Smaltino
Explore
English 简体中文

Image

AI Image Editor

Edit or restyle an existing image with a prompt and reference images. Input requirements vary by model; Qwen local editing also needs a mask.

01Prompt

02Model
Loading models…
  • A concrete and glass house cantilevered over a misty sea
    Cliff house
  • Portrait of a young woman lit softly by a window
    Window light
  • Overhead view of colorful spice baskets in a market
    Spice market
  • A glowing jellyfish drifting in deep blue water
    Deep blue
  • An astronaut on the lunar surface with Earth on the horizon
    Earthrise
  • A snow leopard resting on a rock ledge in light snow
    Snow leopard
  • Koi fish in a dark pond with floating cherry blossom petals
    Koi and petals
  • A neon-lit Tokyo alley after rain with a figure under a clear umbrella
    Rain in the alley
  • A white horse galloping through shallow water on a beach
    Silver shoreline
  • Still life of pears and a ceramic jug on linen
    Three pears
  • A wooden cabin glowing on a snowy ridge at blue hour
    Cabin on the ridge
  • Sunbeams through a misty pine forest with a small figure on the path
    Morning pines
  • A humpback whale breaching at dawn with golden spray
    Breach at dawn
  • A lone traveler crossing pale sand dunes at sunset
    Dune walker

Why use it on Smaltino

  • 01

    Edit with words

    Describe the objects, colors or background to change. Use reference-image editing, or prepare a mask when the selected model requires one.

  • 02

    Restyle in one step

    Turn a photo into illustration, painting or product-shot styles in one run.

  • 03

    Keeps the structure

    The original composition stays recognizable, so edits look like the same picture.

  • 04

    Iterate quickly

    Results land in your creations, where you can redo or refine them.

Models you can choose

Pick a model in the generator above. Each one has its own page with parameters and pricing.

How it works

  1. Add an image and a prompt

    Upload a reference image and describe what should happen to it.

  2. Pick a model and settings

    Choose a model and adjust its settings. The credit cost is shown before you run.

  3. Generate and download

    Run it, then download the result or find it in your creations when it is ready.

Apps for your next project

Apps open in a new tab, keeping your current tool input. Review the prompt, model and settings before generating.

Workflows this tool can help with

  • Sellers exploring a product-photo edit

    Bring a product reference and describe the background or styling change. Compare the result with the original for accurate shape, labels and color before using it.

  • Designers planning a local retouch

    For Qwen 2.1 Inpaint, prepare a source image and a separate matching-size mask. Mark the area to edit, then inspect both edited and preserved areas in the output.

Frequently asked questions

How do I edit a photo with AI?

Choose a model, provide its required images and describe the change. Generate an edit, then compare it with the source; details outside the requested edit may also change.

Can I change the style of an image?

You can request a style such as watercolor, film photo or 3D render. How closely the result follows the request depends on the model and input; inspect it before use.

When do I need reference images or a mask?

The default Nano Banana Pro needs at least one reference image and a prompt; reference order can be changed. Qwen 2.1 Inpaint instead needs a separate source image and black-and-white mask with matching dimensions. White marks the area to edit and black the area to preserve. Its current output is one 1K PNG. Areas marked to preserve can still change, so inspect the result.

Which AI models are available?

You can choose between 2 models: Nano Banana Pro, Qwen 2.1 Inpaint.

How many credits does one run cost?

It depends on the model: from 8 to 12 credits per run. The default model, Nano Banana Pro, costs 12 credits. The exact cost is shown before you generate.

Am I charged if a generation fails?

Requests rejected before task admission are not charged. When a task finally fails, the system releases reserved credits against its original debit record. Check the task and credit records for the settled balance. A status-query error alone is not a final task failure; task credit releases are not payment or subscription cash refunds.

Where can I check progress and see the delivered result?

After submission, the page shows processing, completion, content blocking or failure. If the status lookup fails, retry the status query; that does not mean the task itself failed. A successful task with an asset link shows an image or an audio/video player and a download link below. You can switch between multiple delivered outputs when present. If completion has no preview, check My Creations. No fixed completion time or percentage is promised.

Do changing settings or viewing results start another generation?

No. Reading FAQs, changing inputs or settings, and viewing, playing or downloading existing outputs do not submit a generation. Uploading and checking a quote are preparation steps. Review the selected model, inputs and current credit cost, then click Generate to submit. Clicking Generate again is a new request with its own cost; retries of a status query only check the existing task.

Ready to create?

Pay only for what you generate. Failed or blocked runs return their credits.

Start creating See pricing