AI Headshot Generation API — Professional Profile Photos Programmatically

An HR onboarding tool that needs a consistent, professional-looking headshot for every new hire's directory profile can't rely on employees submitting a studio photo — remote hires rarely have one, and asking them to book a photographer adds friction to onboarding. Generating a polished headshot from a casual selfie via API removes that dependency entirely.

Direct answer: a reference-guided image generation API can take one casual photo of a person and output a professional-style headshot — consistent lighting, neutral background, business attire — without a photo studio, by passing the source image alongside a style prompt.

How Does an API Turn a Selfie Into a Professional Headshot?

Reference-guided generation takes an uploaded photo as an image input and a text prompt describing the target style, then produces a new image that preserves the subject's likeness while applying the requested lighting, background, and framing changes. This differs from pure text-to-image, which has no source photo to anchor facial identity — headshot generation specifically needs a model that supports image-plus-prompt input, not text-only generation.

The output is a new image, not an edited version of the original file — the model recomposes the scene around the reference face rather than applying filters on top of the existing background.

Which Model Fits Headshot Generation Best?

Nano Banana Pro's reference-guided mode is the stronger fit for headshots because it's built to hold facial identity consistent while changing background, lighting, and attire around it; GPT-image-2 works for simpler instruction-based touch-ups on an existing headshot rather than a full studio-style regeneration.

Model	Mode	Resolution / price	Facial consistency	Best for
Nano Banana Pro	Reference-guided	1K $0.030 / 2K $0.030 / 4K $0.042	High	Selfie-to-professional-headshot generation
Nano Banana 2	Reference-guided	1K $0.020 / 4K $0.036	Moderate	Lower-cost draft headshots at volume
GPT-image-2	Instruction editing	$0.014/image	N/A (edits existing photo)	Background swap or lighting touch-up on an existing headshot

For a directory of hundreds of employee photos where identity accuracy matters more than unit cost, Nano Banana Pro is the safer default. For a quick background cleanup on a photo that's already headshot-quality, GPT-image-2's edit mode is cheaper per call.

How Do I Generate a Headshot From a Reference Photo?

Submit the reference image URL along with a prompt describing the target studio style — lighting, background, and attire — and the model returns a new image built around the subject's face.

import requests

API_KEY = "YOUR_GENRELAY_KEY"
BASE_URL = "https://genrelay.ai/v1"

def generate_headshot(reference_image_url, style_prompt, resolution="2K"):
    r = requests.post(
        f"{BASE_URL}/images/generations",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={
            "model": "nano-banana-pro",
            "mode": "reference-guided",
            "reference_image_url": reference_image_url,
            "prompt": style_prompt,
            "resolution": resolution,
        },
        timeout=60,
    )
    return r.json()["output_url"]

headshot_url = generate_headshot(
    "https://cdn.example.com/hr/new-hire-selfie.jpg",
    "Professional corporate headshot, soft studio lighting, neutral gray background, "
    "business casual attire, direct eye contact, shoulders-up framing",
)
print(headshot_url)

How Do I Generate Headshots for an Entire Team in One Batch?

Loop the same reference-guided call over a list of employee photos with a shared style prompt so every headshot in the directory matches the same lighting and background treatment.

from concurrent.futures import ThreadPoolExecutor

STYLE_PROMPT = (
    "Professional corporate headshot, soft studio lighting, neutral gray background, "
    "business casual attire, shoulders-up framing"
)

NEW_HIRES = [
    ("emp-1042", "https://cdn.example.com/hr/emp-1042-selfie.jpg"),
    ("emp-1043", "https://cdn.example.com/hr/emp-1043-selfie.jpg"),
    ("emp-1044", "https://cdn.example.com/hr/emp-1044-selfie.jpg"),
]

def process(entry):
    emp_id, photo_url = entry
    return emp_id, generate_headshot(photo_url, STYLE_PROMPT)

with ThreadPoolExecutor(max_workers=4) as pool:
    results = list(pool.map(process, NEW_HIRES))

for emp_id, url in results:
    print(f"{emp_id}: {url}")

See the guide to consistent AI-generated images for prompt patterns that keep style uniform across a larger batch.

What Does Generating Headshots at Scale Cost?

As of September 2026, GenRelay pricing: Nano Banana Pro is $0.030 per image at 2K resolution, and Nano Banana 2 is $0.020 per image at 1K.

A directory refresh for 200 employees at 2K via Nano Banana Pro costs 200 × $0.030 = $6.00. Running a lower-cost draft pass first with Nano Banana 2 at 1K — to let each employee pick a preferred variant before the final render — adds 200 × $0.020 = $4.00 for a first round, bringing the two-pass total to $10.00 for 200 headshots.

Approach	Volume	Cost
Direct 2K generation (Nano Banana Pro)	200 employees	$6.00
Draft pass (Nano Banana 2, 1K) + final (Nano Banana Pro, 2K)	200 employees	$10.00
Single re-generation for 20 unsatisfied results	20 employees	$0.60

At that cost, generating headshots for an entire company directory remains cheaper than a single half-day photography session for a mid-sized team.

Internal Links

• Nano Banana Pro on GenRelay
• AI avatar generation API
• How to maintain visual consistency in AI-generated images via API

FAQ

Does the API guarantee the output looks exactly like the reference photo?
No. Reference-guided generation preserves the subject's general likeness closely but is not a pixel-level edit — expect close resemblance, not an identical face pasted into a new scene, and budget for a manual review step before publishing to a public directory.

Can it remove glasses or change hairstyle from the reference photo?
Yes, if requested explicitly in the prompt, though results vary — test a small batch before committing to a full-team rollout with any single instruction.

What image formats and sizes does the reference photo need to be?
Standard JPEG or PNG at typical selfie resolution works; the API does not require pre-cropping or specific aspect ratios before submission.

Is there a free tier to test headshot quality before rolling out to a team?
Yes. GenRelay includes free credits on signup, enough to generate and compare several sample headshots before running a full-team batch.

How does this differ from a generic background removal tool?
Background removal tools cut out the existing background from a photo; this API regenerates the entire scene — lighting, background, and often attire — around the subject's face using a text prompt.

---

As of September 2026. Pricing subject to change — verify current rates at genrelay.ai.