Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
To get more control over an image, give the model a few purposeful visual references and explain what to take from each one. Use the prompt to map the references to the job, describe the new scene, and identify what must stay unchanged. This is more useful than trying to recreate every visual detail in prose—but references guide generation; they do not guarantee an exact match.
Why a reference sheet can work better than a longer prompt
A text prompt has to describe the subject, style, composition, colors, materials, and scene. If you already have images that show some of those qualities, use them as visual specifications instead. The prompt then does the important connecting work: it tells the model which image contributes which traits and how those traits should combine.
OpenAI Academy advises that one to three clear sentences are often enough for an image prompt. That is product guidance, not a universal length limit or a measured guarantee of better results. The useful lesson is to keep the request focused, whether it takes one sentence or several: OpenAI Academy’s image-creation guidance.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsA reference sheet does not need to be a designed collage. It can simply be a small set of uploaded images. Choose references that answer different questions rather than several images that all show the same thing.
#1 Best Overall
Choose references for distinct jobs
Before uploading, decide what each image should tell the model. Common roles include:
- Subject or identity: the character, person, or object whose recognizable features should carry into the result.
- Style: the medium, visual treatment, or overall feel to borrow.
- Composition: the layout, pose, viewpoint, or placement of elements.
- Palette or material: colors, surface qualities, or textures to carry over.
- Setting: the environment or background, if it should come from a reference rather than the new scenario.
Give each uploaded image a clear label or number in your prompt. Do not assume the model will know whether a reference is meant to control the character, the style, or the layout. OpenAI’s guidance recommends referring to images by order and explaining their relationship; Google also describes using references for style, color, composition, character consistency, and combining a product with a new environment: OpenAI image guidance and Google’s prompt guidance.
Build the prompt in layers
Use the following elements as a scaffold, not a required formula. Leave out anything that does not matter to the image.
- Goal: Name the deliverable or intended use, such as a character illustration or a product image.
- Subject and action: Say who or what appears and what is happening.
- Scene and framing: Specify the setting, viewpoint, composition, and spatial relationships that matter.
- Reference map: Assign each image a role and name the traits to borrow from it.
- Invariants: State which identity, layout, objects, or visual features should remain consistent.
- Change request: Describe the new scenario or the one edit you want.
- Exclusions: Mention unwanted text, logos, objects, or background elements only when relevant.
For example, this is a prompt template, not a tested result:
Image 1 is the character reference; Image 2 is the watercolor style reference. Show the same character waiting at a train station, keeping the face and clothing from Image 1 while using Image 2’s palette and brush texture.
The first sentence assigns the references; the second describes the new scene and explains how to combine them. Google’s published scaffold for a blank-canvas prompt is subject, action, location or context, composition, and style. For reference-guided work, its guidance adds the references, their relationship, and the new scenario: Google Docs Editors Help.
For edits, separate what changes from what stays fixed
When revising an existing image, state the requested change directly and name the important parts to preserve. For example: “Change only the jacket color. Keep the person’s identity, pose, framing, and background.” This makes the scope of the edit clear without asking the model to reinterpret the entire image.
Make one targeted revision at a time, then inspect the result. If the lighting is wrong but the subject and composition work, ask for a lighting adjustment and restate the key elements that should remain fixed. OpenAI Academy recommends small, targeted refinements with clear preservation instructions; Google likewise recommends reviewing the output and adjusting particular aspects such as color, lighting, or objects: OpenAI Academy and Google’s image-generation guidance.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.When the output drifts, simplify the instruction
If the result starts borrowing the wrong traits or changing elements you wanted to preserve, adding more references or a longer description may make the request harder to follow. Try this reset:
- Restate only the most important reference roles and constraints.
- Remove redundant images so each remaining reference has a distinct purpose.
- Adjust one variable at a time, such as style, lighting, or a single object.
- Review the new output before adding further detail.
OpenAI says a small set of uploaded images is usually easier to manage than a large one. That is practical product guidance, not a claim that any specific number of images is best across all tools: OpenAI Academy.
Pick a tool based on how you work
Image tools overlap in their support for visual references, but their controls and workflows differ. Compare the features that matter for your task rather than assuming that one tool will produce better images in every case.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →| Workflow question | Why it matters | Documented examples |
|---|---|---|
| Can you distinguish style from composition references? | Useful when the visual treatment should come from one image and the layout or arrangement from another. | Adobe Photoshop Help describes style and composition reference choices: Photoshop reference-image guidance. |
| Can you provide multiple references and explain their relationship? | Useful for assigning separate jobs to subject, palette, style, or setting images. | OpenAI and Google document multi-image reference workflows: OpenAI image guidance and Google prompt guidance. |
| Can you make targeted edits and state what must be preserved? | Useful when changing one element without requesting a new interpretation of the full image. | OpenAI describes reference-based editing, while Google provides examples of targeted transformations: OpenAI image prompting documentation and Google image-generation guidance. |
| Does the workflow fit your task? | A conversational generator, a document-oriented workflow, and a full image editor may expose different controls and suit different working styles. | OpenAI, Google, and Adobe describe features in their own products; their guidance is not an independent image-quality comparison. |
Availability and interface steps may vary by product, account, and region. Check the current help page for the tool you use before relying on a specific control.
What reference images cannot promise
References guide a model; they do not guarantee exact preservation of a face, outfit, pose, layout, or other detail, nor do they guarantee identical results across tools. The official guidance cited here provides workflows and examples, not a cross-model benchmark for fidelity, time savings, or quality improvement. Treat each output as something to inspect and refine rather than as a guaranteed reproduction.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
