Restaurant table tent art: Ideogram 4.5 at 3:4, specials quoted
Make table tent artwork for a restaurant: a 3:4 Ideogram 4.5 image with each special quoted in the prompt, a price check, and the real cost per tent on Sume.

For a table tent, generate a portrait 3:4 image with Ideogram 4.5 on Sume, put every line of text you want printed inside quotation marks in the prompt, and proofread the output against your menu before it goes to the printer. A tent is read at arm's length over a meal, so large type and few words work better than a full menu.
The Sume Image API docs quote the fal list price for Ideogram 4.5 as $0.03, $0.06 or $0.22 per image by quality, the same for all sizes (read 2026-10-06, fal). On Sume the id is ideogram/ideogram-v4.5, quality defaults to medium, and resolution is 1K or 2K; the Image API docs list all of it. Prices on Sume are the list price times 1.25, so medium is $0.075 an image.
Plan the text before the image
Write the copy first, as short as it can be. Decide which lines are art and which are facts. A restaurant name and a dish name are fine to render; a price or an allergen note is the kind of line to add yourself in print software, because a wrong digit on a table is a real error. Quote each rendered line exactly as it should appear, including capitals and punctuation. If the restaurant uses a special character, such as an accented letter in a dish name, put it in the quoted text and check the result closely, because an image model can drop an accent.
A good prompt names the layout in words: the headline at the top, the dish name large in the middle, a small line under it, and a plain area at the bottom. Say what the paper and style are, for example warm cream paper with a hand-lettered feel. Ask for the same look across a series by keeping the style sentence identical from week to week and changing only the quoted dish line, which keeps the set coherent on the tables.
| Plan | Images | Billed (0.075 each) |
|---|---|---|
| One design, one draft | 1 | $0.075 |
| Four drafts to pick from | 4 | $0.30 |
| Weekly specials for 12 weeks, one image each | 12 | $0.90 |
A call that quotes every line
The request below asks for 3:4, quotes three lines and leaves the price out. Run it with quality set to medium first and move up only if the type is weak.
import os
import requests
prompt = (
'Portrait table tent for a bistro on cream paper. Headline "THIS WEEK" '
'at the top, large dish name "Roast Chicken Pie" in the middle, '
'small line "with seasonal greens" below. Plain empty band at the bottom.'
)
r = requests.post(
"https://api.sume.com/v1/images",
headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
json={"model": "ideogram/ideogram-v4.5", "prompt": prompt,
"aspect_ratio": "3:4", "quality": "medium", "resolution": "2K"},
timeout=60,
)
print(r.status_code)
print(r.json()["data"][0]["url"] if r.status_code == 200 else r.text[:300])How edits and references work
On Sume, Ideogram 4.5 without input_references generates from text. With references it edits the first image and uses up to four more as references, five in total, and an edit without aspect_ratio keeps the shape of the source image. That is why the weekly workflow below works: you keep last week's approved tent as the first reference, ask for the new dish line, and the layout stays put (Sume Image API docs).
The resolution field takes 1K or 2K, and quality takes low, medium or high, defaulting to medium if you leave it out. Since the list price depends on quality and not on size, there is no cost reason to pick 1K over 2K for a print piece; pick the size the printer needs.
Proof, then print
Compare each rendered line with your list character by character. If one word is wrong, fix it with an edit call that quotes the old and new text rather than regenerating the whole tent; the specials change weekly, so keep the approved image as the template and edit the dish line each week.
Print size is a separate question. A 2K render is enough for a small tent at normal print resolution, but check your printer's required pixel size and bleed, and read the catalog for the largest size the model lists. Ask your printer what pixel size and bleed they need and work backwards.
Prompt habits that help text
A few habits keep restaurant text clean, and none of them depend on a particular model version.
- Quote each line and name its role: headline, dish name, small line.
- Keep to three text elements; every extra line is another place for a typo.
- Ask for empty bands where prices and legal lines will go, then add them in your layout software.
- Say the words you do not want, such as no extra text and no logos, in positive phrasing: only the quoted lines appear.
- Run low or medium quality for drafts and keep high for a final only if you can see a difference.
Limits worth knowing
The model does not know your menu, your health rules or your brand fonts. Treat the output as artwork, not a finished file. If you need a QR code on the tent, add it afterwards; AI event ticket art with a working QR code shows the pattern.
Sources
Related posts
More in Use cases
- SaaS launch week asset budget: images, voiceovers and clips on Sume
A launch week of 12 Ideogram images, six 450-character voiceovers, six 5-second Grok clips and one music track costs about $1.61 on Sume. The per-line bill.
- Sale badge on a product image: draw it in Pillow, not in the prompt
Generate a clean product photo on Sume, then draw the sale badge and percentage from your catalog data in Pillow, so a price change never costs a regeneration.
- Search what was said across clips: SQLite FTS5 from STT segments
Request sentence segmentation on Sume STT, load each segment into a SQLite FTS5 table, and find the clip and second where a phrase was spoken.
- How many characters is a 15, 30, 45 or 60 second AI narration
Using 750 characters per minute of speech, a 15-second voiceover is 188 characters, 30 seconds is 375, 60 seconds is 750. Sume TTS cost for each.
Written by Sume