Ideogram 4 Structured JSON Reconstruction for Image Cleaning

This is an AI image inpainting application based on Ideogram 4 + Reference Image Analysis + Structured JSON Prompt Reverse-Engineering + Image Inpaint Generation.

Users only need to upload a reference image, such as a cover, poster, product shot, visual design, advertisement, logo, tattoo, or social media graphic, and the system will automatically analyze the subject, background, composition, text, color, lighting, and style, organizing them into a structured JSON Prompt better understood by Ideogram 4.

Unlike ordinary image-to-image workflows of "upload image + write prompt", this process is more like first "understanding the image", then "reorganizing the image", and finally regenerating the result.
In other words, it doesn't just make simple latent space modifications based on the original image; it breaks the original image down into structured information and hands it over to Ideogram 4 to redraw from scratch.

The entire workflow automatically completes:

• Reference image analysis
• Structured JSON reverse-engineering
• Aspect ratio organization
• Prompt encoding
• Ideogram 4 scheduling and sampling
• Image inpainting output

The final output is a newly generated image, suitable for layout replication, cover redrawing, product image refreshing, poster reorganization, ad regeneration, tattoo redrawing, and batch production of social media graphics.

If you are packaging it as an RunningHub application, its biggest appeal is:

Upload a reference image, and the system automatically understands the visual structure and generates a new, high-quality inpainting result with one click.