It has been over half a month since Google released Nano Banana Pro, but its creative use cases are still emerging one after another.
Previously, when using AI to create graphics, the most troublesome issues of consistency and Chinese text accuracy were solved, and prompts no longer had to be written like a mini-essay.
I found a method that can be experienced for free in China—Jiaotu AI. Generating 4K ultra-HD images only takes 1-2 minutes, image creation is super fast, and exported images have no watermark.
After playing with it for a while and trying countless prompts, I put together a set of routines for making e-commerce design graphics, from key visuals, product detail pages to campaign posters, different sizes of e-commerce design needs can be handled just by typing.
01. E-commerce Design
Open the Jiaotu AI official website, and select the Banana Pro model in the input box.
Official website:https://www.jiaotuai.cn
By the way, Jiaotu AI is not limited to this one model. The latest mainstream models such as Seedream 4.5, Qwen Image, Veo3.1, and Sora 2 can all be used directly.
Gameplay 1 White Background Image
Prompt: Based on a real photo of the product, generate a professional commercial product photography image; pure white background, studio-grade soft lighting, realistic light and shadow with natural reflections, high-end brand advertising composition, ultra-clear details, clean image, no cluttered elements.
Photos taken casually can directly generate professional commercial photography effect images. After the Nano Banana Pro upgrade, the text rendering accuracy is truly impressive. All 3 types of text on this hand cream are rendered very accurately, exactly the same as the original product.
Gameplay 2 Product Photography
Prompt: High-end product photography, floating composition, dark gradient background, soft studio lighting, cool tone, premium feel, brand official website key visual, realistic photographic texture.
Prompt: Cinematic product photography, single-direction light, strong light-dark contrast, geometric light and shadow, dark background, warm-colored floor, high-end still life advertising style.
Prompt: Based on a real photo of the product, generate high-end skincare cinematic still life photography, dark silk backdrop fabric, natural folds forming soft light-dark transitions, directional light source from the upper side, a beam of light cutting into the frame, low ambient illumination, large dark areas retained, the product as the sole visual focus, realistic photographic texture.
The presentation of different light and shadow effects is very good, with basically no physical errors. The overall composition structure is also very good, allowing people to grasp the visual focus at a glance.
Gameplay 3 Scene Image
Prompt: Based on a real photo of the product, generate a scene image. Rose petal elements as the main visual language of the image, [the product placed in clear water, with a few rose petals floating on the water surface as embellishment]. The light is soft and concentrated, highlighting the product subject, suitable for e-commerce detail page display.
By changing the description in [], you can replace the background and decorations to generate various scene real-shot images. The generated rose petals are very realistic, and there is a background blur effect like that taken by a camera, making it look more like a real shot.
Prompt: Based on a real photo of the product, generate an Ins-style image.
If only the style is specified, we can get a whole set of real-shot images with a consistent style, especially suitable for promotion on social platforms.
We can also generate a gift box set for the product:
Prompt: Based on real product photos, generate a gift box scene image of the product. Shot in a real indoor desktop environment, natural light illuminates the gift box and product, the image is bright and clean, and the desktop retains natural shadows and spatial depth. The overall composition has clear primary and secondary elements and negative space; the image looks brand-designed rather than casually arranged, presenting the realistic photographic quality of a high-end skincare gift box scene.
Prompt: The overall visual style remains consistent, but with each generation the way the gift box is opened, the shooting angle, the composition, and the positions of elements change naturally, similar to different frames obtained by pressing the shutter during the same brand photography shoot.
Play 4 Model Display
Prompt: Add an Asian female model in Image 1. The model's shoulder serves as a blurred foreground, in the center of the frame, and the model holds the hand cream from Image 2 in one hand.
Prompt: Based on real product photos, generate a model image. The foreground of the image is the outline of the model's shoulder and is clearly blurred to create a sense of occlusion, [in the center of the frame the model's hand naturally holds the product], the product is clear and sharp as the only visual focus, soft indoor light, and the background presents a living room environment but is slightly blurred, suitable for e-commerce detail page usage scenario display.
Modify the prompt in [ ] to specify the model's action, so that the product can be displayed in different ways.
Play 5 Selling Point Image
Prompt: Based on real product photos, generate an e-commerce detail page efficacy key visual, add a hand model display, describe the product's functions, and help users fully understand the product.
A common detail page display image can be generated with one sentence. Jiaotu AI will automatically summarize the product's selling points, and at the same time can clearly and beautifully highlight the selling point information.
Prompt: Generate an e-commerce detail page efficacy key visual for the product in the image, with molecular-level display of the product's core ingredients, expressing the product's core ingredients from a microscopic perspective.
Not just physical display; Jiaotu AI can also handle abstract expressions such as molecular-level structures and mechanisms of action very well.
Play 6 Detail Image
Prompt: Generate an e-commerce detail page efficacy key visual for the product in the image, through a hand model display, introduce the product's usage in 3 steps, and help users fully understand the product.
Prompt: Generate an e-commerce detail page efficacy key visual for the product in the image, expressing the product's core ingredients from a microscopic perspective. The core ingredients include: urea, glycerin, panthenol, etc., displayed in different bubbles for distinction. The image uses an asymmetrical composition, with the hand cream product on one side of the image and occupying a relatively small proportion of the image. Suitable for the ingredient explanation module of an e-commerce detail page.
For traditional art designers to make such an ingredient display image, they first have to figure out the main ingredients, then search for various 3D materials of molecular structures, and then adjust lighting, transparency, and so on step by step. Making one image takes at least twenty to thirty minutes... Jiaotu AI will automatically combine elements such as molecules, bubbles, and light effects, and get it done in 2 minutes.
Play 7 Change Product Color
Prompt: Based on real product photos, generate an e-commerce scene image, modern life aesthetics style, clean, natural, with negative space. The product in the image is the main featured version, add multiple versions with only the packaging color changed (pink, purple, blue, green, black, yellow), with proportions and texture completely consistent. Multiple products are arranged in an orderly staggered manner, with a clear visual focus, and the lighting is soft natural light or diffused light, with a real life scene feel but professionally controlled. The overall image is unified, restrained, and high-end, suitable for e-commerce product combination display and series SKU scene images.
The product color change effect generated by Jiaotu is merely a color change, but the product's proportions and structure are completely identical to the original, and the texture has not changed at all.
Gameplay 8 Translation
Prompt: Keep the product unchanged, only translate the following Chinese information on the product packaging: "Urea Hand Care Cream", "Moisturizing and Repairing Skin", "Anti-dry, Anti-crack, Anti-freeze", translate the text into [Korean].
The language has changed, but the overall layout structure is the same as the original. Even with different language lengths, the packaging visual consistency is maintained, and the translations are relatively accurate. The generated text and image have a high degree of integration.
Gameplay 9 Promotional Poster
Prompt: A modern and fashionable Christmas promotional poster, full of vibrant and playful energy. The background is a highly saturated festive red, paired with graphic elements such as Christmas snowflakes and gifts, creating a strong holiday atmosphere. A high-quality photo of an Asian woman full of dynamism and vitality. Her fashion is modern-style winter clothing. The photo is presented in a bright, full-color style, with natural and realistic skin tones subtly reflecting the warm tones of the red background. The woman is physically interacting with the huge number "50," her arm passing through the "0" and leaning on the "5." She is holding the hand cream from the image. At the bottom of the page, the text headline "50% OFF" and the subtitle "Christmas Carnival" are arranged in a bold, dynamic, and lively typographic combination, using complementary modern fonts. The numbers and symbols are treated as graphic elements, varying in size and overlapping each other, creating a sense of movement and excitement. The font is a clean and geometric sans-serif.
The visual weight of the promotional information is very accurate. You can see the 50% discount at a glance. The festive elements are dotted in the image, making it clear at a glance that it is a holiday event.
Gameplay 10 Canvas Image Editing
Jiaotu AI is not only great at generating images, but also very good at editing them. All generated images are placed on an infinite canvas, and clicking on an image allows you to perform various operations.
Outpainting, enhancing clarity, and removing unwanted elements—these functions are not actually novel if discussed individually, but Jiaotu AI integrates them all into the canvas.
From the initial concept to partial completion and then overall adjustment, the entire creative process for an image is coherent and smooth.
02. More Gameplay
Virtual Try-On
Prompt: Change the girl's jacket in Image 1 to that in Image 2
Other character details can also be modified through reference images:
Prompt: Change the girl's hat in Image 1 to that in Image 2, keep everything else unchanged
Prompt: Change the girl's jacket in Image 1 to that in Image 2, keep the person completely consistent (face, hairstyle, skin tone, body shape, and temperament unchanged). Under the premise of maintaining the person's identity consistency, randomly generate different shooting angles and poses, with a natural and relaxed posture, realistic human body structure, normal proportions, and movements that conform to daily shooting logic. The lighting is natural and coordinated, consistent with the environment, and the overall style is realistic.
By repeatedly using this prompt, you can get a complete set of try-on images of the same model and scene.
Zootopia Group Photo
Prompt: Keep the person in the reference image unchanged, with consistent facial features, skin tone, hairstyle, and temperament. Generate a vivid, natural, photorealistic group photo. The scene is inside a movie theater, with dim, soft lighting; the faint reflection from the screen makes the characters' expressions look soft and vivid. The girl in the reference image sits in the middle, holding a bucket of popcorn, her body slightly relaxed and leaning forward, with a relaxed expression and a natural smile. On the left is Nick Wilde, the fox from Zootopia, and on the right is Judy Hopps, the rabbit. Both naturally lean toward the girl in the middle, with relaxed postures and smiles and body language carrying a slight exaggerated cinematic-character feel, like friends watching a movie together. Nick and Judy are highly photorealistic, with realistic and delicate fur texture, reasonable proportions, and natural expressions. The composition is balanced, the atmosphere relaxed and warm, with a photorealistic photographic quality, realistic depth of field, and lighting.
Magazine cover
Prompt: Naturally integrate the characters in the reference image into a real scene at Shanghai Disneyland, letting them sightsee, interact, and take photos in the park as tourists. The overall layout of the image should be in the style of a beautiful magazine cover, with refined composition and clear layering. As the cover design of a high-quality magazine or travel album, emphasize the sense of integration between the characters and the scene. The image should have strong visual appeal, with bright but not harsh colors, vivid and natural characters, and iconic Disney elements in the scene.
The check-in location can also be changed to any precise latitude and longitude; we can check in at any corner of the Earth.
Prompt: Naturally integrate the characters in the reference image into a scene, presenting them sightseeing at the location with latitude and longitude [63.615623, -19.988685]. The overall layout of the image should be in the style of a beautiful magazine cover, with refined composition and clear layering. As the cover design of a high-quality magazine or travel album, emphasize the sense of integration between the characters and the scene. Please design this page as a very important showcase page, so that viewers immediately become interested at a glance and are willing to keep learning about this location.
03. Some sharing
A top-tier model like Nano Banana Pro is already very capable, but because of limited usage environments, not many people have actually tried it. At Jiaotu AI (www.jiaotuai.cn），you can directly experience top-tier image and video models in China. New registrations get free usage credits, and logging in daily also lets you get a few free uses of Nano Banana Pro.。
This allows more ordinary people to create images close to professional standards with almost no learning cost.
In high-frequency, fast-paced scenarios such as e-commerce and marketing materials, many images do not require extreme creativity; instead, stability, efficiency, and reusability are valued more.
Key visuals, detail page display images, SKU color variations, promotional posters, and multilingual version switching—these steps that used to consume a lot of manpower can now all be handled at relatively low cost.
From a longer-term perspective, the learning path for designers may change. Whether someone can use a particular software may no longer be the only measure; being able to think through requirements clearly and express them clearly may instead become the new core ability.
Original link:Once you master Nano Banana Pro, PS really is not necessary