Skip to main content

Create Images from Prompts

Images is the prompt-led image generation area in Riverflow.

Written by Erika

Use Images when you want to start from a written idea rather than a Scene. It is useful for concept exploration, campaign visuals, reference sheets, backgrounds, props, product-adjacent imagery, or any image where the written brief is the main input. The Images section is particularly useful when you want to create an image that does NOT include your product.

Use Photoshoots when you want to generate a product-led image from a selected Scene. Use Editor when you need to make a precise edit to one existing asset. Use Batch when you need to apply the same edit or adaptation across many assets or SKUs.

Before you start

  • Select the correct Brand.

  • Decide what image you want to create.

  • Prepare any optional visual references you want to use.

  • Prepare any optional Style you want to apply.

  • Decide the aspect ratio and resolution you need.

Write your prompt

In Images, the prompt is the main brief. Use it to describe the image you want Riverflow to create.

A good prompt usually explains:

  • the main subject

  • the setting or background

  • the composition or camera angle

  • the lighting

  • the mood or visual treatment

  • any important colours, materials, or styling details

  • anything Riverflow should avoid

Good example:

Create a premium editorial image for a summer skincare campaign. Show a warm bathroom shelf scene with soft morning sunlight, pale stone tiles, clean reflections, and natural shadows. Leave space on the right side for ad text. The image should feel calm, bright, premium, and realistic.

Enhancing your prompt

For help with your prompt, use the Enhance Prompt button. This feature rewrites your prompt for free, enabling you to produce more accurate and high quality image outputs.

Example input prompt:

Crate a macro close up of some spinach, lime and ginger ingredients. Editorial feel

Enhanced prompt:

Create a realistic editorial macro food image featuring fresh spinach, lime, and ginger ingredients.

Subject:
Fresh spinach leaves, cut lime, and raw ginger root as the main visible ingredients. The spinach should feel leafy and textured, the lime should look juicy and fresh, and the ginger should show natural rough skin and fibrous cut surfaces.

Product accuracy:
Keep the ingredients botanically believable and true to life. Do not make the spinach, lime, or ginger look artificial, plastic, oversized, or unrecognizable.

Setting:
Arrange the ingredients on a simple natural surface such as matte stone, ceramic, or lightly textured neutral countertop. Keep the background minimal and softly blurred so the ingredients remain the focus.

Composition and framing:
Macro close-up framing with the ingredients filling most of the image. Use layered composition: spinach leaves in the foreground and midground, lime wedges or halves partially overlapping, and ginger root placed naturally among them. Keep the sharpest focus on the most appetizing textures: lime pulp, spinach veins, and ginger surface detail.

Camera:
Macro food photography with a full-frame camera feel, shallow depth of field, close focusing distance, realistic perspective, and crisp detail in the focal plane. Slight editorial crop, not overly staged.

Lighting:
Soft directional daylight from camera-left, gentle shadow falloff, natural highlights on the lime pulp, and subtle texture across the spinach and ginger. Keep exposure realistic with no harsh glare.

Style:
Editorial food photography, clean and natural, fresh ingredient story, premium but not overly polished.

People:
No people or hands visible.

Materials and realism:
Make the ingredients feel physically real: natural moisture on the lime, subtle surface imperfections, realistic spinach leaf texture, rough ginger skin, believable scale, soft contact shadows, and true-to-life color.

Avoid:
Avoid plastic-looking food, fake shine, oversaturated greens, impossible ingredient shapes, artificial glow, messy clutter, or overly smooth AI textures.

Add references or a Style

Use references and Styles when you want more control over the result.

  • Add visual references: use existing images, product images, uploads, folder assets, logos, or a new upload to guide the output.

  • Add a style: apply a saved Style to guide lighting, colour, mood, and visual treatment.

References are guidance, not exact copy instructions. If you want Riverflow to follow a specific part of a reference, say so in the prompt.

Example:

Use the reference image for the brushed metal texture and soft reflections. Do not copy the exact object, background, or layout.

You can add multiple reference images. The maximum depends on the selected model, and Riverflow will warn you if you add too many. Using more than four references can take longer to generate.

Choose model, resolution, aspect ratio, and output count

Before generating, review the controls at the bottom of the Images panel.

  • Model: choose the image model. The default model is usually the best starting point.

  • Resolution: choose the output resolution. Some models may not support every resolution.

  • Aspect ratio: choose the shape of the image.

  • Output count: choose how many images to generate. You can generate 1 to 4 outputs at once.

The credit cost is shown on the Generate button before you run the job.

Generate the image

  1. Open Images from the side navigation.

  2. Write your prompt in Describe your idea.

  3. Add visual references if needed.

  4. Add a Style if needed.

  5. Choose the model, resolution, aspect ratio, and output count.

  6. Select Generate.

The finished images will appear in the results grid.

Review the output

After generation, check whether the image matches the brief.

Look for:

  • the right subject and setting

  • the right composition and crop

  • the right lighting and mood

  • useful space for text or design if needed

  • correct use of references or Style

  • unwanted objects, text, people, logos, or layout details

  • whether the image is suitable for its intended channel

If the result is close, open the image and continue with edit, upscale, regenerate, remix, or save it into a Folder.

Remix an Image Result

Use Remix when an Images output is close and you want to create more variations from the same setup.

Remix loads the Images control panel with the ingredients that were used to generate that asset. This can include the original prompt, visual references, Style, model, resolution, aspect ratio, and output count where available.

  1. Open the image result you want to build from.

  2. Select Remix.

  3. Check the loaded Images control panel.

  4. Adjust the prompt, references, Style, model, resolution, aspect ratio, or output count.

  5. Generate again.

Remix creates a new generation from the original setup. It does not directly edit the existing image. Use Review and Edit Images when you need a precise edit to the current asset.

When to use Images instead of Photoshoots

Use Images when the written idea is the main input and you do not want to start from a Scene.

Good uses for Images include:

  • creating campaign concepts from a written brief

  • creating reference sheets

  • creating background or prop ideas

  • creating visual directions before saving a Scene or Style

  • generating assets that are not tied to one specific product SKU

  • testing composition, lighting, or mood quickly

Use Photoshoots when you want Riverflow to adapt a selected Scene around a product.

If you need product accuracy

Images can use product images as visual references, but it is not the most controlled way to preserve a commercial SKU.

If the product must be accurate, start with Photoshoots and select the product and Scene. If you need to make a precise change to an existing image, use Editor. If you need to make the same precise change across many assets, use Batch.

If the result does not follow your prompt

Try one or more of these fixes:

  • make the prompt more specific

  • say which part of each reference matters

  • remove references that are adding conflicting information

  • add a Style if the issue is lighting, colour, mood, or overall treatment

  • generate more than one output so you can compare options

  • simplify the request if the image is trying to do too many things at once

What to do next

Use approved Images outputs as standalone assets, as references for future generations, or as starting points for editing. Save useful results into Folders so they are easy to find later.

Did this answer your question?