diff --git a/docs/project-memory/shared-memory/pitfalls.md b/docs/project-memory/shared-memory/pitfalls.md
index 7ff9389c7..dd2c8758d 100644
--- a/docs/project-memory/shared-memory/pitfalls.md
+++ b/docs/project-memory/shared-memory/pitfalls.md
@@ -368,6 +368,14 @@
 - 验证：`npm run test -- src/services/image-editor/editorProjectClient.test.ts src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.test.tsx -- --runInBand`；后端验证至少覆盖 `editor_image_edit_request_omits_price_mud_points` 和 `editor_image_edit_can_complete_by_replacing_target_layer`。
 - 关联：`src/services/image-editor/editorProjectClient.ts`、`src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.ts`、`server-rs/crates/api-server/src/editor_project.rs`。
 
+## 图片画布快速编辑尺寸要区分业务目标和 provider 对齐尺寸
+
+- 现象：原图经过快速编辑后 Resolution 变成近似比例的 1K / 2K 预设；原图或框选标记图宽高不是 16 的倍数时，VectorEngine edits 直接拒绝请求。
+- 原因：前端已有源图精确 `originalWidth/originalHeight`，提交时却按最近常用比例和 K 档重新计算 `size`；后端又把非 16 倍数的目标尺寸和原始参考图字节直接放进 multipart，并以 provider 回图宽高落库和覆盖画布图层。
+- 处理：画布快速编辑提交源图精确尺寸作为业务目标；api-server 只在 provider 边界向右、向下复制边缘像素，把每张参考图和目标尺寸临时补齐到 16 的倍数，收到回图后裁回业务目标尺寸再持久化。若上游异常返回其他尺寸，先按目标比例裁切缩放；临时对齐尺寸不能进入 OSS 元数据、`editor_project_resource`、`editor_asset` 或画布 Resolution。前端 inline 回填也保留源图显示尺寸和 Resolution，避免旧回包再次放大图层。
+- 验证：`npm run test -- src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts src/components/image-editor/ImageCanvasGenerationLayerModel.test.ts src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.test.tsx` 覆盖 `1537x1025` 精确提交和源图回填；`cargo test -p api-server editor_image_edit_aligns_provider_images_and_restores_source_dimensions --manifest-path server-rs/Cargo.toml` 覆盖 provider `1552x1040`、额外参考图独立对齐和回图恢复。
+- 关联：`src/components/image-editor/ImageCanvasGenerationSubmissionModel.ts`、`src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.ts`、`src/components/image-editor/ImageCanvasGenerationLayerModel.ts`、`server-rs/crates/api-server/src/editor_project.rs`。
+
 ## 图片画布快速编辑元数据必须记录原图引用
 
 - 现象：快速编辑生成的新图可以替换画布，但打开图片信息时“生成输入”里看不到被修改的原图。
diff --git a/docs/technical/【前端架构】图片画布编辑器MVP接入方案-2026-06-11.md b/docs/technical/【前端架构】图片画布编辑器MVP接入方案-2026-06-11.md
index 8b45587f6..9e3c44482 100644
--- a/docs/technical/【前端架构】图片画布编辑器MVP接入方案-2026-06-11.md
+++ b/docs/technical/【前端架构】图片画布编辑器MVP接入方案-2026-06-11.md
@@ -87,7 +87,7 @@
 - `POST /api/editor/images/background-removals`：接收当前图片源，校验登录态后由 api-server 解析为图片文件并转发到 BiRefNet 去背景服务；请求可携带 `projectId`、`targetLayerId`、`assetFolderId`、`assetLabel`、`sourceResourceId` 和 `canvasCompletion`，有 `canvasCompletion` 时完成后按生成占位写入结果图层，否则沿用旧的目标图层替换路径；响应返回 `imageSrc`、`objectKey`、`assetObjectId`、`width`、`height`、`taskId`、`elapsedMs`、`provider` 和可选 `project` 快照。服务地址由 `GENARRATIVE_EDITOR_BACKGROUND_REMOVAL_BASE_URL` 配置，令牌只在服务端通过 `GENARRATIVE_EDITOR_BACKGROUND_REMOVAL_TOKEN` 注入。
 - `POST /api/editor/icon-spritesheets/generations`：按图标规范图和素材描述数组生成 spritesheet，生成成功后 api-server 先保存带纯色背景 spritesheet 源图，再调用 BgFilter 生成透明 spritesheet。请求支持 `model`、`screenColor`、`segModel`、`aspectRatio`、`imageSize`、`priceMudPoints`、`projectId`、`assetFolderId` 和 `generationInputs`；`priceMudPoints` 必须来自编辑器生成计费配置中对应生图模型的尺寸档位（如 `nanobanana2` 的 `0.5K / 1K / 2K` 或 `gpt-image-2` 的 `1K / 2K`），后端用 `editor_generation_config` 校验后才调用上游；`nanobanana2` 走原生 `generateContent` 并写入 `generationConfig.imageConfig.aspectRatio/imageSize`，`0.5K` 传 `"512"`；`gpt-image-2` 走 `/v1/images/edits`。后端保存透明 spritesheet project resource / 账号素材，并随响应返回对应快照。
 - `POST /api/editor/ui-designs/assets/extractions`：前端把红色框选轮廓绘入本地临时图后，先将该图上传 OSS 并确认 asset object，再以返回的 `objectKey` 作为参考图入队；Data URL / Blob URL 只允许停留在上传前的浏览器临时态。接口固定 `gpt-image-2` 和自动决策纯色背景素材提取提示词生成素材 spritesheet，生成成功后 api-server 先保存带纯色背景 spritesheet 源图，再调用 BgFilter 生成透明 spritesheet，并按连通域自动拆分为 `素材 1..N`，返回结构复用图标 spritesheet 响应。请求必须携带 `screenColor`、`segModel`、`aspectRatio: "1:1"`、`imageSize: "1K" | "2K"` 和 `priceMudPoints`；框选数量不超过 6 个时前端按 `1:1·1K` 与 gpt-image-2 1K 价格提交，超过 6 个时按 `1:1·2K` 与 2K 价格提交。后端必须在调用上游前校验比例、尺寸和泥点价格，只允许 `1:1 / 1K / 2K`。请求可携带 `projectId`、`assetFolderId`、`generationInputs` 和 `spritesheetLabel`，后端保存 spritesheet / 拆分素材并返回对应 resource / asset 快照；前端必须把 spritesheet 原图与拆分素材都加入画布。
-- `POST /api/editor/images/edits`：按提示词和当前图片的已登记 `objectKey` / `resourceId` 调用 VectorEngine edits，返回新的生成图片元数据；图片快速编辑当前只提交 `sourceImageSrc`，不提交隐藏的 `referenceImageSrcs`。本地红框标记图必须先上传再提交 objectKey；请求携带 project / asset 上下文时由后端创建新 resource / asset，前端只消费响应快照。
+- `POST /api/editor/images/edits`：按提示词和当前图片的已登记 `objectKey` / `resourceId` 调用 VectorEngine edits，返回新的生成图片元数据；图片快速编辑当前只提交 `sourceImageSrc`，不提交隐藏的 `referenceImageSrcs`。画布快速编辑必须把源图精确 `originalWidth x originalHeight` 作为业务目标 `size` 提交，不能重新映射为近似比例或 1K / 2K 预设；api-server 在 VectorEngine provider 边界把目标尺寸和所有 multipart 参考图临时补齐到 16 的倍数，回图后恢复到业务目标精确尺寸，再落 OSS、project resource、账号素材和画布快照。16 对齐尺寸不得泄漏到响应、持久化资源或图层 Resolution。本地红框标记图必须先上传再提交 objectKey；请求携带 project / asset 上下文时由后端创建新 resource / asset，前端只消费响应快照。
 - `POST /api/editor/videos/generations`：按视频描述、模型、比例、时长、分辨率、模式、声音、默认联网搜索标记和泥点价格生成视频。前端可选模型为 `seedance2.0-fast`、`seedance2.0`、`kling3.0`、`kling3.0-omni`，默认 `seedance2.0-fast`；后端必须将 `seedance2.0-fast` 映射到 `doubao-seedance-2-0-fast-260128`，将 `seedance2.0` 映射到 `doubao-seedance-2-0-260128`，两者不得混用。后端允许 6 类比例、4 到 15 秒整数、`480p / 720p / 1080p`，并拒绝 `seedance2.0-fast + 1080p`；`sound=on/off` 映射 Ark `generate_audio=true/false`。后端复用 Ark / VectorEngine content generation task 轮询链路，下载最终视频并持久化到 OSS；请求携带 `projectId` / `assetFolderId` 时同步创建 project resource / 账号素材并返回 `project` / `asset` 快照，基础响应返回 `videoSrc`、尺寸、prompt、model、provider、taskId、durationSeconds、resolution 和 `priceMudPoints`。
 - `POST /api/editor/audios/sound-effects/generations` 与 `POST /api/editor/audios/background-music/generations`：按音效 / 背景音乐参数生成音频并持久化到 OSS；请求携带 `projectId` / `assetFolderId` 时同步创建 project resource / 账号素材并返回 `project` / `resource` / `asset` 快照，基础响应返回 `audioSrc`、prompt、model、provider、taskId、duration、歌词和 `priceMudPoints`。
 
diff --git a/server-rs/crates/api-server/src/editor_project.rs b/server-rs/crates/api-server/src/editor_project.rs
index 7c81b2ba9..9969b55aa 100644
--- a/server-rs/crates/api-server/src/editor_project.rs
+++ b/server-rs/crates/api-server/src/editor_project.rs
@@ -1910,6 +1910,148 @@ fn editor_generation_provider_image_size_for_model(
     image_size
 }
 
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+struct EditorImageEditDimensions {
+    target_width: u32,
+    target_height: u32,
+    provider_width: u32,
+    provider_height: u32,
+}
+
+fn align_editor_image_edit_dimension(value: u32) -> u32 {
+    value.saturating_add(15) / 16 * 16
+}
+
+fn encode_editor_image_edit_png(
+    image: image::DynamicImage,
+    failure_status: StatusCode,
+    failure_message: &str,
+) -> Result<Vec<u8>, AppError> {
+    let mut bytes = Cursor::new(Vec::new());
+    image
+        .write_to(&mut bytes, image::ImageFormat::Png)
+        .map_err(|error| {
+            AppError::from_status(failure_status).with_details(json!({
+                "provider": "editor-image-edit",
+                "message": format!("{failure_message}：{error}"),
+            }))
+        })?;
+    Ok(bytes.into_inner())
+}
+
+fn prepare_editor_image_edit_references(
+    reference_images: &mut [OpenAiReferenceImage],
+    target_size: &str,
+) -> Result<EditorImageEditDimensions, AppError> {
+    let (target_width, target_height) = target_size.split_once('x').ok_or_else(|| {
+        AppError::from_status(StatusCode::BAD_REQUEST).with_details(json!({
+            "provider": "editor-image-edit",
+            "message": "图片改造目标尺寸无效",
+        }))
+    })?;
+    let target_width = target_width.parse::<u32>().map_err(|_| {
+        AppError::from_status(StatusCode::BAD_REQUEST).with_details(json!({
+            "provider": "editor-image-edit",
+            "message": "图片改造目标宽度无效",
+        }))
+    })?;
+    let target_height = target_height.parse::<u32>().map_err(|_| {
+        AppError::from_status(StatusCode::BAD_REQUEST).with_details(json!({
+            "provider": "editor-image-edit",
+            "message": "图片改造目标高度无效",
+        }))
+    })?;
+    let provider_width = align_editor_image_edit_dimension(target_width);
+    let provider_height = align_editor_image_edit_dimension(target_height);
+    if reference_images.is_empty() {
+        return Err(
+            AppError::from_status(StatusCode::BAD_REQUEST).with_details(json!({
+                "provider": "editor-image-edit",
+                "message": "图片改造缺少原图",
+            })),
+        );
+    }
+    for (index, reference) in reference_images.iter_mut().enumerate() {
+        let decoded = image::load_from_memory(reference.bytes.as_slice()).map_err(|error| {
+            AppError::from_status(StatusCode::BAD_REQUEST).with_details(json!({
+                "provider": "editor-image-edit",
+                "message": format!("图片改造参考图不是有效图片：{error}"),
+                "referenceIndex": index,
+            }))
+        })?;
+        let width = decoded.width();
+        let height = decoded.height();
+        let reference_provider_width = align_editor_image_edit_dimension(width);
+        let reference_provider_height = align_editor_image_edit_dimension(height);
+        if reference_provider_width == width && reference_provider_height == height {
+            continue;
+        }
+
+        // 中文注释：只在 provider 边界向右、向下复制边缘像素补齐到 16 倍数，
+        // 不缩放、不裁切用户原图；生成结果持久化前会恢复到业务目标精确尺寸。
+        let source = decoded.to_rgba8();
+        let mut aligned =
+            image::RgbaImage::new(reference_provider_width, reference_provider_height);
+        for y in 0..reference_provider_height {
+            let source_y = y.min(height.saturating_sub(1));
+            for x in 0..reference_provider_width {
+                let source_x = x.min(width.saturating_sub(1));
+                aligned.put_pixel(x, y, *source.get_pixel(source_x, source_y));
+            }
+        }
+        reference.bytes = encode_editor_image_edit_png(
+            image::DynamicImage::ImageRgba8(aligned),
+            StatusCode::BAD_REQUEST,
+            "图片改造参考图 16 对齐失败",
+        )?;
+        reference.mime_type = "image/png".to_string();
+        reference.file_name = format!("editor-image-edit-reference-{}.png", index + 1);
+    }
+
+    Ok(EditorImageEditDimensions {
+        target_width,
+        target_height,
+        provider_width,
+        provider_height,
+    })
+}
+
+fn restore_editor_image_edit_output_dimensions(
+    output: DownloadedOpenAiImage,
+    dimensions: &EditorImageEditDimensions,
+) -> Result<DownloadedOpenAiImage, AppError> {
+    let decoded = image::load_from_memory(output.bytes.as_slice()).map_err(|error| {
+        AppError::from_status(StatusCode::BAD_GATEWAY).with_details(json!({
+            "provider": "vector-engine",
+            "message": format!("图片改造结果不是有效图片：{error}"),
+        }))
+    })?;
+    if decoded.width() == dimensions.target_width && decoded.height() == dimensions.target_height {
+        return Ok(output);
+    }
+
+    let restored = if decoded.width() == dimensions.provider_width
+        && decoded.height() == dimensions.provider_height
+    {
+        decoded.crop_imm(0, 0, dimensions.target_width, dimensions.target_height)
+    } else {
+        decoded.resize_to_fill(
+            dimensions.target_width,
+            dimensions.target_height,
+            image::imageops::FilterType::Lanczos3,
+        )
+    };
+    Ok(DownloadedOpenAiImage {
+        bytes: encode_editor_image_edit_png(
+            restored,
+            StatusCode::BAD_GATEWAY,
+            "恢复图片改造结果目标尺寸失败",
+        )?,
+        mime_type: "image/png".to_string(),
+        extension: "png".to_string(),
+    })
+}
+
 pub async fn edit_editor_image(
     State(state): State<AppState>,
     Extension(request_context): Extension<RequestContext>,
@@ -1965,7 +2107,7 @@ pub(crate) async fn edit_editor_image_for_owner(
     }
     let generation_options =
         normalize_editor_generation_options(payload.model.as_deref(), None, None);
-    let image_size = normalize_editor_image_generation_size(payload.size.as_deref());
+    let requested_image_size = normalize_editor_image_generation_size(payload.size.as_deref());
     let mut reference_images =
         Vec::with_capacity(1 + payload.reference_image_srcs.as_ref().map_or(0, Vec::len));
     reference_images.push(
@@ -1983,10 +2125,6 @@ pub(crate) async fn edit_editor_image_for_owner(
             parse_editor_reference_image(state, caller.owner_user_id.as_str(), source).await?,
         );
     }
-    let expected_price_mud_points =
-        resolve_editor_image_edit_price(state, generation_options.model, image_size.as_ref())
-            .await?;
-
     let settings = require_openai_image_settings(state)?.with_external_api_audit_context(
         request_context,
         caller.audit_subject_user_id.clone(),
@@ -1996,6 +2134,15 @@ pub(crate) async fn edit_editor_image_for_owner(
             .or_else(|| payload.project_id.clone()),
     );
     let http_client = build_openai_image_http_client(&settings)?;
+    let edit_dimensions =
+        prepare_editor_image_edit_references(&mut reference_images, requested_image_size.as_ref())?;
+    let image_size = format!(
+        "{}x{}",
+        edit_dimensions.provider_width, edit_dimensions.provider_height
+    );
+    let expected_price_mud_points =
+        resolve_editor_image_edit_price(state, generation_options.model, image_size.as_ref())
+            .await?;
     let generated = execute_billable_asset_operation_with_cost(
         state,
         caller.owner_user_id.as_str(),
@@ -2021,6 +2168,7 @@ pub(crate) async fn edit_editor_image_for_owner(
             "message": "VectorEngine 未返回图片",
         }))
     })?;
+    let image = restore_editor_image_edit_output_dimensions(image, &edit_dimensions)?;
     let (width, height) = image::load_from_memory(image.bytes.as_slice())
         .map(|image| (image.width(), image.height()))
         .unwrap_or((1024, 1024));
@@ -6435,6 +6583,59 @@ mod tests {
         );
     }
 
+    #[test]
+    fn editor_image_edit_aligns_provider_images_and_restores_source_dimensions() {
+        fn encode_png(width: u32, height: u32) -> Vec<u8> {
+            let image = image::DynamicImage::new_rgba8(width, height);
+            let mut bytes = Cursor::new(Vec::new());
+            image
+                .write_to(&mut bytes, image::ImageFormat::Png)
+                .expect("test image should encode");
+            bytes.into_inner()
+        }
+
+        let mut references = vec![
+            OpenAiReferenceImage {
+                bytes: encode_png(1537, 1025),
+                mime_type: "image/png".to_string(),
+                file_name: "source.png".to_string(),
+            },
+            OpenAiReferenceImage {
+                bytes: encode_png(501, 333),
+                mime_type: "image/png".to_string(),
+                file_name: "style.png".to_string(),
+            },
+        ];
+
+        let dimensions = prepare_editor_image_edit_references(&mut references, "1537x1025")
+            .expect("references should align for provider");
+        assert_eq!(dimensions.target_width, 1537);
+        assert_eq!(dimensions.target_height, 1025);
+        assert_eq!(dimensions.provider_width, 1552);
+        assert_eq!(dimensions.provider_height, 1040);
+        let source = image::load_from_memory(references[0].bytes.as_slice()).unwrap();
+        assert_eq!((source.width(), source.height()), (1552, 1040));
+        let extra = image::load_from_memory(references[1].bytes.as_slice()).unwrap();
+        assert_eq!((extra.width(), extra.height()), (512, 336));
+
+        let restored = restore_editor_image_edit_output_dimensions(
+            DownloadedOpenAiImage {
+                bytes: encode_png(1552, 1040),
+                mime_type: "image/png".to_string(),
+                extension: "png".to_string(),
+            },
+            &dimensions,
+        )
+        .expect("provider output should restore source dimensions");
+        let restored_image = image::load_from_memory(restored.bytes.as_slice()).unwrap();
+        assert_eq!(
+            (restored_image.width(), restored_image.height()),
+            (1537, 1025)
+        );
+        assert_eq!(restored.mime_type, "image/png");
+        assert_eq!(restored.extension, "png");
+    }
+
     #[test]
     fn editor_generation_dimensions_follow_model_options() {
         let default_generation = normalize_editor_generation_options(None, Some("1:1"), Some("1K"));
diff --git a/src/components/image-editor/ImageCanvasGenerationLayerModel.test.ts b/src/components/image-editor/ImageCanvasGenerationLayerModel.test.ts
index dfa3993aa..88f006286 100644
--- a/src/components/image-editor/ImageCanvasGenerationLayerModel.test.ts
+++ b/src/components/image-editor/ImageCanvasGenerationLayerModel.test.ts
@@ -287,8 +287,8 @@ describe('ImageCanvasGenerationLayerModel', () => {
     const sourceLayer = createSourceLayer({
       width: 320,
       height: 240,
-      originalWidth: 1024,
-      originalHeight: 768,
+      originalWidth: 1537,
+      originalHeight: 1025,
     });
     const layer = applyImageEditResultToSourceLayer({
       generated: createGenerated({
@@ -313,10 +313,10 @@ describe('ImageCanvasGenerationLayerModel', () => {
       id: 'layer-source',
       title: '源图',
       src: 'data:image/png;base64,edited',
-      width: 1536,
-      height: 1024,
-      originalWidth: 1536,
-      originalHeight: 1024,
+      width: 320,
+      height: 240,
+      originalWidth: 1537,
+      originalHeight: 1025,
       resourceId: 'resource-edited',
       sourceResourceId: 'resource-source',
       objectKey: 'generated/edited.png',
diff --git a/src/components/image-editor/ImageCanvasGenerationLayerModel.ts b/src/components/image-editor/ImageCanvasGenerationLayerModel.ts
index d13b56521..b88025f29 100644
--- a/src/components/image-editor/ImageCanvasGenerationLayerModel.ts
+++ b/src/components/image-editor/ImageCanvasGenerationLayerModel.ts
@@ -247,18 +247,10 @@ export function applyImageEditResultToSourceLayer({
   sourceLayer: CanvasLayer;
   generationInputs: CanvasGenerationInputs;
 }): CanvasLayer {
-  const originalWidth =
-    generated.width || sourceLayer.originalWidth || sourceLayer.width;
-  const originalHeight =
-    generated.height || sourceLayer.originalHeight || sourceLayer.height;
-  const { width, height } = resolveLayerResolutionSize(
-    originalWidth,
-    originalHeight,
-    {
-      width: sourceLayer.width,
-      height: sourceLayer.height,
-    },
-  );
+  const originalWidth = sourceLayer.originalWidth || sourceLayer.width;
+  const originalHeight = sourceLayer.originalHeight || sourceLayer.height;
+  const width = sourceLayer.width;
+  const height = sourceLayer.height;
   return applyGeneratedMetadata(
     {
       ...sourceLayer,
diff --git a/src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts b/src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts
index 85c6078e4..fc221c4c4 100644
--- a/src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts
+++ b/src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts
@@ -122,6 +122,8 @@ describe('ImageCanvasGenerationSubmissionModel', () => {
     const sourceLayer = createLayer({
       objectKey: 'generated-character-drafts/editor/source.png',
       assetKind: 'character',
+      originalWidth: 1537,
+      originalHeight: 1025,
     });
 
     const plan = buildImageGenerationSubmissionPlan({
@@ -150,7 +152,7 @@ describe('ImageCanvasGenerationSubmissionModel', () => {
       normalizedPrompt: '把当前图改成雨天',
       sourceLayer,
       editInput: {
-        size: '1024x768',
+        size: '1537x1025',
         model: IMAGE_MODEL_GPT_IMAGE_2,
       },
       result: {
diff --git a/src/components/image-editor/ImageCanvasGenerationSubmissionModel.ts b/src/components/image-editor/ImageCanvasGenerationSubmissionModel.ts
index 6f2d1c139..7552c7837 100644
--- a/src/components/image-editor/ImageCanvasGenerationSubmissionModel.ts
+++ b/src/components/image-editor/ImageCanvasGenerationSubmissionModel.ts
@@ -41,11 +41,8 @@ import {
   DEFAULT_VIDEO_WEB_SEARCH_ENABLED,
   ICON_DESCRIPTION_LIMIT,
   IMAGE_MODEL_GPT_IMAGE_2,
-  inferEditorImageAspectRatio,
-  inferEditorImageSizeLabel,
   normalizeEditorImageModel,
   resolveCharacterAnimationSourceImageSrc,
-  resolveEditorImageGenerationPixelSize,
   SEEDANCE_VIDEO_REFERENCE_LIMITS,
   SPEC_GENERATION_ASPECT_RATIO,
   SPEC_GENERATION_IMAGE_SIZE,
@@ -260,29 +257,12 @@ export function buildImageGenerationSubmissionPlan({
     const basePrompt = dialog.prompt.trim() || '快速编辑图片';
     const normalizedQuickEditPrompt = basePrompt;
     const imageModel = IMAGE_MODEL_GPT_IMAGE_2;
-    const aspectRatio =
-      dialog.aspectRatio ??
-      inferEditorImageAspectRatio(
-        sourceLayer.originalWidth,
-        sourceLayer.originalHeight,
-      );
-    const imageSize =
-      dialog.imageSize ??
-      inferEditorImageSizeLabel(
-        sourceLayer.originalWidth,
-        sourceLayer.originalHeight,
-      );
-    const outputSize = resolveEditorImageGenerationPixelSize({
-      model: imageModel,
-      aspectRatio,
-      imageSize,
-    });
     return {
       kind: 'quick-edit',
       normalizedPrompt: normalizedQuickEditPrompt,
       sourceLayer,
       editInput: {
-        size: `${outputSize.width}x${outputSize.height}`,
+        size: `${sourceLayer.originalWidth}x${sourceLayer.originalHeight}`,
         model: imageModel,
       },
       result: {
diff --git a/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.test.tsx b/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.test.tsx
index 2261f329c..2d548bc06 100644
--- a/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.test.tsx
+++ b/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.test.tsx
@@ -535,7 +535,13 @@ describe('useImageCanvasGenerationSubmissionWorkflow', () => {
     editEditorImageMock.mockResolvedValueOnce(
       createGenerated({ prompt: '快速修图' }),
     );
-    render(<SubmissionWorkflowHarness />);
+    render(
+      <SubmissionWorkflowHarness
+        initialLayers={[
+          createLayer({ originalWidth: 1537, originalHeight: 1025 }),
+        ]}
+      />,
+    );
 
     fireEvent.click(screen.getByRole('button', { name: '打开快速编辑' }));
     fireEvent.click(screen.getByRole('button', { name: '填写快速编辑' }));
@@ -550,7 +556,7 @@ describe('useImageCanvasGenerationSubmissionWorkflow', () => {
         expect.objectContaining({
           prompt: '快速修图',
           sourceImageSrc: 'resource-source',
-          size: '1024x768',
+          size: '1537x1025',
           model: 'gpt-image-2',
         }),
       );
diff --git a/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.ts b/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.ts
index 61f493198..69f6a1e35 100644
--- a/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.ts
+++ b/src/components/image-editor/useImageCanvasGenerationSubmissionWorkflow.ts
@@ -1268,12 +1268,11 @@ export function useImageCanvasGenerationSubmissionWorkflow({
         quickEditSourceLayer.originalWidth,
         quickEditSourceLayer.originalHeight,
       );
-    const quickEditOutputSize = resolveEditorImageGenerationPixelSize({
-      model: quickEditPanel.model,
-      aspectRatio: quickEditAspectRatio,
-      imageSize: quickEditImageSize,
-    });
-    const quickEditSize = `${quickEditOutputSize.width}x${quickEditOutputSize.height}`;
+    const quickEditOutputSize = {
+      width: quickEditSourceLayer.originalWidth,
+      height: quickEditSourceLayer.originalHeight,
+    };
+    const quickEditSize = `${quickEditSourceLayer.originalWidth}x${quickEditSourceLayer.originalHeight}`;
     const basePrompt =
       quickEditPanel.prompt.trim() ||
       (panelMode === 'redraw' ? '重绘图片' : '快速编辑图片');
@@ -1678,6 +1677,7 @@ export function useImageCanvasGenerationSubmissionWorkflow({
             editEditorImage({
               prompt: submissionPlan.normalizedPrompt,
               sourceImageSrc: referenceImageSrc,
+              size: `${submissionPlan.sourceLayer.originalWidth}x${submissionPlan.sourceLayer.originalHeight}`,
               projectId,
               assetKind: submissionPlan.sourceLayer.assetKind,
               generationInputs: submissionPlan.generationInputs,