topazlabs/image-upscale - AI Image Tools Tool

Overview

topazlabs/image-upscale is a professionally oriented AI image upscaler hosted on Replicate that exposes Topaz Labs’ upscaling models as an API and playground for programmatic and interactive use. The model family supports multiple enhancement modes — Standard V2, Low Resolution V2, CGI, High Fidelity V2 and Text Refine — and can scale images up to 6× while offering dedicated facial-enhancement controls for portraits and blurry faces. The Replicate-hosted model is intended for use cases that need high-quality enlargements (photography, digital art, scanned text/images) without local GPU setup, and it returns image outputs sized by megapixel tiers with pay-per-output pricing. ([replicate.com](https://replicate.com/topazlabs/image-upscale/readme)) In parallel, Topaz distributes similar technology in its Gigapixel / Topaz Studio products for desktop and cloud workflows; those products expose many of the same enhancement modes and are offered through Topaz’s subscription and app-bundle plans. Community feedback generally praises Topaz’s detail preservation and face recovery but raises concerns about processing time on lower-end hardware, occasional artifacts, and the company’s move toward subscription bundles — all important considerations when selecting a deployment path. ([topazlabs.com](https://www.topazlabs.com/topaz-gigapixel?utm_source=openai))

Key Features

• Multiple enhancement models: Standard, Low Resolution, CGI, High Fidelity, Text Refine.
• Upscales images up to 6× while targeting quality preservation for photos and art.
• Adjustable facial-enhancement parameters for improved portrait restoration and face clarity.
• API-hosted on Replicate with pay-per-output (megapixel-tier) pricing for cloud inference.
• Optimized Text Refine model for sharp, legible results on text-heavy or scanned images.

Example Usage

Example (python):

# Example: run topazlabs/image-upscale from Replicate using the Python client.
# Install: pip install replicate
# Set your REPLICATE_API_TOKEN in environment variables before running.

import os
import replicate
import requests

# Authenticate with Replicate token (set REPLICATE_API_TOKEN env var)
client = replicate.Client(api_token=os.environ.get("REPLICATE_API_TOKEN"))

# Select the Topaz model
model = client.models.get("topazlabs/image-upscale")

# Local input image path
input_path = "input.jpg"

# Run prediction. Parameter names may vary per model — check the model README on Replicate.
# Common inputs: image (file or URL), scale (2/4/6), enhancement_model (one of the available models), face_enhance (bool).
prediction = model.predict(
    image=open(input_path, "rb"),
    scale=4,
    enhancement_model="Standard V2",
    face_enhance=True
)

# The API typically returns URL(s) to the processed image(s)
print("Prediction output:", prediction)

# If the model returns an image URL, download it:
if isinstance(prediction, list) and prediction:
    out_url = prediction[0]
    r = requests.get(out_url, stream=True)
    if r.status_code == 200:
        with open("upscaled.jpg", "wb") as f:
            for chunk in r.iter_content(1024):
                f.write(chunk)
        print("Saved upscaled image to upscaled.jpg")
    else:
        print("Failed to download result, status:", r.status_code)

# Notes:
# - The exact input field names and return format vary by model version; consult the model README on Replicate.
# - This example follows common Replicate client patterns. See Replicate Python client docs for advanced usage. 
# References: Replicate model README and Python client docs. ([replicate.com](https://replicate.com/topazlabs/image-upscale/readme))

Pricing

Replicate-hosted model: pay-per-output pricing by output megapixel tiers (example: $0.05 for 12–24MP outputs; see per-tier table on the model page). Topaz’s direct products (Gigapixel/Topaz Studio) are sold under Topaz’s subscription and bundle plans (sample Gigapixel personal plan listed at $17/month on Topaz’s pricing page). Topaz has also communicated changes to licensing and bundling in 2025; check Topaz’s pricing page and official announcements for up-to-date commercial terms. ([replicate.com](https://replicate.com/topazlabs/image-upscale/readme))

Benchmarks

Replicate usage (model runs): 1.2M runs (Source: https://replicate.com/topazlabs/image-upscale)

Replicate per-output pricing (example): $0.05 for 12–24 MP outputs (per run tier) (Source: https://replicate.com/topazlabs/image-upscale)

Topaz Gigapixel (personal subscription sample): $17 / month (personal, annual commitment listed on Topaz site) (Source: https://www.topazlabs.com/pricing)

Customer rating (Trustpilot): 3.9 / 5 (55,347 reviews, aggregated) (Source: https://www.trustpilot.com/review/www.topazlabs.com)

Key Information

• Category: Image Tools
• Type: AI Image Tools Tool