reference image AI generator

Turn the feeling of a reference image into new AI visuals.

Instead of translating taste into complex prompts, upload a reference image. DOBIDOBI reads the light, composition, color, and mood, then creates new product, model, and ad visuals.

Start from the visualPreserve the key moodMove it onto your productExtend into a series
Full-body model in a champagne satin dress
Model wearing a black top and white skirt with a black handbag

Selected site showcases: dress drape, handbag proportions and outerwear texture.

01

Upload assets, get shots

Product, background, and model images can all become the starting point.

02

Generate around the reference

A clear visual direction means fewer random attempts.

03

No complex prompt needed

Style, lighting, and composition follow the reference.

Case examples

Clothing, texture and collection imagery.

Full-body model in a champagne satin dress

Dress drape and light

A satin dress showcase with soft light, visible folds and a clear silhouette.

Model wearing a black top and white skirt with a black handbag

Handbag and outfit

A black handbag in a complete monochrome outfit, showing its scale alongside clothing.

Generated brown suede jacket image

Outerwear texture

A generated jacket image from a verified project, showing suede texture, tailoring and neutral tones.

Beige trench coat generated using the same photographic Style

The same Style on a trench coat

A saved photographic Style used for another product, retaining soft light and a neutral background.

Three-view lookbook of a pastel outfit with pink trousers

Multiple lookbook views

A site lookbook example combining full-body, closer and rear compositions.

Case breakdown

One reference image becomes a full editorial set.

The same look can continue into front, side, back, detail, and full-body shots.

Reference

Lookbook input showing pink trousers and a pale top

Single-image extension

A pastel lookbook in multiple views

The existing lookbook example extends one input into a set of full-body, closer and rear views. Review garment and accessory details before delivery.

Three generated views of the same pastel outfit
Generated multi-view result

Reference

Brown jacket result using the saved Style

Same Style, another product

Reuse a photographic Style from jacket to trench

A saved Style from a real project was used for a jacket and then a trench coat. These are generated results for two different products.

Generated beige trench coat in soft light against a neutral background
Trench result using the same Style

Why it works

A reference image is often the most stable visual language for a team.

Brand, design, operations, and photography teams may interpret words differently, but one reference can align taste quickly. DOBIDOBI turns that direction into a generative workflow.

01

Upload the reference image

Use a sample shot, moodboard, competitor style, product photo, or portrait.

02

Define the output goal

Choose whether you need product images, model visuals, ad assets, lookbook frames, or video stills.

03

Generate and keep editing

Add strong results to the canvas for background changes, outpainting, recomposition, or video.

Do not start from blank. Let the reference speak first.

Turn existing taste into a visual workflow that can be generated, compared, and edited.

Will it copy the reference image exactly?+

The goal is to understand and extend the visual direction, not copy the original. Confirm reference rights and platform rules before commercial use.

Can a product photo be a reference?+

Yes. A product photo can be the subject, while another image can provide style, scene, or lighting direction.

Who is this useful for?+

Brands, ecommerce teams, designers, photographers, and content teams use it to align direction and create candidate assets quickly.

Practical production guide

Read the step-by-step guide