Image workflow
Use a specific scene description, then decide whether your goal is a new image or an edit that must preserve a reference. Wavespeed helps you frame that choice; check the available model and controls at the destination before generating.
The useful distinction is not a style label: A starts from text alone, while B starts from an image whose details may need to survive.
A: Describe a new campaign scene without supplying a reference. Specify subject, lighting, framing, and unwanted elements.
A seedream brief can explore a direction, but it cannot preserve a particular photographed object's identity without source material.
wavespeed qwen image editB: Start with an approved product photo and request a different background while keeping the packaging legible.
The v5 edit brief needs an explicit list of details that must remain unchanged; inspect the result rather than assuming preservation.
wavespeed ai modelsB: Improve presentation of an existing image without inventing missing product details.
Separate editing from enlargement: an upscaled image may look sharper without becoming a faithful reconstruction.
wavespeed upscaleA and B: Test a fresh concept, then compare it with an edit based on approved campaign art.
Review both outputs against the same brief before deciding which workflow better serves the assignment.
wavespeed ai reviewTurning a reference into a text description is convenient, but a prompt is not a complete record of the original image.
Before writing a v5 prompt, note the reference's crop, product markings, object count, and colors. Save the original so you can compare it with any later output.
Name the one change you want, then state what must stay fixed. A general request such as “make it cinematic” may leave important brand or identity details open to interpretation.
A text-only reconstruction can lose small lettering, exact geometry, or the relationship between objects. If those details matter, use a reference-capable workflow when available and inspect the result closely.
Bring a prepared brief to the destination, then confirm which image models and reference controls are actually offered there. Wavespeed does not establish that a particular v5 variant is available.
Compare the output against the source and the written brief, not just against your memory of the intended look.
Reference check
Output check
This page cannot verify that the destination offers Seedream or a specific v5 release.
Workaround
Read the model name and version displayed at the destination before submitting a job.
Even a detailed edit request may change lettering, faces, proportions, or product features.
Workaround
Keep the reference and compare every detail that matters to your use case.
The paired images here do not establish output quality, edit accuracy, or performance for your prompt.
Workaround
Run a small test with your own permitted source image and evaluate the actual result.
Write the subject, desired change, and details to preserve before opening the generation destination. Confirm its current model choices and terms there, then judge the resulting image against your original brief.
Explore image toolsThe name is used in searches for a Seedream image-model variant. A search term alone does not confirm the variant's release status, specifications, or availability at a particular service. Check the provider's current model listing for an exact version name.
A prompt-only image and an edit based on a reference are different workflows. Whether both are supported depends on the model and controls available where you generate. Check for a reference upload or editing option rather than assuming a text prompt will preserve an existing image.
Do not assume it will. Small text and distinctive marks can change in generated images, even when the prompt asks to retain them. Compare the output with the approved source before publishing it.
Look for the model identifier in the service's job details or saved generation record. If that information is unavailable, the image itself is not reliable proof of a particular version. Save the prompt and model selection alongside any output you need to document.