AI greeting card generator: a card front from your photo

To make a greeting card with AI, generate the card's front as artwork, from a prompt or from your own photo sent as a reference with a style ("a watercolor portrait of this dog in a party hat"), at the card's shape, and write the greeting yourself inside or in a layout tool. A folded 5 × 7-inch card's front is 5 × 7 inches, so on Sume, ChatGPT Image 2.5 at 1600 × 2240 pixels covers it at 320 pixels per inch.

Sume facts come from the Image API docs, whose own image-to-image example is "make this scene look like a watercolor painting", and the request schemas in the Sume API reference, read on 2026-09-28. Card sizes are plain arithmetic.

How do I turn my photo into a card front?

Put the photo at a public HTTPS URL and send it to POST /v1/images in input_references, with the style and the occasion in the prompt. Keep image_size at the front's size: it wins over aspect_ratio, so a landscape snapshot still becomes a portrait card.

• Name the style and the occasion: "a watercolor portrait of this dog in a party hat, balloons around him, pastel background".
• Birthday, holiday, and thank-you cards work the same way; only the prompt changes, such as "our family from the photo as a snowy watercolor scene".
• Say what must stay: "keep his white chest patch, his pose, and his ears".
• Ask for up to 4 versions with n, this model's ceiling today. Each completed image is billed.
• Use a model that edits from references, such as ChatGPT Image 2.5, which takes up to 16. A model whose input_references range is {"min": 0, "max": 0} is text-to-image only and rejects the photo.

curl -X POST "https://api.sume.com/v1/images" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-image-2.5",
    "prompt": "A watercolor portrait of this dog in a party hat, balloons around him, pastel background. Keep his white chest patch, his pose, and his ears. No text.",
    "input_references": [
      { "type": "image_url", "image_url": { "url": "https://example.com/dog.jpg" } }
    ],
    "image_size": { "width": 1600, "height": 2240 },
    "n": 4
  }'

What size should the card front be?

A card that folds along its side is printed on a sheet twice the width of its front, then folded in half, so the artwork only has to cover the front. Each front below is a legal ChatGPT Image 2.5 custom size at 320 pixels per inch; AI invitation generator explains why 320 works and lists flat card sizes.

Should the AI write the greeting?

Write it yourself. A short word on the front, such as "Thanks!", can be drawn if you proofread it, but a name or a longer message can come out misspelled, and the image has no text layer to fix. Type the inside message in a layout tool before printing, or write it by hand. AI image with text covers drawn lettering.

Will my dog or my family still look like themselves?

Not guaranteed. The model draws a new image from your photo, so faces, markings, and small details can change. Name the features that matter in the prompt, generate a few versions, and compare each with the photo before you print. If the card must show the real faces, place the original photo in the layout tool and use the generated art around it. Photo to painting AI covers more style prompts.

What does it cost, and what are the limits?

• Each completed image is billed, and a failed or canceled one is not. ChatGPT Image 2.5 estimates each request's price from its size and quality, and usage.cost in the response is the USD amount billed.
• High quality, ChatGPT Image 2.5's default, and a large n are two of the slow settings most likely to return 202 with a job instead of the images. Poll the job's status until it is terminal, then read its result.
• Today the front comes back as a PNG, JPEG, or WebP file at a Sume-hosted, signed URL, so download it before you lay out the card.