How to Use AI to Create Stunning Qixi Festival Designs in Minutes

Since I started using AI tools for design tasks, my first reaction to any new brief is to ask whether AI can help, what tasks it can assist with, and how much time it can save. This article uses a Qixi (Chinese Valentine’s Day) theme to demonstrate how AI can rapidly produce diverse design assets and provides the corresponding prompts.

The tool used is “Instant Inspiration”, which works directly in the browser without the need for a VPN or downloads, supports Chinese input, and offers free high‑resolution downloads, making it beginner‑friendly. It also features the new SDXL 1.0 model and many other CG and illustration models.

Before designing, browse design sites to decide on the desired Qixi style and choose the appropriate model combination. The right model combo reduces keyword tweaking and yields higher‑quality, more stable results. The same set of keywords can produce surprising variations when paired with different models.

1. Traditional “Ancient‑Style” Theme

For a traditional festival look, focus on keywords like “ancient style” and “ink wash”.

Model selection:

Prompt (Chinese): 插画海报设计，平面设计，美丽的中国城市，祥云，中国服装夫妇坐在喜鹊上，夸张的波浪，喜鹊，桥

Prompt (English): Illustration poster design, graphic design, beautiful Chinese city, auspicious clouds, Chinese costume couple sitting on a magpie, exaggerated wave, magpie, bridge

The pure ancient‑style output looks great, but many brands may need a more neutral look. In that case, combine the ancient model with a different style to tone down the strong cultural elements.

Alternative model combo:

Use the same description to generate a completely different style.

2. High‑Impact “3D Portrait” Style

For a vivid, three‑dimensional look, choose models that produce 2.5D street‑art style images and apply a pastel pink palette suitable for the Qixi theme.

Model selection:

Prompt (Chinese): 一个男孩和一个女孩拥抱在一起，头对头，在海里，周围有许多彩色的泡泡，镜头前的透明泡泡，卡通插画，晶莹剔透，五颜六色，生动，轻盈，梦幻，头像，8K

Prompt (English): a boy and a girl hugging each other, head to head, in the sea, surrounded by many colorful bubbles, clear bubbles in front of the camera, cartoon illustration, crystal clear, colorful, vivid, light, dreamy, head portrait, 8K

3. Creative “Paper‑Cut” Style

Another unconventional approach is to use a paper‑cut aesthetic.

Model selection: General‑purpose SD XL 1.0

Prompt (Chinese): 一对情侣在热气球里告白，镜头前有植物和花朵，简约风格，剪纸作品，剪纸设计，粉色，白色背景，3D，8k

Prompt (English): A couple confesses in a hot air balloon, with plants and flowers in front of the camera, simple style, paper cut work, paper cut design, pink, white background, 3D, 8k

4. Versatile “Hand‑Drawn Illustration”

The most common and flexible style is hand‑drawn illustration. Simply select the “Hand‑drawn Background Enhancement” base model, optionally stacking additional style models.

Model selection: Hand‑drawn Background Enhancement

Prompt (Chinese): 情人节站在花店前开心大笑的情侣，全身镜头，粉色，清新柔和的颜色，2D 插画，大师作品，最好画质，更高质量

5. Generating “Symbolic Elements”

Beyond portrait styles, you can generate love‑related symbols such as flowers, roses, hearts, or balloons.

Model selection: CG Portrait Enhancement

Prompt (Chinese): 粉色的 3D 玫瑰花，C4D 质感，清新明亮的颜色，白色背景，没有人物，白色的背景，超高品质

Prompt (English): Pink 3D roses, C4D texture, fresh and bright colors, white background, no characters, super high quality

All the above assets were generated with “Instant Inspiration”. You can copy the prompts and try them yourself.

After AI creates the raw assets, the remaining work is simple: use any familiar design tool for layout. Two main layout approaches are:

Use the generated images as backgrounds, paying attention to typography and decorative element placement, adding gradients if needed.

Extract individual elements via AI‑assisted cut‑out (e.g., the AI‑generated rose) and use them as focal points.

Both methods can be performed directly in the “Instant Design” plugin without opening Photoshop.

Finally, crop the composition, add text and decorative elements, and the poster is complete.

One extra tip: to make AI write specific text, treat the text as a graphic shape, upload it as a reference image, and enable “Edge Detection” control. The latest SDXL 1.0 can read short English words but not Chinese, so this workaround works for both languages.

Steps for text‑in‑image:

Create a blank canvas and write the desired text.

In “Instant Inspiration”, enable “Image Control”, upload the reference, and select “Edge Detection”.

Enter the full prompt.

Experiment with different fonts and control strengths for varied results.