Style Reference
What is Style Reference?
A style reference is an image you show an AI generation tool to communicate what you want the output to look like: it is easier and more accurate to show the AI an example of the visual style you want than to try to describe it entirely in words.
At a glance
- Also known as
- Visual referenceStyle imageReference imageAesthetic reference
- Used for
- Communicating target visual aesthetic to AI generation modelsEstablishing consistent visual language across a projectGuiding colour palette and lighting quality in generated contentTranslating mood board references into generation inputs
- Key features
- Provides direct visual aesthetic information rather than text descriptionConditions generation outputs toward a specific look and feelCan be combined with text prompts for precise style controlSupports visual consistency across multiple generation sessions
- Related terms
- Style transferReference imageMood boardIP-adapterStyle guideLoRA
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
How it compares
Compared with related concepts
A style reference and a content reference serve different functions in generation workflows. A style reference communicates how generated content should look: its aesthetic, treatment, and visual character. A content reference communicates what the generated content should contain: a specific character's appearance, a location's physical features, an object's design. Many generation workflows use both simultaneously: a content reference anchors the subject matter, while a style reference guides how that subject matter is rendered. Conflating the two can produce unintended results, such as a character reference image that inadvertently conditions the style of the output toward the aesthetic of the reference photograph rather than the target visual treatment.
Think of it like…
A style reference works like showing a painter a photograph and saying "paint whatever subject you like, but render it the way this image is rendered" — the reference communicates not what to paint but how to paint it, directing the visual treatment, the quality of light, the handling of colour, and the texture of the surface without specifying the content.
Pro tip
Isolate the aspect of visual style you want the reference to communicate and choose your image accordingly. If you want to specify colour palette, find a reference with a very clear, intentional colour treatment but otherwise neutral content. If you want to specify lighting quality, find a reference with strong, characteristic lighting in a context that won't pull the generation toward unintended subject matter. Using a reference image whose content closely resembles your intended generation subject will condition both style and content simultaneously, which can be powerful but requires awareness that both aspects of the reference are influencing the output.
Types and variations
- Style references can be categorised by the aspect of visual style they communicate.
- Colour references provide examples of specific palette relationships: the warm amber and shadow combination of a particular film, the desaturated cool tones of a documentary aesthetic.
- Lighting references communicate the quality, direction, and contrast of light in the target output.
- Compositional references demonstrate framing conventions and spatial organisation.
- Texture references show surface qualities: painterly brushwork, photographic grain, clean digital rendering, analogue degradation.
- A sophisticated generation workflow might use multiple reference images simultaneously, each addressing a different dimension of the target aesthetic, rather than relying on any single image to communicate the entire visual direction.
Ready to make your first scene in Morphic?
Try MorphicCommon use cases
- Style references are used whenever consistent visual aesthetic is important across a body of generated work.
- Brand content production uses style references to ensure all generated assets align with the brand's established visual identity.
- Narrative film and video production uses references to maintain the cinematographic language of the project across clips generated in different sessions.
- Advertising production uses references to anchor generated content to the approved visual direction of a campaign.
- Character artists use style references to establish the illustration style that all character designs must conform to.
- Any creative workflow involving multiple generation sessions, multiple team members, or multiple output formats benefits from the visual alignment that shared style references provide.
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
FAQs
Different AI systems handle style references in different ways, but most encode the reference image into a vector representation that captures its visual characteristics: colour distribution, texture, spatial frequency, lighting quality: and then use this representation to condition the generation process alongside the text prompt. The degree to which the reference influences the output versus the text prompt is often controlled by a strength or weight parameter, allowing creators to blend reference conditioning with prompt direction at different ratios.
A good style reference clearly represents the target aesthetic without containing competing visual styles or distracting content. It should be technically clean: sharp, well-exposed, and free of compression artefacts. It should be relevant to the type of content being generated: a reference from a closely related genre, medium, or production context will condition outputs more effectively than one with a different visual language. High-contrast, stylistically distinctive references tend to produce stronger conditioning effects than images with neutral or averaged aesthetics.
Film stills are among the most commonly used style references in AI generation workflows, as they efficiently communicate cinematographic language including colour grade, lighting quality, lens characteristics, and compositional approach. A well-chosen frame from a film whose aesthetic matches the target visual direction can condition AI generation outputs toward that cinematic look more precisely than extended text description. When using film stills, be aware that the content of the frame ( its characters, environment, and staging ) may also influence the generation content, not only the visual style.
The optimal number depends on the generation system and the complexity of the target aesthetic. Single references work well when the target style is unified and clearly represented by one image. Multiple references allow different visual dimensions to be specified separately: colour from one reference, lighting from another, texture from a third: but increase the risk that conflicting visual information produces an averaged or incoherent result. Most generation tools support two to four simultaneous references effectively; beyond this, the conditioning signals tend to interfere with each other.
A style reference conditions a single generation session by providing visual information at inference time. A LoRA is a fine-tuned model component trained on a set of style examples that encodes that style into the model's weights, affecting every generation without needing a reference image at each session. LoRAs produce stronger and more consistent style conditioning than reference images for well-defined styles, but require a training process and a sufficient body of training examples. Style references are more flexible and require no training, making them the default approach for style conditioning and LoRAs the appropriate tool when a specific style needs to be applied consistently at production scale.
Yes, and style references are particularly valuable in video generation because maintaining consistent visual aesthetic across multiple clips in a production is more challenging than applying it to a single image. Providing the same style reference across all generation sessions for a project anchors the visual language in a way that text prompts alone cannot reliably sustain. Some video generation platforms allow style references to condition not only the colour and light quality of the output but also the motion character and camera movement aesthetic, extending style conditioning beyond the static visual treatment into the temporal dimension of the content.
Morphic stores style references in the Assets tab of a project, alongside character references, location references, and other input materials. Organising all style references in the project's Assets tab at the outset of a production ensures they are available consistently across all generation sessions within the project, and that all team members working on the project have access to the same reference materials. Naming and annotating reference images in the Assets tab helps maintain clarity about which reference communicates which aspect of the visual direction as the project grows.
A text prompt describing a style communicates aesthetic qualities through language, which the model interprets based on associations learned during training. A style reference communicates visual qualities directly through the actual visual data of the image. Text descriptions are imprecise: different people mean different things by words like "cinematic," "moody," or "painterly" — while a reference image communicates exact colour relationships, contrast ratios, and textural qualities without ambiguity. The most effective approach combines both: a style reference anchors the visual treatment while text prompts add specificity about subject, context, and aspects of the style that the reference alone cannot communicate.
