Upscale an image
Planned as a local AI tool: enlarge an image with detail that plain interpolation cannot produce.
まだ利用できません
このツールはロードマップ上にあります。このページは処理の流れ・対応フォーマット・関連ツールを記録するために存在しており、実際の処理は行われず、結果も生成されません。
This page has not been translated yet and is shown in English. It is excluded from search indexing for this language until it has been.
Conventional resampling can make an image bigger but not sharper — it has no information to add, so it spreads the existing pixels more thinly. Super-resolution models predict plausible detail instead, which is why an AI upscale looks different in kind from a bicubic one.
Super-resolution models are small enough to be a realistic local-AI candidate, which is why this is planned as Tier B rather than a cloud call. It ships when on-device inference is dependable.
できること
2× and 4× targets
The standard super-resolution factors, chosen per image.
On-device inference planned
A small enough model to run in a browser with WebGPU or WebAssembly.
Tiled processing
Large images processed in tiles to keep memory use bounded.
使い方
- 1
Capability check
WebGPU or WebAssembly availability is confirmed before the feature is offered.
- 2
Tiled inference
The image is processed in overlapping tiles inside a Web Worker.
- 3
Reassemble
Tiles are recombined onto the canvas and the result enters the history.
仕様
- This tool is not implemented yet. Nothing is processed, no image is uploaded, and no result is produced. The page documents the intended workflow and is excluded from search indexing until the feature actually works.
- Planned as Tier B (local AI): ONNX Runtime Web with tiled inference.
KaiConvert を選ぶ理由
- ローカルツールはアップロード不要
- ページ読み込み後はオフラインでも動作
- クラウド AI は開始前に必ず明示