AI poster generator: print sizes and room for your title

Generate AI poster art at your poster's shape with room for the title, upscale it for a 300 DPI print, then add the words in a layout tool.

5 min readSume
All posts

To make a poster with an AI poster generator, generate the artwork at your poster's shape with calm space left for the title, upscale it to the pixel count your print size needs, then add the title, date, and venue yourself in a layout tool. On Sume, ChatGPT Image 2.5 takes custom sizes up to 2336 × 3504 pixels in the 2:3 shape of a 24 × 36-inch poster, and Image Upscale 1.0 enlarges an image by any factor from 1 to 4.

Sume facts come from the Image API docs and the request schemas in the Sume API reference, read on 2026-09-28. Poster sizes and pixel counts are plain arithmetic, the same math AI image for print walks through for any print.

What size should an AI poster be?

Choose the print size first and generate at its exact shape, so nothing gets cropped. For the pixel count, multiply each edge in inches by the DPI your print shop asks for: at 300 DPI, a 24 × 36-inch poster needs 7,200 × 10,800 pixels. That is far above the 8,294,400-pixel limit on a ChatGPT Image 2.5 custom size, so generate the largest size in the right shape, then upscale it by the factor in the last column.

The factors are rounded up to two decimals so the print is not a few pixels short.

Largest sizes computed from the ChatGPT Image 2.5 custom-size rules in Image API; factors use the 1–4 upscale_factor range in the Sume API reference. Read 2026-09-28.
PosterShapeGenerate atPixels at 300 DPIUpscale by
11 × 17 in11:172288 × 35363,300 × 5,1001.45×
18 × 24 in3:42448 × 32645,400 × 7,2002.21×
24 × 36 in2:32336 × 35047,200 × 10,8003.09×

How do I generate the poster art?

Send POST /v1/images with openai/gpt-image-2.5 and the size as image_size. Write the prompt as a brief for the picture, not for the whole poster:

  • Subject, mood, and style: a jazz trio under string lights, warm dusk colors, screen-print texture.
  • Where the words go: "leave the top third as plain night sky for a title and a quiet band along the bottom for the details". For wall art with no words, skip this line.
  • "No text", so the model does not paint in a title you would have to cover up.
  • n for up to 4 versions per call, this model's ceiling today. Each completed image is billed.
  • High quality (this model'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.
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": "Poster art for a summer jazz night: a trio on a small outdoor stage under string lights, warm dusk colors, screen-print texture. Leave the top third as plain night sky for a title and a quiet band along the bottom. No text.",
    "image_size": { "width": 2336, "height": 3504 },
    "n": 4
  }'

Should the AI write the title on the poster?

Only if you proofread every letter: generated lettering can come out misspelled, and today the file comes back as a flat PNG, JPEG, or WebP with no text layer to fix, as AI image with text explains. For an event poster, set the title, date, and venue in a design tool on top of the upscaled art, so you can check each detail and reuse the art for the next date.

Can I make a poster from my own photos?

Yes. Put each photo at a public HTTPS URL and list it in input_references; ChatGPT Image 2.5 takes up to 16. Say what each photo shows and how to use it, such as "the band photo as the main image, redrawn as a two-color screen print". Keep image_size in the request: it wins over aspect_ratio, so the result stays poster-shaped even when the photos are landscape. The model redraws the photos, so check faces and small details.

How do I make the file big enough to print?

Upscale the art before you add the words: send its URL from the generation result as image_url to POST /v1/image-upscale-1.0/upscale, with the upscale_factor from the table. At 3.09, a 2336 × 3504 image should come back just over 7,200 × 10,800. The docs don't say how a decimal factor is rounded, so check the pixel size of the file you get back before you order the print. AI image for print covers the rest of the upscale request and the job it returns.

  • API pricing lists Image upscale at $0.20 per image, plus a 5.5% agent fee by default. ChatGPT Image 2.5 estimates each generation's price from its size and quality, usage.cost in the response is the USD amount billed, and a failed generation is not billed.
  • /v1/images results are Sume-hosted, signed URLs, so download the art you keep.
  • Bleed, trim, and color settings are your print shop's specs. Set them when you export from your layout tool.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume