ShopixHelp Center

Editing images in Canvas

Put an image on the board, select it, and say how to change it: new background, new text, a new scene.

10 steps · 6 minCost: Brain usage and images are billed separately. Text-only chat can incur brain charges; image generation and editing cost extra by model, quality and resolution.
Open Canvas

What you bring

The result

One sentence: “A lifestyle shot for Amazon: bright home, a cat drinking from it, headline ‘Keep your pets drinking’”.

Steps

  1. Step 1 of 10

    How the Canvas brain charges

    The brain is the AI assistant on the right: it understands requests, reasons and calls tools. Low / Medium / High controls brain effort, not image quality. Brain input and output usage is billed even for text-only chat; images cost extra. Equipping inspiration, skills or teams is free.

  2. Step 2 of 10

    Low: off-peak and peak rates

    At October 2026 rates, Low costs 0.35 credits per 1,000 input tokens and 1.4 per 1,000 output tokens off-peak. Peak rates are 0.7 and 2.8 respectively. Peak hours are 09:00–12:00 and 14:00–18:00 daily, Beijing / Hong Kong time (UTC+8); other hours use off-peak rates.

  3. Step 3 of 10

    Medium and High: same rates, different usage

    At October 2026 rates, Medium and High both cost 1.75 credits per 1,000 input tokens and 8.75 per 1,000 output tokens. High may use more reasoning and conversation history, increasing the total. These are usage rates, not fixed prices per message.

  4. Step 4 of 10

    How and when charges are calculated

    Tokens measure model usage, not character count. Input includes system instructions, conversation history and tool results, not just your message. Output usage comes from the model response. After a response reports usage, the cost is input tokens ÷ 1,000 × input rate plus output tokens ÷ 1,000 × output rate. One request may involve several calls. Its cumulative cost is rounded up to whole credits, then credits already charged for that request are subtracted to avoid charging twice.

  5. Step 5 of 10

    Example: 6 brain credits plus an image

    A Low off-peak call with 12,775 input tokens and 604 output tokens costs 12.775 × 0.35 + 0.604 × 1.4 = 5.31685 credits, rounded up to 6. If that is the only brain call in the request and it generates one Sunburst Light 1K image (50 credits), the total is 6 + 50 credits. This is an example, not a fixed quote: additional calls, longer content or different image settings change the total. See the price table in Models and quality for other image tiers.

  6. Step 6 of 10

    Know the board

    The left is the board with your images and every result; the right is the assistant — type requests in the box at the bottom. The bottom toolbar uploads, adds text and marks. Not sure where to start? Tap a direction or a sample request in the assistant.

  7. Step 7 of 10

    Put the product on the board

    Press “Upload” in the bottom toolbar, or drag an image onto the board. A new image is selected automatically and an image tag appears in the input, meaning the assistant will use it.

  8. Step 8 of 10

    Say everything in one go

    Say which image, what to make, the scene and any text on it, then press “Send”. Rather talk than type? Press the mic next to “Send”.

    TipExample: use this fountain for an Amazon lifestyle image, bright clean home, a cat drinking from it, headline “Keep your pets drinking”.

  9. Step 9 of 10

    Keep editing the result

    The new image lands on the board, selected. Say “change the background to a light oak table, keep everything else” and only that changes.

  10. Step 10 of 10

    Tools and download

    Select an image and the toolbar above it offers HD upscale, remove background, text edits, expand, angles and more; “Download” saves it and “Platforms” exports marketplace sizes.

Image models: credits per image

Find the per-image price by model, quality tier and resolution. For multiple images with the same settings, multiply by the image count. These prices exclude the brain usage described above.

Model and tierSD 1KHD 2KUHD 4K
Ultimate 2.5 Sunburst3006301020
Superfine 2.5 Sunburst155305480
High Quality 2.5 Sunburst105190290
Standard 2.5 Sunburst6080105
Light 2.5 Sunburst506070
Ultimate 2.5 Flare3006301020
Superfine 2.5 Flare155305480
High Quality 2.5 Flare105190290
Standard 2.5 Flare6080105
Light 2.5 Flare506070
Exceptional GPT Image 2 Ultra3006301020
Ultimate GPT Image Pro120180360
Versatile GPT Image 25075150
High Quality Nano Banana 2.180100140
Standard Nano Banana 2.16590125
Light Nano Banana 2.15580110
High Quality Nano Banana Pro120180360
Ultra-fast Nano Banana Lite100150300
Balanced Nano Banana 25075150

Credits per image.

  • Unit: credits per image. A dash means the resolution is not publicly supported for that model.
  • For example, Sunburst Light 1K costs 50 credits per image, or 150 for 3 images. Standard 2K costs 80 per image, or 240 for 3.
  • Sunburst and Flare have identical prices at matching tiers. Only public models are listed; available options depend on the current Canvas.

Not happy with the result?

The assistant didn’t use my image
Make sure an image tag is in the input before sending; if not, click the image on the board first.
I asked for one change and everything changed
Say “only change X, keep everything else”; or use “Scribble” in the toolbar to circle the area.
I want to spend less
Set “Low / Medium / High” (left of “Send”) to “Low” and keep the default Light tier.

Related questions

Did this lesson help?