Yes. SVG はピクセルではなく XML の幾何学であり、ぼかしなしに任意のサイズにスケールできますが、サンプルにピクセルグリッドはありません。. Compression artefacts around a subject are exactly what a segmentation model has to disambiguate, so cleaner input gives a cleaner edge.
SVGに特別なものは？
+
Yes — ファイルはテキストとして編集されますので 結果は無限に拡張可能で バージョン管理においては 変更可能です. That shapes both what the model sees and what the output can carry.
Yes: free accounts process images up to 5 MB each, which an SVG essentially never approaches — a complex illustration is usually a few hundred kilobytes of text. In practice the thing that makes an SVG heavy is thousands of path nodes from an auto-trace, not the file size, and that is a different problem from the cap.
透明度の決定は，透明度の変更を行う段階である。 That is the step most people actually came for, and doing it after SVG の背景を削除 rather than before it means the conversion works from the file you finally settled on.
