✣ RUNWARE
Explore models
English

Before you choose

Choosing runware models for an image project

Start with the result you need, not a model name. Write down whether you are creating an image from text, changing an existing image or working toward a video. Keep a sample prompt or source image ready, along with any requirements you can verify in the tool you intend to use. Runware is the subject of this guide; the button below opens a separate partner site, so check its available models and controls there rather than assuming the two services offer the same choices.

Explore image generation
Runware landing-page visual

Selection workflow

One full run-through

  1. 1

    Define the job

    Suppose you need a clean product concept image of a desk lamp. Specify the subject, background, lighting and intended use before looking through model names. If you already have a photograph to alter, mark that as an input requirement: a text-only workflow will not substitute for a model that accepts images. This first distinction narrows the choices more reliably than a broad search for the newest option.

  2. 2

    Check the available controls

    Open the model information in the service you are using and check supported inputs, output type and any exposed settings. For the lamp example, choose an image-generation option that accepts a text prompt. Treat resolution, aspect ratio and reference-image support as questions to verify, not features to assume from a model name. Runware model availability and a partner site's selection may differ.

  3. 3

    Make a controlled comparison

    Test one prompt, then change only the model if the interface permits it. Compare whether the lamp shape stays coherent, the requested lighting appears and the background suits your use. Record the prompt and settings so you can identify what changed. If the result misses the brief, revise a specific instruction before switching again; otherwise, prompt changes and model changes become difficult to separate.

Options table

Use these task categories as a comparison checklist. They describe what to look for in a model listing, not a verified catalog of currently available options.

First Runware landing-page feature illustration
01

Start from text

Choose this path when the starting point is a written description and the desired result is a new still image. Compare how well candidates follow your subject, composition and style instructions. For a fair test, keep the prompt constant and inspect details that matter to your project rather than judging only the overall impression. Check the selected interface for its actual output and setting options.

  • Input to check: text prompt
  • Output to check: still image
  • Best test: repeat one brief across candidates
Second Runware landing-page feature illustration
02

Work from an existing image

Choose this path when preserving part of a source image matters more than inventing a scene from scratch. Look for explicit support for an image input and the kind of change you need. A request to alter a background, for example, differs from a request for an unrelated new picture. Check what the interface lets you provide before drafting a workflow around a reference image.

  • Input to check: source or reference image
  • Output to check: edited image
  • Best test: inspect what was preserved
Third Runware landing-page feature illustration
03

Plan for motion

Choose this path only when your deliverable needs movement rather than a single frame. Verify that the available model and interface support video output, then check their stated input requirements. A strong still image does not establish that an image model can produce motion. If you are comparing results, evaluate movement and scene consistency separately from the appearance of an individual frame.

  • Input to check: the required prompt or source
  • Output to check: video rather than a still
  • Best test: review motion and consistency

What fails

Avoid a model-task mismatch

A detailed prompt cannot make a text-only input accept a source image, and a still-image result cannot satisfy a motion brief. Vague comparisons fail too: if you change the prompt, settings and model together, you cannot tell which change helped. Define the deliverable, verify the available inputs and outputs, and keep your test consistent. The partner destination may present different choices from Runware, so confirm its current capabilities before committing to a workflow.

Explore model options
  • Check inputs before writing the full brief
  • Compare like-for-like results
  • Verify options at the destination

FAQ

The phrase refers to the model choices someone may encounter while exploring AI generation through Runware. The useful distinction is what each available option takes as input and produces as output. Consult the current model listing for specific names and supported tasks.

Begin with your deliverable: a new still image, a change to an existing image or a motion result. Then verify that a candidate supports your required input and output. Test the same brief across eligible choices instead of comparing unrelated prompts.

No such capability should be assumed from the word model alone. Check whether the particular option and interface accept an image input and support the change you want. If they only accept text, choose a different supported workflow.

Do not assume it is. The call to action here opens a separate partner site, whose available models, controls and requirements may differ from Runware. Check its current options before following a Runware-specific workflow there.

Explore models
Explore models