Skip to content
VYRA

AI image generator
for creators & brands

Nano Banana, Seedream, Flux and more. Reference images, sharp text, up to 4K. Free Flux drafts every month.

How to make an AI image on VYRA

Start with the image’s purpose, not a pile of style words. A social post, a product concept and a video start frame each need a different composition. Decide where the subject should sit and whether you need room for a headline. Write one clear sentence describing the subject and scene, then add the light, medium and camera perspective. Generate an initial draft and review it at full size. Change one requirement at a time so each revision teaches you something useful.

Choose an aspect ratio before you generate. A square crop suits many feed posts and product studies, a tall image works for stories and vertical video, and a wide frame works for banners. Leaving extra space around the subject can help with later cropping, but it is not a substitute for planning. Check the selected model’s controls: reference support, resolution and available settings differ. Features supported by one model are not automatically supported by all the others in this studio.

Choose the right image model for the job

The model selector lets you compare approaches without moving between separate accounts. Try Grok Imagine for a text-led visual concept, or explore Seedream when you need a model with reference-image support. Read each model’s page for its current integration rather than relying on a version name alone. Auto can choose a model from your prompt, but manually selecting one is useful for a controlled comparison. Keep the subject, crop and prompt stable when you want to learn how two models differ.

Do not assume that the newest name is the right engine for every job. A cheaper draft can answer a composition question before you spend on a final image. A reference-capable model may be a better fit for an existing product than a text-only model. If typography is central, test a small portion of the layout and inspect spelling. No model choice removes the need to check the result. Factual diagrams, product claims and real-world identifiers require independent verification, not just visual approval.

Use reference images deliberately

On models that accept references, upload a clear image you own or have permission to use. Explain what should remain and what should change: keep the shape of the bottle, replace the background with a stone counter, preserve the camera angle. This is more helpful than adding a photo with an unrelated prompt. Avoid feeding several conflicting references into the first test. Begin with one reliable view, then add more only if they communicate a necessary product detail or consistent visual direction.

A reference is guidance, not a guarantee of a pixel-perfect copy. Review labels, geometry, colours and small details after generation. If an existing logo or legal label must be exact, retain or apply that artwork in an appropriate design workflow rather than accepting a near match. For faces and personal photographs, get consent and follow the content policy. Do not upload material merely because it is easy to find online. Rights, privacy and the suitability of the output remain important even during an early concept stage.

Prompt examples you can adapt

For a product study, try: a matte green ceramic mug on a light oak table, soft window light from the right, close studio photograph, warm neutral background, clear empty space above the handle. The brief defines the object, material, light and layout. Change the material or background first if the overall composition works. Avoid simultaneously changing the camera angle, palette and subject, because you will lose the ability to tell why the next result looks different.

For a landscape concept, try: a small cabin beside a still mountain lake, early morning mist, one warm window, wide composition, natural colours, quiet editorial photography. For an illustration, try: an original fictional cyclist wearing an orange jacket, simple city shapes in the background, limited colour palette, flat graphic poster style. These are example prompts, not existing generated outputs or performance promises. Use them as starting points, then adapt the framing and subject to your actual project rather than copying blindly.

Understand credits before generating

New VYRA accounts receive 20 free credits. This is a limited signup allowance, not unlimited access to every image model. The exact quote depends on the model and your selected options. Read the price on the Generate button before submitting. If you are planning several variations, reserve some credits for revisions and compare the cost of a single draft with your available balance. Review the pricing page before choosing a plan or buying more credits; do not infer current prices from an old tutorial.

A failed generation returns its credits automatically. A successful generation that you do not like is a different situation, so think about the question each test should answer. Save a promising draft, write down what needs changing and make a targeted revision. This keeps experimentation intentional. When comparing two models, compare the quote as well as the images. A higher-cost job can be worthwhile for a particular brief, but price alone does not establish that the result will meet your requirements.

From still image to a finished creative

Inspect your chosen image for unintended details before downloading. Look at hands, reflections, product edges, background objects and every line of text. A small preview can conceal errors. Keep the prompt with the asset so the visual direction is understandable later. If you need a transparent background or an upscale, use the relevant Edit tools and inspect the processed file too. Each extra transformation can change details, so review the final deliverable rather than assuming the first approved thumbnail is still representative.

To animate a still, open Video Studio and select a model that accepts a start frame. Describe one action and a camera move that fit the image. A slow push-in on a product is a more controlled first experiment than a complete scene transformation. Image creation and video generation are separate jobs with separate quotes. Plan the budget for both stages. For motion transfer or supported video editing, explore Motion; for published examples and remix ideas, browse Explore and read the prompt alongside the result.