From a prompt or a photo to a new image.
Send a text prompt — or an existing image plus an instruction — and get back a generated image. Identity, subscription status and usage limits are checked on the way; your app only shows the result.
A watercolor of a lighthouse at dawn, soft light, muted colors.
Repaint this photo in a watercolor style.
Art and wallpapers
Prompt in, image out — the pattern behind Pixy. Styles and sizes are parameters, not separate services.
Photo restyling
Take the user’s photo and apply a style or an edit instruction. Same endpoint, with an image attached.
Avatars and product shots
A reference image plus a prompt for consistent variations — profile pictures, mockups, thumbnails.
One endpoint. Two inputs.
- 1Send a prompt — optionally with an image.Text only for generation; add an image file and the same call becomes a transformation. No provider key in the app.
- 2Backapp checks the user and runs the job.Identity, subscription status and usage limit are verified first; declined requests never reach the provider.
- 3Read the image URL from the job.Poll the job until
completed, then load the image into your view.
// CONTENT PENDING: replace with the verified example
const form = new FormData();
form.append('prompt', 'A watercolor of a lighthouse at dawn, soft light');
// image-to-image: attach the source photo — same endpoint
// form.append('image', { uri, name: 'photo.jpg', type: 'image/jpeg' });
const res = await fetch(`${BACKAPP_API_URL}/ai/images`, {
method: 'POST',
headers: { Authorization: `Bearer ${userToken}` },
body: form,
});
const job = await res.json();
const result = await waitForJob(job.id);
setImageUrl(result.image_url);{
"id": "job_01HZY…",
"status": "completed",
"workflow": "image_generation",
"result": {
"image_url": "https://…/out.png",
"width": 1024,
"height": 768,
"mode": "text_to_image",
"prompt": "A watercolor of a lighthouse at dawn, soft light"
}
}What comes back.
A job with the image and the parameters it was made with, so you can show, save or regenerate.
image_url- Where to load the generated image from.
width / height- Output size in pixels.
mode- text_to_image or image_to_image.
prompt- The prompt or instruction the image was made with.
status- queued, processing, completed or failed.
Text-to-image and image-to-image
Generate from a prompt, or transform an existing image with an instruction. Supported sizes, formats and the model list are in the docs.
Multiple outputs, upscaling and inpainting
Several candidates per request, resolution upscaling and masked edits are being scoped and will be listed here when confirmed.
The generation behind our own image app.
This workflow is the one Pixy runs on. Same API, same limits, same subscription check.
Questions about image generation?
Are text-to-image and image-to-image separate services?
No. One endpoint handles both: send a prompt alone to generate, or attach an image to transform it. They share limits and pricing.
Which sizes and formats are supported?
Common aspect ratios and PNG/JPEG output. The exact size list and the model options are in the docs.
How long does a generation take?
Generation runs as a job, typically a few seconds. Upload, poll until completed, then load the image URL.
Does a declined request count against the limit?
No. Identity, subscription and limit checks happen before the job is created. Only accepted jobs count.
Who pays for the AI provider?
Either you, with your own provider key in your Backapp project, or Backapp, billed on usage. Both options are described on the main page.
Add image generation to your app.
Identity, subscriptions and AI workflows from one backend.