create a feature branch

Generate an image of a banana wearing a costume.

Okay, here is a banana wearing a costume for you:

**说明**: 用自然语言直接“指挥”模型修改图片。不再需要复杂的P图软件，只需要告诉它“给这个女孩戴上一副墨镜”或者“把天空换成梵高星空的风格”，模型就能理解并执行你的指令。

**说明**: 在广告制作初期，快速生成大量不同风格、不同创意的“起始帧”或“概念图”。这能帮助创意团队在几分钟内将想法可视化，进行提案和筛选，极大地提升了效率。

5. 风格迁移 (Move styles from one image to another)

Refer to the video above, open the cloned folder with Antigravity, and use the prompt below to generate images.

**Transparent pixels** show the background color → distance = max → alpha = 0

When using AI image generation tools, follow this workflow to create transparent PNGs:

Generate your subject on a **neutral solid color background** (not white or black) that contrasts with the subject.

2. Create a new API Key

User prompt: "hkust entrance piazza in cyberpunk future"

Using the uploaded floor plan as the base image, arrange furniture and soft furnishings in each space according to the functional room labels indicated by the text annotations in the original drawing. Maintain all wall structures, door locations, window positions, and architectural elements exactly as shown without any modifications to the building configuration. Do not add any new walls or partitions to the layout. Preserve the black and white line drawing style and monochromatic palette throughout the composition. Remove all text annotations and labels from the final output, showing only the architectural elements and newly added furniture arrangements.

Ultra-realistic 3D render of a cute, miniature [Uploaded images] building.

Whether you’re looking for photorealistic portraits, stylized aesthetics, or complex creative experiments, you will find a broad set of effective examples here to unlock the expressive potential of generative models.

**Photorealistic Portraits

5. Adjust resolution, composition, and style modifiers as needed.

Subject → Style anchor → Lighting → Camera hint → Detail modifiers → Composition context

✔ When refining existing images, use image-to-image suffixes that emphasize texture clarity and lighting correction.

"create a streetwear t-shirt design with skulls and flames for my Etsy shop" "generate an instagram post image for my artisanal coffee brand" "make a cute kawaii cat design for coffee mugs" "I need a minimalist poster design with just the word BREATHE"

generate an instagram post image for my artisanal coffee brand

make a cute kawaii cat design for coffee mugs

I need a minimalist poster design with just the word BREATHE