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
Image-to-image AI gives you another way to steer a generation: provide a reference image for visual guidance and use text to describe the content or changes you want. Depending on the tool and reference mode, the image may influence composition, color, style, shape, or other visual structure. It is influence, not a guarantee of an exact copy or pixel-perfect preservation.
What image-to-image AI lets you control
A text prompt describes an image in words. An image-conditioned workflow adds an existing visual input, so the model has visual information that can be difficult to spell out precisely. In the Plug-and-Play Diffusion paper, the guidance image provides layout while text guides semantics and appearance. The authors demonstrate translating sketches and drawings, changing an object’s appearance, and modifying lighting or color (Plug-and-Play Diffusion Features for Text-Driven Image-to-Image Translation).
The reference’s role depends on how a tool interprets it. It might contribute content or composition, or it might be used chiefly for style or structure. A useful prompt therefore separates the two jobs: describe what should appear or change in text, and choose a reference mode suited to the visual quality you want to guide.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesChoose the reference for the job
Content and composition
An image prompt can give the model a visual starting point or inspiration for a new composition. Midjourney says its image prompts and references guide new creations rather than copy an input exactly. For precise changes to an existing image, its documentation points users to the Editor instead (Midjourney Image Prompts documentation).
#1 Best Overall
Style and visual consistency
A style reference is intended to guide look and feel rather than dictate the exact subject. Adobe describes this mode as a way to guide generated variations and develop consistent-looking assets (Adobe Style Image Reference documentation). Midjourney likewise treats style references as a distinct control, separate from image prompts (Midjourney Style Reference documentation).
Structure, pose, and spatial layout
Structure references can guide characteristics such as an image’s outline and depth. Adobe’s example varies strength and generates four alternatives to compare; the API documentation does not establish that every Adobe interface exposes identical controls (Adobe Structure Image Reference documentation).
For more explicit spatial guidance, ControlNet research describes conditioning text-to-image models with edges, depth, segmentation, human pose, and scribbles. These inputs can convey layout, shape, or pose more directly than prose alone. The authors identify complex layouts, poses, shapes, and forms as cases that can be difficult to specify through text alone (Zhang, Rao, and Agrawala, “Adding Conditional Control to Text-to-Image Diffusion Models”).
How the main reference controls differ
The documented controls below do not share a common scale. Their numbers describe settings within each product, not comparable amounts of influence.
| Tool and control | What it guides | Documented setting | Practical qualification |
|---|---|---|---|
| Midjourney image weight | How strongly an image prompt affects the result | Default 1; range 0–3 for versions 8.1 and 7, and 0–2 for Niji 7 | Ranges depend on model version and may change. This is not the style-weight control. |
Midjourney style weight (--sw) |
Influence of a style reference | Default 100; range 0–1000 | Midjourney advises keeping the text prompt simple and describing the desired content rather than telling the model how to modify the reference. |
| Adobe Firefly style-reference strength | Influence of a style reference on look and feel | Default 50; range 1–100 in the API documentation | Adobe’s API value is not numerically comparable with Midjourney’s settings. |
| Adobe Firefly structure reference | Structural characteristics such as outline and depth | A strength setting is shown; a numeric range and default are not stated in the cited documentation | The example generates four variations for comparison; that does not establish identical controls in every interface. |
Midjourney’s image-weight figures are documented for the named versions, and its style-weight setting is a separate control. Adobe’s strength values belong to its API documentation. Treat each setting as a tool-specific way to adjust its own workflow, not as a universal “denoise” equivalent (Midjourney Image Prompts; Midjourney Style Reference; Adobe Style Image Reference; Adobe Structure Image Reference).
How to get more useful results from a reference
- Decide what should carry over. Choose an image for composition, style, or structure according to the available reference modes. One image need not be expected to guide every quality equally.
- Describe the target in text. State the subject, desired changes, and relevant appearance. For Midjourney style references, its documentation recommends describing the content you want and keeping the prompt simple rather than instructing the model how to edit the reference.
- Adjust only the relevant control. Change image weight when tuning an image prompt’s influence, or style weight when tuning Midjourney’s style reference. Use the selected product’s own range and model-version guidance; do not transfer a number from another tool.
- Generate alternatives and compare. A setting changes influence, not certainty. Compare variations for the qualities that matter, then revise the prompt, reference, or relevant strength if the results miss the target. Adobe’s structure-reference example explicitly illustrates generating four variations to compare.
- Use an editing workflow when exact local changes matter. Midjourney directs users to its Editor for precise changes to their own images; an ordinary image prompt or reference should not be treated as guaranteed preservation.
Why a reference does not guarantee an exact result
A model uses the reference as guidance in a generation process, not necessarily as a protected layer to reproduce pixel for pixel. The chosen mode, the prompt, and the model’s interpretation all affect the output. Midjourney explicitly frames references as inspiration for new creations rather than exact copying, while spatial-conditioning research demonstrates ways to guide structure without promising perfect adherence (Midjourney Image Prompts; ControlNet paper).
So, when a creator asks for the Midjourney equivalent of a setting from another image-generation workflow, there is no universal one-to-one conversion established here. First identify what the other setting is meant to preserve or change; then choose the closest Midjourney control, if one exists, and tune it through variations. Image weight and style weight affect different reference functions, and neither should be assumed equivalent to another product’s numerical scale.
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.
