The previous local AI job ended before completion. Smart Fast v3.3 records the selected model, AI pass, disk checkpoint, and encoding stage without silently lowering scale.

Private AI Upscaler · Smart Fast v3.3

Smart Fast 2x, 4x, or 8x compact Real-ESRGAN output for practical iPhone speed, with optional Exact Full Resolution mode. Your photo is never uploaded by this app.

🔒 100% client-side processing

1. Choose an image

JPG, JPEG, PNG, WebP, HEIC, HEIF, AVIF, GIF, BMP, JPEG XL, TIFF, and other browser-supported photos.

🖼️

Drop your image here or select it from your device

No file selected

2. Upscale settings

v3.3 keeps the validated 4.87 MB compact Real-ESRGAN model and adds Smart Fast input planning so large iPhone photos finish in practical time.

Image category

Processing mode

Scale

2x 4x 8x

Quality level

Download format

Mobile engine policy

Processing engine Checking…

The worker is checking local WASM runtime and cache status.

Photo decoding, AI inference, disk streaming, and final encoding use separated stages to avoid overlapping memory peaks.

Engine diagnostics and exact errors

No engine errors yet.

Smart plan & iPhone safety Waiting for image

Select an image to see original size, planned AI input, final dimensions, tile count, storage, and expected speed reduction.

Default: Smart Fast 2x High JPG with Faster 2-thread WASM. Large photos are intelligently prepared before genuine Real-ESRGAN so output remains useful without creating unnecessary 100–200 MP files.

Preparing… 0%

Timing starts when Convert is tapped.

Screen wake lock inactive.

AI model cache status will appear after the app starts.

3. Before / After

Move the slider to inspect the AI-generated result.

↔

BEFORE AFTER