The image generator on this site turns a written prompt into a finished picture, and it can rework an image you already have. This guide covers the two modes, how to pick a model, how to write prompts that land on the first or second attempt, and what all of it costs in credits.
wavespeed.lol is an independent third-party experience for exploring WaveSpeedAI models. It is not operated by, endorsed by, or affiliated with the WaveSpeed AI team.
Two modes in one generator
Open the image generator and you get two tabs.
Text to image takes a written description and returns an image. You describe the subject, the setting, and the look you are after, and the model fills in the rest.
Image to image starts from a picture you upload. Instead of describing the whole scene from nothing, you describe the change: a variation of a shot, another angle on a product, a different treatment of existing artwork. The form accepts more than one reference image, which is useful when you want the model to borrow a subject from one frame and a style from another.
Both modes cost the same: 2 credits per image, drawn from the credits included in your plan. There is no separate charge for reference images.
Picking a model
The image side of the model library includes FLUX and Qwen Image, and the model selector lists everything the platform currently exposes for image work. Models behave differently on the same prompt - they differ in how literally they read instructions and in how much detail they add on their own - so the practical way to choose is a small comparison:
- Write one prompt that describes your subject, setting, and style.
- Run it on two different models.
- Look at the same detail in both results: how the subject's features came out, how legible any text is, whether the lighting matches what you asked for.
- Keep the model that needed fewer retries, and keep the prompt that produced it, because that pairing is what you will reuse.
Because every image costs 2 credits regardless of model, comparing two models costs 4 credits. On the Starter plan that is a small slice of the 600 credits included each month.
If you are still deciding where to start with the model list itself, the documentation lists the library and the pricing page shows what each plan includes.
Writing prompts that work
Prompts are plain description. The form does not ask for a special syntax, so writing the brief you would give a photographer is the right approach. Four things make the biggest difference.
Subject first. Lead with the thing you want to see: a person, a product, a place, a scene. Everything after it is a modifier.
Setting and context. Where is the subject, what is behind it, what time of day is it, what is the weather or the room like. Vague backgrounds produce vague images.
Style and treatment. Photo or illustration, film or digital, clean or textured, the mood you want. Camera and lighting words belong here: lens length, light direction, soft or hard shadows - the kind of notes that would go on a shot list.
Constraints, sparingly. If the image must not contain something, say so once. Long lists of negatives tend to fight each other, and the form warns you when a prompt gets too long, so tightening the sentence usually helps more than adding another clause.
When you get a result you like, change one word at a time. If you change the style and the framing and the lighting together, you cannot tell which change did the work.
Working from a reference image
Image to image rewards a different habit from text to image. Describe the delta, not the scene. If you upload a product photo and want it on marble under warmer light, the useful prompt is about the surface and the light - the product is already in the frame. Referencing the subject again tends to make the model re-invent it.
Keep the reference clean. A reference with heavy compression artefacts or a busy background will carry those qualities into the result, especially when you ask for a slight change rather than a transformation.
What an image costs, and what a plan covers
An image costs 2 credits in either mode. Plans renew each month:
| Plan | Price | Credits | Images per month |
|---|---|---|---|
| Starter | $19 / month | 600 | about 300 |
| Pro | $49 / month | 1,800 | about 900 |
| Studio | $99 / month | 4,000 | about 2,000 |
Annual billing costs less per month. Generating requires an active subscription - new accounts do not start with free credits - and the remaining balance is shown above the generator, so you always know what a session has used. The pricing and credits guide works through the numbers in more detail, and the video generator and music generator share the same account and the same balance if you want to mix formats.
FAQ
How much does one image cost? 2 credits per image, whether you start from a prompt or from a reference image.
Which image models can I choose from? The image models in the WaveSpeedAI library, including FLUX and Qwen Image.
Do I need a subscription to generate images? Yes. Plans start at $19 per month with 600 credits, which covers about 300 images.