Tuesday, December 9, 2025 / Clicky News
Studio quality imagery using Nano Banana Pro & JSON prompts for an incredible level of detail & endless variation
Studio quality imagery using Nano Banana Pro & JSON prompts for an incredible level of detail & endless variation
Giving detailed prompts in JSON to Google Nano Banana Pro means getting an almost perfect image. As ever it's all about the prompt, but get it right and the results are impressive, staggeringly detailed and endlessly variable.
Giving detailed prompts in JSON to Google Nano Banana Pro means getting an almost perfect image. As ever it's all about the prompt, but get it right and the results are impressive, staggeringly detailed and endlessly variable.




Author
Author
Oli Yeates
Oli Yeates
The evolution of AI imagery is moving away from guesswork and towards absolute precision. With Nano Banana Pro, powered by Google’s Gemini 3 Pro model, we have moved beyond standard text prompts to structured JSON data. This shift allows us to achieve an incredible level of detail that traditional prompting simply cannot match.
Here is how we use structured data to create high-end editorial assets.
The Prompt
Instead of writing a long, winding sentence, we feed the model specific key-value pairs. This acts as a checklist for the AI, ensuring every element is accounted for.
Take a look at this prompt we used for a minimal editorial shoot featuring Bala weights:
JSON
{ "scene": "Studio photoshoot, soft warm beige seamless background, minimal editorial", "subject": "Female model standing, arms in shot, sleek ponytail, blush beige bodysuit", "product": "Blush beige Bala weights worn on wrists, visible logo, neutral 2lb model", "tech": "Soft studio lighting, 50mm f/2.8, 8k ultra photorealistic, shallow depth of field", "negative": "glasses, sunglasses, color mismatch, distorted anatomy, cartoon, low res" }
Why It Works
This format provides granular control over the final output.
Scene & Subject: We define the exact aesthetic ("minimal editorial") and styling details ("sleek ponytail") separately. This prevents the background style from bleeding into the model's look.
Product Precision: By isolating the
productfield, we ensure the Bala weights are not just an accessory but the focal point, with specific instructions for the "visible logo" and "2lb model".Technical Specs: The
techfield acts as our virtual camera. specifying "50mm f/2.8" tells the engine exactly how to handle depth of field, giving us that professional, soft-focus background while keeping the product sharp.Negative Prompts: Crucially, we tell the model what not to do. This filters out common errors like "distorted anatomy" or unwanted accessories before the image is even generated.
The Result
The output is a consistent, high-fidelity image that sits perfectly within a brand's visual identity. It is not just about making a pretty picture; it is about creating a usable commercial asset that requires minimal post-production.

Nano Banana Pro allows us to replicate this setup infinitely. We could change the product colour to "sage green" or the subject to "blonde female" while keeping the camera settings identical. In this example the model is holding another weight, we could omit this in the next prompt by adding it to the negative element but for now we will keep it.

We could change the product colour to "blush pink" or the scene to "directly face the camera" while keeping the other settings identical.

Now lets add a second model to the scene one blonde and one brunette…

This is the future of digital production. creative freedom backed by technical rigour. The possibilities and variations are endless and limitless. Love it or hate it, these really are unprecedented times in creative possibilities.
The evolution of AI imagery is moving away from guesswork and towards absolute precision. With Nano Banana Pro, powered by Google’s Gemini 3 Pro model, we have moved beyond standard text prompts to structured JSON data. This shift allows us to achieve an incredible level of detail that traditional prompting simply cannot match.
Here is how we use structured data to create high-end editorial assets.
The Prompt
Instead of writing a long, winding sentence, we feed the model specific key-value pairs. This acts as a checklist for the AI, ensuring every element is accounted for.
Take a look at this prompt we used for a minimal editorial shoot featuring Bala weights:
JSON
{ "scene": "Studio photoshoot, soft warm beige seamless background, minimal editorial", "subject": "Female model standing, arms in shot, sleek ponytail, blush beige bodysuit", "product": "Blush beige Bala weights worn on wrists, visible logo, neutral 2lb model", "tech": "Soft studio lighting, 50mm f/2.8, 8k ultra photorealistic, shallow depth of field", "negative": "glasses, sunglasses, color mismatch, distorted anatomy, cartoon, low res" }
Why It Works
This format provides granular control over the final output.
Scene & Subject: We define the exact aesthetic ("minimal editorial") and styling details ("sleek ponytail") separately. This prevents the background style from bleeding into the model's look.
Product Precision: By isolating the
productfield, we ensure the Bala weights are not just an accessory but the focal point, with specific instructions for the "visible logo" and "2lb model".Technical Specs: The
techfield acts as our virtual camera. specifying "50mm f/2.8" tells the engine exactly how to handle depth of field, giving us that professional, soft-focus background while keeping the product sharp.Negative Prompts: Crucially, we tell the model what not to do. This filters out common errors like "distorted anatomy" or unwanted accessories before the image is even generated.
The Result
The output is a consistent, high-fidelity image that sits perfectly within a brand's visual identity. It is not just about making a pretty picture; it is about creating a usable commercial asset that requires minimal post-production.

Nano Banana Pro allows us to replicate this setup infinitely. We could change the product colour to "sage green" or the subject to "blonde female" while keeping the camera settings identical. In this example the model is holding another weight, we could omit this in the next prompt by adding it to the negative element but for now we will keep it.

We could change the product colour to "blush pink" or the scene to "directly face the camera" while keeping the other settings identical.

Now lets add a second model to the scene one blonde and one brunette…

This is the future of digital production. creative freedom backed by technical rigour. The possibilities and variations are endless and limitless. Love it or hate it, these really are unprecedented times in creative possibilities.
more BLogs
more BLogs





