Ir para o conteúdo
KaiConvert
IA localPlanejado

Upscale an image

Planned as a local AI tool: enlarge an image with detail that plain interpolation cannot produce.

Ainda não disponível

Esta ferramenta está no roteiro. A página existe para documentar o fluxo, os formatos e as ferramentas relacionadas — mas nada é processado e nenhum resultado é produzido.

IA local
Ver as ferramentas que já funcionam

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.

O que você ganha

  • 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.

Como funciona

  1. 1

    Capability check

    WebGPU or WebAssembly availability is confirmed before the feature is offered.

  2. 2

    Tiled inference

    The image is processed in overlapping tiles inside a Web Worker.

  3. 3

    Reassemble

    Tiles are recombined onto the canvas and the result enters the history.

Especificações

  • 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.

Por que usar o KaiConvert

  • Sem upload nas ferramentas locais
  • Funciona offline depois que a página carrega
  • A IA em nuvem é sempre sinalizada antes de começar

Perguntas frequentes

Ferramentas relacionadas