AI architecture concept render from a sketch via API on Sume

To get a concept render from an architectural sketch or a photo of a physical model, use a reference edit on the Sume image API: send the drawing in input_references and describe the materials, light and camera. Sixteen catalog rows take references, read 2026-10-03, and 16:9 is listed by every row. ChatGPT Image 2.5 takes up to 16 references, so you can add material samples and a site photo next to the sketch.

Treat the output as a visual concept. It does not carry dimensions, structure or code compliance, and a model can invent windows or move walls. Label renders accordingly when you share them.

What to send

• Reference 1: the sketch or massing photo (public HTTPS URL).
• Optional references: a site photo, a material swatch, a mood reference. Ten references on most rows, sixteen on ChatGPT Image 2.5, five on Ideogram 4.5.
• Prompt: what must stay (the silhouette, window rhythm, roof line) and what changes (materials, time of day, landscaping).

Rows for the job

A request

Ask for several angles at once with n: 4 and keep the preserve list in every prompt. A call that is still running after the 30-second wait returns 202 with a job envelope instead of 200, so check the status code before reading data; see Jobs and results.

import os
import requests

resp = requests.post(
    "https://api.sume.com/v1/images",
    headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
    json={
        "model": "openai/gpt-image-2.5",
        "prompt": "photoreal concept render of the building in the sketch, keep the "
                  "silhouette and window rhythm, timber cladding, late afternoon light",
        "input_references": [
            {"type": "image_url", "image_url": {"url": "https://example.com/sketch.png"}}
        ],
        "aspect_ratio": "16:9",
        "quality": "medium",
    },
    timeout=60,
)
resp.raise_for_status()
print(resp.status_code, resp.json()["data"])

Cost

Flux 2 Pro at $0.0375 makes ten angles for $0.375. ChatGPT Image 2.5 lists $0.065875 at high quality and 1024×1024; the reserve rises with quality, size and reference count, so quote a real call before budgeting a large batch.