Choose a source
Use an image you have permission to work with. Pick one where the subject you want to retain is easy to see.
Reference-led creation
Use an ai image generator from image references
Start with a picture you already have, then describe the change you want. Imagegen helps you think through the source-to-result workflow, whether you are changing a style, setting, or composition.
What this variant is
Image-to-image generation uses an existing picture as a visual reference and a written instruction to guide a new result. The reference gives the process something concrete to preserve or reinterpret.
Why turn one image into another?
A reference can communicate shapes, colors, and framing faster than a long description. A prompt then identifies which parts should change, so the result has a clearer direction.
Separate fixed from flexible
Name the elements to keep, such as silhouette or camera angle, then specify the new setting, lighting, or style.
Review the result
Compare the output with your source. If a defining detail changed, make that constraint more explicit in your next instruction.
The tool block Source and result specifications
For an image-to-image task, these are the practical differences between what you provide and what you inspect afterward. Exact supported formats and output limits depend on the tool you open.
Reference image
Generated image
1
Role
Reference image
Provides visual starting information
Generated image
Shows the requested reinterpretation
2
Subject
Reference image
Contains the original subject
Generated image
Should retain the subject if the prompt asks for it
3
Composition
Reference image
Establishes framing and placement
Generated image
May follow or depart from that framing
4
Style
Reference image
Shows the source treatment
Generated image
Can apply a newly described treatment
5
Background
Reference image
Shows the original environment
Generated image
Can depict a different environment
6
Review focus
Reference image
Identify details worth preserving
Generated image
Check those details and look for unwanted changes
Spec table in practice
The most useful prompt depends on what the source contributes. Each workflow below starts by identifying that contribution before asking for a change.
Product maker
You have a clear product photo but need to explore a different setting.
Keep the product silhouette and viewpoint; describe the new surface, background, and light. For a concept with no source photo, start from a written description instead.
ai image generator from text freeIllustrator
A rough composition communicates the scene, but not its final visual style.
Preserve the placement of major subjects while describing the intended palette and treatment. Specific examples can help you write that instruction.
image generation prompt examplesContent creator
An existing scene has the right subject but needs another mood.
Ask for a change in lighting and atmosphere, then check whether the subject still reads clearly. A step-by-step workflow helps with revisions.
How to do image generation?Make a reference-led variation
Give your source image a new direction
Describe what must stay recognizable and what should change. Open the tool, add your reference there, and compare its output with your original before using it.
- Start with a source you can use
- State the details to preserve
- Check the result against your request
Variant FAQ
It uses an existing picture to guide the creation of another image. Your written instruction can ask for changes to style, lighting, background, or composition while identifying details you want to keep.
Not necessarily. A generated result may alter small features even when you ask to preserve them, so compare important shapes, text, and identifying details with the source.
You can request that, but the result still needs inspection. Describe the subject features and camera angle you want retained, as well as the replacement background.
Choose one you have permission to use and that clearly shows the elements you want carried forward. A source with an unobstructed subject makes it easier to state what should remain recognizable.
Image-to-image begins with visual information already present in a source picture. Text-only generation begins with your description, so you must explain the subject and composition without a visual reference.