拆分编辑器生成提交模型

抽出图片生成请求与结果快照构建逻辑

补充生成提交模型单测

更新 TRACKING.md 记录第三十五阶段验证

TRACKING.md

@@ -151,3 +151,4 @@
 - 2026-06-17 前端拆分第三十二阶段：继续收口 `ImageCanvasGenerationComposerView`，新增 `ImageCanvasBasicGenerationComposerView`、`ImageCanvasCharacterGenerationComposerView` 和 `ImageCanvasEditGenerationModalView`，把普通生图跟随框、角色形象生成面板和修改图片弹窗从 Composer 内联 JSX 中抽出；Composer 降至 312 行，只保留生成模式分流、portal 菜单和各面板装配。新增三组子视图单测覆盖普通生图 prompt / 参考图 / 提交 / 关闭、角色参考图菜单 / 状态恢复 / 提交、修改图片弹窗提示词 / 失败 / 关闭。统一验证命令：`npm run test -- src/components/image-editor/ImageCanvasSidebarView.test.tsx src/components/image-editor/ImageCanvasBasicGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasCharacterGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasEditGenerationModalView.test.tsx src/components/image-editor/ImageCanvasEditorShellView.test.tsx src/components/image-editor/ImageCanvasBottomToolbarView.test.tsx src/components/image-editor/ImageCanvasPanelDockView.test.tsx src/components/image-editor/ImageCanvasContextMenusView.test.tsx src/components/image-editor/ImageCanvasSelectedLayerToolbarView.test.tsx src/components/image-editor/ImageCanvasIconSpritesheetComposerView.test.tsx src/components/image-editor/ImageCanvasSpecGenerationPanelView.test.tsx src/components/image-editor/ImageCanvasGenerationImageOptionsView.test.tsx src/components/image-editor/ImageCanvasQuickEditPanelView.test.tsx src/components/image-editor/ImageCanvasCharacterAnimationPanelView.test.tsx src/components/image-editor/useImageCanvasGenerationSurface.test.tsx src/components/image-editor/useImageCanvasGenerationWorkflow.test.tsx src/components/image-editor/ImageCanvasEditorView.test.tsx`、`npm run typecheck`、`npm run check:encoding`、`git diff --check`；浏览器回归：`http://127.0.0.1:10007/editor/canvas` 临时 `XDG_CONFIG_HOME` 下 Chrome headless 打开成功，未登录显示 `账号入口` 且控制台仅有预期 `/api/auth/refresh` 401；关闭登录后 `画布背景设置` 可打开，点击 `暖灰` 后 viewport 背景为 `rgb(243, 240, 234)` 且 `background-image: none`；`打开图层` 切换后侧栏显示 `图层`，`AI画布工具栏` 仍可见；点击 `生成工具` 后 `Image Generator`、`生成图片` 对话框和底部工具栏均可见。
 - 2026-06-17 前端拆分第三十三阶段：继续收口 `ImageCanvasAssetLibraryPanelView`，新增 `ImageCanvasAssetFolderSectionView` 和 `ImageCanvasAssetRowView`，把素材库文件夹头 / 文件夹 drop 区域、素材卡片 / 上传进度 / 重命名 / 选择模式从素材库父面板中拆成两个完整 surface；素材库父面板降至 279 行，只保留素材列表容器、新建文件夹表单、批量操作栏和框选遮罩。新增素材行单测覆盖普通点击加入画布、选择模式改为选中、重命名 Enter 提交和上传中禁用 / 进度显示。统一验证命令：`npm run test -- src/components/image-editor/ImageCanvasAssetRowView.test.tsx src/components/image-editor/ImageCanvasSidebarView.test.tsx src/components/image-editor/ImageCanvasBasicGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasCharacterGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasEditGenerationModalView.test.tsx src/components/image-editor/ImageCanvasEditorShellView.test.tsx src/components/image-editor/ImageCanvasBottomToolbarView.test.tsx src/components/image-editor/ImageCanvasPanelDockView.test.tsx src/components/image-editor/ImageCanvasContextMenusView.test.tsx src/components/image-editor/ImageCanvasSelectedLayerToolbarView.test.tsx src/components/image-editor/ImageCanvasIconSpritesheetComposerView.test.tsx src/components/image-editor/ImageCanvasSpecGenerationPanelView.test.tsx src/components/image-editor/ImageCanvasGenerationImageOptionsView.test.tsx src/components/image-editor/ImageCanvasQuickEditPanelView.test.tsx src/components/image-editor/ImageCanvasCharacterAnimationPanelView.test.tsx src/components/image-editor/useImageCanvasGenerationSurface.test.tsx src/components/image-editor/useImageCanvasGenerationWorkflow.test.tsx src/components/image-editor/ImageCanvasEditorView.test.tsx`、`npm run typecheck`、`npm run check:encoding`、`git diff --check`；浏览器回归：`http://127.0.0.1:10007/editor/canvas` 临时 `XDG_CONFIG_HOME` 下 Chrome headless 打开成功，未登录显示 `账号入口` 且控制台仅有预期 `/api/auth/refresh` 401；默认素材栏显示 `项目素材` 文件夹、上传入口和底部 `AI画布工具栏`；关闭登录后 `画布背景设置` 可打开，点击 `暖灰` 后 viewport 背景为 `rgb(243, 240, 234)` 且 `background-image: none`；`打开图层` 切换后侧栏显示 `图层`；点击 `生成工具` 后 `Image Generator`、`生成图片` 对话框和底部工具栏均可见。
 - 2026-06-17 前端拆分第三十四阶段：继续收口 `ImageCanvasStageView`，新增 `ImageCanvasWorldView`，把画布世界表面、吸附参考线、可见图层排序 / 悬浮 / 选中 / 锁定 / 生成态、元数据角标、框选矩形、生成占位框和浮动生成状态从 StageView 内联 JSX 中抽出；StageView 降至 324 行，继续保留 viewport 宿主、drop overlay、左下 dock、底部工具栏、右键菜单和选中图片工具栏装配。新增 world view 单测覆盖隐藏图层过滤、悬浮尺寸、生成态、元数据按钮、吸附线、框选矩形和生成占位框事件。统一验证命令：`npm run test -- src/components/image-editor/ImageCanvasWorldView.test.tsx`、`npm run test -- src/components/image-editor/ImageCanvasAssetRowView.test.tsx src/components/image-editor/ImageCanvasSidebarView.test.tsx src/components/image-editor/ImageCanvasBasicGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasCharacterGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasEditGenerationModalView.test.tsx src/components/image-editor/ImageCanvasEditorShellView.test.tsx src/components/image-editor/ImageCanvasBottomToolbarView.test.tsx src/components/image-editor/ImageCanvasPanelDockView.test.tsx src/components/image-editor/ImageCanvasContextMenusView.test.tsx src/components/image-editor/ImageCanvasSelectedLayerToolbarView.test.tsx src/components/image-editor/ImageCanvasIconSpritesheetComposerView.test.tsx src/components/image-editor/ImageCanvasSpecGenerationPanelView.test.tsx src/components/image-editor/ImageCanvasGenerationImageOptionsView.test.tsx src/components/image-editor/ImageCanvasQuickEditPanelView.test.tsx src/components/image-editor/ImageCanvasCharacterAnimationPanelView.test.tsx src/components/image-editor/ImageCanvasWorldView.test.tsx src/components/image-editor/useImageCanvasGenerationSurface.test.tsx src/components/image-editor/useImageCanvasGenerationWorkflow.test.tsx src/components/image-editor/ImageCanvasEditorView.test.tsx`、`npm run typecheck`、`npm run check:encoding`、`git diff --check`；浏览器回归：`http://127.0.0.1:10007/editor/canvas` 临时 `XDG_CONFIG_HOME` 下 Chrome headless 打开成功，未登录显示 `账号入口`；关闭后 `画布背景设置` 可打开，点击 `暖灰` 后 viewport 背景为 `rgb(243, 240, 234)` 且 `background-image: none`，`AI画布工具栏` 保持可见；`打开图层` 切换后侧栏显示 `图层`；点击 `生成工具` 后 `Image Generator`、`生成图片` 对话框和底部工具栏均可见，控制台仅有预期的未登录 `/api/auth/refresh` 401。
+- 2026-06-17 前端拆分第三十五阶段：继续收口 `useImageCanvasGenerationWorkflow`，新增 `ImageCanvasGenerationSubmissionModel`，把普通生图、修改图片、规范生成和角色形象生成的请求 payload、标准化 prompt、结果图层标题 / assetKind 和 generationInputs 快照构建从 workflow hook 中抽成纯模型；workflow hook 保留对话状态、真实 API 调用、图片引用解析、结果落图、选中和 fit 副作用，避免拆散生成生命周期。新增模型单测覆盖普通生图、修改图、带参考图的规范生成、带规范 / 常规参考图的角色生成和缺失源图异常；`useImageCanvasGenerationWorkflow` 从 1167 行降至 1104 行。统一验证命令：`npm run test -- src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts src/components/image-editor/useImageCanvasGenerationWorkflow.test.tsx`、`npm run test -- src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts src/components/image-editor/ImageCanvasAssetRowView.test.tsx src/components/image-editor/ImageCanvasSidebarView.test.tsx src/components/image-editor/ImageCanvasBasicGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasCharacterGenerationComposerView.test.tsx src/components/image-editor/ImageCanvasEditGenerationModalView.test.tsx src/components/image-editor/ImageCanvasEditorShellView.test.tsx src/components/image-editor/ImageCanvasBottomToolbarView.test.tsx src/components/image-editor/ImageCanvasPanelDockView.test.tsx src/components/image-editor/ImageCanvasContextMenusView.test.tsx src/components/image-editor/ImageCanvasSelectedLayerToolbarView.test.tsx src/components/image-editor/ImageCanvasIconSpritesheetComposerView.test.tsx src/components/image-editor/ImageCanvasSpecGenerationPanelView.test.tsx src/components/image-editor/ImageCanvasGenerationImageOptionsView.test.tsx src/components/image-editor/ImageCanvasQuickEditPanelView.test.tsx src/components/image-editor/ImageCanvasCharacterAnimationPanelView.test.tsx src/components/image-editor/ImageCanvasWorldView.test.tsx src/components/image-editor/useImageCanvasGenerationSurface.test.tsx src/components/image-editor/useImageCanvasGenerationWorkflow.test.tsx src/components/image-editor/ImageCanvasEditorView.test.tsx`、`npm run typecheck`、`npm run check:encoding`、`git diff --check`；浏览器回归：`http://127.0.0.1:10007/editor/canvas` 临时 `XDG_CONFIG_HOME` 下 Chrome headless 打开成功，未登录显示 `账号入口`；关闭后 `画布背景设置` 可打开，点击 `暖灰` 后 viewport 背景为 `rgb(243, 240, 234)` 且 `background-image: none`；点击 `生成工具` 后 `Image Generator`、`生成图片` 对话框和 `AI画布工具栏` 均可见，控制台仅有预期的未登录 `/api/auth/refresh` 401。

src/components/image-editor/ImageCanvasGenerationSubmissionModel.test.ts

@@ -0,0 +1,212 @@
+import { describe, expect, it } from 'vitest';
+
+import type { CanvasLayer } from './ImageCanvasEditorTypes';
+import { buildImageGenerationSubmissionPlan } from './ImageCanvasGenerationSubmissionModel';
+
+function createLayer(overrides: Partial<CanvasLayer> = {}): CanvasLayer {
+  return {
+    id: 'layer-source',
+    resourceId: 'resource-source',
+    title: '源图',
+    src: 'data:image/png;base64,source',
+    x: 120,
+    y: 140,
+    width: 320,
+    height: 240,
+    originalWidth: 1024,
+    originalHeight: 768,
+    zIndex: 2,
+    sourceType: 'uploaded',
+    ...overrides,
+  };
+}
+
+describe('ImageCanvasGenerationSubmissionModel', () => {
+  it('builds normal image generation submission plans', () => {
+    const plan = buildImageGenerationSubmissionPlan({
+      dialog: {
+        mode: 'generate',
+        prompt: '  一张发光主视觉  ',
+        status: 'idle',
+      },
+      layers: [],
+      nextGeneratedIndex: 3,
+    });
+
+    expect(plan).toEqual({
+      kind: 'image',
+      normalizedPrompt: '一张发光主视觉',
+      input: {
+        prompt: '一张发光主视觉',
+      },
+      result: {
+        generationInputs: {
+          fields: [{ title: '生成提示词', value: '一张发光主视觉' }],
+          references: [],
+        },
+      },
+    });
+  });
+
+  it('builds edit submission plans with the source layer snapshot', () => {
+    const sourceLayer = createLayer();
+
+    const plan = buildImageGenerationSubmissionPlan({
+      dialog: {
+        mode: 'edit',
+        prompt: '',
+        status: 'idle',
+        sourceLayerId: sourceLayer.id,
+      },
+      layers: [sourceLayer],
+      nextGeneratedIndex: 4,
+    });
+
+    expect(plan).toMatchObject({
+      kind: 'edit',
+      normalizedPrompt: '修改当前图片',
+      sourceLayer,
+      generationInputs: {
+        fields: [{ title: '修改要求', value: '修改当前图片' }],
+        references: [
+          {
+            title: '参考图',
+            label: '源图',
+            src: 'data:image/png;base64,source',
+          },
+        ],
+      },
+    });
+  });
+
+  it('builds spec generation plans with reference prompt semantics', () => {
+    const plan = buildImageGenerationSubmissionPlan({
+      dialog: {
+        mode: 'spec',
+        prompt: '',
+        status: 'idle',
+        specType: 'icon',
+        specValues: {
+          playSetting: '休闲消除',
+          artStyle: '清爽卡通',
+          bodyRatio: '3',
+          characterView: '',
+          customPrompt: '',
+        },
+        specReference: {
+          id: 'spec-ref',
+          label: '参考.png',
+          src: 'data:image/png;base64,ref',
+        },
+      },
+      layers: [],
+      nextGeneratedIndex: 8,
+    });
+
+    expect(plan).toMatchObject({
+      kind: 'image',
+      normalizedPrompt: 'AI 生成图片',
+      input: {
+        size: '2048x1152',
+        kind: 'spec',
+        referenceImageSrcs: ['data:image/png;base64,ref'],
+        prompt: expect.stringContaining('参考图生成规范'),
+      },
+      result: {
+        assetKind: 'icon-spec',
+        title: '图标素材规范 8',
+      },
+    });
+    expect(plan.kind === 'image' ? plan.result.generationInputs : null).toEqual({
+      fields: [
+        { title: '玩法设定', value: '休闲消除' },
+        { title: '美术风格', value: '清爽卡通' },
+      ],
+      references: [
+        {
+          title: '参考图',
+          label: '参考.png',
+          src: 'data:image/png;base64,ref',
+        },
+      ],
+    });
+  });
+
+  it('builds character plans with references and remembered model', () => {
+    const plan = buildImageGenerationSubmissionPlan({
+      dialog: {
+        mode: 'character',
+        prompt: '  白发骑士  ',
+        status: 'idle',
+        imageModel: 'gpt-image-2',
+        aspectRatio: '2:3',
+        imageSize: '2K',
+        characterSpecReference: {
+          id: 'spec',
+          label: '角色规范',
+          src: 'data:image/png;base64,spec',
+        },
+        characterReferences: [
+          {
+            id: 'ref-1',
+            label: '盔甲参考',
+            src: 'data:image/png;base64,armor',
+          },
+        ],
+      },
+      layers: [],
+      nextGeneratedIndex: 9,
+    });
+
+    expect(plan).toMatchObject({
+      kind: 'image',
+      normalizedPrompt: '白发骑士',
+      input: {
+        prompt: '白发骑士',
+        kind: 'character',
+        model: 'gpt-image-2',
+        aspectRatio: '2:3',
+        imageSize: '2K',
+        referenceImageSrcs: [
+          'data:image/png;base64,spec',
+          'data:image/png;base64,armor',
+        ],
+      },
+      result: {
+        assetKind: 'character',
+        title: '角色形象 9',
+      },
+      rememberImageModel: 'gpt-image-2',
+    });
+    expect(plan.kind === 'image' ? plan.result.generationInputs : null).toEqual({
+      fields: [{ title: '角色设定', value: '白发骑士' }],
+      references: [
+        {
+          title: '角色形象规范',
+          label: '角色规范',
+          src: 'data:image/png;base64,spec',
+        },
+        {
+          title: '常规参考图 1',
+          label: '盔甲参考',
+          src: 'data:image/png;base64,armor',
+        },
+      ],
+    });
+  });
+
+  it('throws when edit source layer is missing', () => {
+    expect(() =>
+      buildImageGenerationSubmissionPlan({
+        dialog: {
+          mode: 'edit',
+          prompt: '修图',
+          status: 'idle',
+          sourceLayerId: 'missing-layer',
+        },
+        layers: [],
+        nextGeneratedIndex: 1,
+      }),
+    ).toThrow('未找到要修改的图片');
+  });
+});

src/components/image-editor/ImageCanvasGenerationSubmissionModel.ts

@@ -0,0 +1,143 @@
+import type {
+  EditorImageGenerationInput,
+} from '../../services/image-editor/editorProjectClient';
+import type {
+  CanvasGenerationInputs,
+  CanvasLayer,
+  GenerateDialogState,
+} from './ImageCanvasEditorTypes';
+import {
+  DEFAULT_IMAGE_MODEL,
+  DEFAULT_SPEC_FORM_VALUES,
+  SPEC_GENERATION_SIZE,
+  SPEC_TYPE_LABEL,
+  buildCharacterGenerationInputs,
+  buildEditGenerationInputs,
+  buildImageGenerationInputs,
+  buildSpecGenerationInputs,
+  buildSpecPrompt,
+} from './ImageCanvasGenerationModel';
+
+type ImageGenerationSubmissionOptions = {
+  dialog: GenerateDialogState;
+  layers: CanvasLayer[];
+  nextGeneratedIndex: number;
+};
+
+export type ImageGenerationSubmissionPlan =
+  | {
+      kind: 'edit';
+      normalizedPrompt: string;
+      sourceLayer: CanvasLayer;
+      generationInputs: CanvasGenerationInputs;
+    }
+  | {
+      kind: 'image';
+      normalizedPrompt: string;
+      input: EditorImageGenerationInput;
+      result: {
+        assetKind?: CanvasLayer['assetKind'];
+        title?: string;
+        generationInputs: CanvasGenerationInputs;
+      };
+      rememberImageModel?: string;
+    };
+
+export function buildImageGenerationSubmissionPlan({
+  dialog,
+  layers,
+  nextGeneratedIndex,
+}: ImageGenerationSubmissionOptions): ImageGenerationSubmissionPlan {
+  const normalizedPrompt =
+    dialog.prompt.trim() ||
+    (dialog.mode === 'edit' ? '修改当前图片' : 'AI 生成图片');
+
+  if (dialog.mode === 'edit') {
+    const sourceLayer = layers.find((layer) => layer.id === dialog.sourceLayerId);
+    if (!sourceLayer) {
+      throw new Error('未找到要修改的图片');
+    }
+    return {
+      kind: 'edit',
+      normalizedPrompt,
+      sourceLayer,
+      generationInputs: buildEditGenerationInputs(
+        '修改要求',
+        normalizedPrompt,
+        sourceLayer,
+      ),
+    };
+  }
+
+  if (dialog.mode === 'spec') {
+    const specType = dialog.specType ?? 'custom';
+    const specValues = dialog.specValues ?? DEFAULT_SPEC_FORM_VALUES[specType];
+    return {
+      kind: 'image',
+      normalizedPrompt,
+      input: {
+        prompt: buildSpecPrompt(
+          specType,
+          specValues,
+          Boolean(dialog.specReference?.src),
+        ),
+        size: SPEC_GENERATION_SIZE,
+        model: DEFAULT_IMAGE_MODEL,
+        kind: 'spec',
+        ...(dialog.specReference?.src
+          ? { referenceImageSrcs: [dialog.specReference.src] }
+          : {}),
+      },
+      result: {
+        assetKind: specType === 'icon' ? 'icon-spec' : 'spec',
+        title: `${SPEC_TYPE_LABEL[specType]} ${nextGeneratedIndex}`,
+        generationInputs: buildSpecGenerationInputs(
+          specType,
+          specValues,
+          dialog.specReference,
+        ),
+      },
+    };
+  }
+
+  if (dialog.mode === 'character') {
+    const referenceImageSrcs = [
+      dialog.characterSpecReference?.src,
+      ...(dialog.characterReferences ?? []).map((reference) => reference.src),
+    ].filter((src): src is string => Boolean(src));
+    const imageModel = dialog.imageModel ?? DEFAULT_IMAGE_MODEL;
+    return {
+      kind: 'image',
+      normalizedPrompt,
+      input: {
+        prompt: normalizedPrompt,
+        kind: 'character',
+        model: imageModel,
+        aspectRatio: dialog.aspectRatio ?? '1:1',
+        imageSize: dialog.imageSize ?? '1K',
+        ...(referenceImageSrcs.length ? { referenceImageSrcs } : {}),
+      },
+      result: {
+        assetKind: 'character',
+        title: `角色形象 ${nextGeneratedIndex}`,
+        generationInputs: buildCharacterGenerationInputs(
+          normalizedPrompt,
+          dialog.characterSpecReference,
+          dialog.characterReferences,
+        ),
+      },
+      rememberImageModel: imageModel,
+    };
+  }
+
+  return {
+    kind: 'image',
+    normalizedPrompt,
+    input: {
+      prompt: normalizedPrompt,
+    },
+    result: {
+      generationInputs: buildImageGenerationInputs(normalizedPrompt),
+    },
+  };
+}

src/components/image-editor/useImageCanvasGenerationWorkflow.ts

@@ -36,22 +36,17 @@ import {
   ICON_FRAME_ORIGINAL_SIZE,
   SPEC_FRAME_DISPLAY_SIZE,
   SPEC_FRAME_ORIGINAL_SIZE,
-  SPEC_GENERATION_SIZE,
-  SPEC_TYPE_LABEL,
-  buildCharacterGenerationInputs,
   buildEditGenerationInputs,
   buildIconGenerationInputs,
-  buildImageGenerationInputs,
   buildQuickEditModelOptions,
   buildQuickEditSizeOptions,
-  buildSpecGenerationInputs,
-  buildSpecPrompt,
   calculateCharacterAnimationPrice,
   createCanvasLayerReference,
   isCanvasGenerationDialog,
   resolveCharacterAnimationSourceImageSrc,
   resolveImageGenerationErrorMessage,
 } from './ImageCanvasGenerationModel';
+import { buildImageGenerationSubmissionPlan } from './ImageCanvasGenerationSubmissionModel';
 import { formatImageSizeValue } from './ImageCanvasEditorModel';
 import type {
   CanvasGenerationDialogState,
@@ -801,92 +796,34 @@ export function useImageCanvasGenerationWorkflow({
       }
 
       try {
-        if (dialog.mode === 'edit') {
+        const submissionPlan = buildImageGenerationSubmissionPlan({
-          const sourceLayer = layers.find(
+          dialog,
-            (layer) => layer.id === dialog.sourceLayerId,
+          layers,
-          );
+          nextGeneratedIndex: layerCounterRef.current + 1,
-          if (!sourceLayer) {
+        });
-            throw new Error('未找到要修改的图片');
+        if (submissionPlan.kind === 'edit') {
-          }
           const referenceImageSrc = await resolveEditorImageReferenceDataUrl(
-            sourceLayer.src,
+            submissionPlan.sourceLayer.src,
           );
           const generated = await editEditorImage({
-            prompt: normalizedPrompt,
+            prompt: submissionPlan.normalizedPrompt,
             sourceImageSrc: referenceImageSrc,
           });
           addGeneratedResultLayer(generated, {
-            sourceLayer,
+            sourceLayer: submissionPlan.sourceLayer,
-            generationInputs: buildEditGenerationInputs(
+            generationInputs: submissionPlan.generationInputs,
-              '修改要求',
-              normalizedPrompt,
-              sourceLayer,
-            ),
-          });
-        } else if (dialog.mode === 'spec') {
-          const specType = dialog.specType ?? 'custom';
-          const specValues =
-            dialog.specValues ?? DEFAULT_SPEC_FORM_VALUES[specType];
-          const specPrompt = buildSpecPrompt(
-            specType,
-            specValues,
-            Boolean(dialog.specReference?.src),
-          );
-          const generated = await generateEditorImage({
-            prompt: specPrompt,
-            size: SPEC_GENERATION_SIZE,
-            model: DEFAULT_IMAGE_MODEL,
-            kind: 'spec',
-            ...(dialog.specReference?.src
-              ? { referenceImageSrcs: [dialog.specReference.src] }
-              : {}),
-          });
-          addGeneratedResultLayer(generated, {
-            frame: getGeneratingDialogPlaceholder(dialog),
-            assetKind: specType === 'icon' ? 'icon-spec' : 'spec',
-            title: `${SPEC_TYPE_LABEL[specType]} ${layerCounterRef.current + 1}`,
-            dialogId: canvasDialog?.id,
-            generationInputs: buildSpecGenerationInputs(
-              specType,
-              specValues,
-              dialog.specReference,
-            ),
-          });
-        } else if (dialog.mode === 'character') {
-          const referenceImageSrcs = [
-            dialog.characterSpecReference?.src,
-            ...(dialog.characterReferences ?? []).map(
-              (reference) => reference.src,
-            ),
-          ].filter((src): src is string => Boolean(src));
-          const generated = await generateEditorImage({
-            prompt: normalizedPrompt,
-            kind: 'character',
-            model: dialog.imageModel ?? DEFAULT_IMAGE_MODEL,
-            aspectRatio: dialog.aspectRatio ?? '1:1',
-            imageSize: dialog.imageSize ?? '1K',
-            ...(referenceImageSrcs.length ? { referenceImageSrcs } : {}),
-          });
-          setLastImageModel(dialog.imageModel ?? DEFAULT_IMAGE_MODEL);
-          addGeneratedResultLayer(generated, {
-            frame: getGeneratingDialogPlaceholder(dialog),
-            assetKind: 'character',
-            title: `角色形象 ${layerCounterRef.current + 1}`,
-            dialogId: canvasDialog?.id,
-            generationInputs: buildCharacterGenerationInputs(
-              normalizedPrompt,
-              dialog.characterSpecReference,
-              dialog.characterReferences,
-            ),
           });
         } else {
-          const generated = await generateEditorImage({
+          const generated = await generateEditorImage(submissionPlan.input);
-            prompt: normalizedPrompt,
+          if (submissionPlan.rememberImageModel) {
-          });
+            setLastImageModel(submissionPlan.rememberImageModel);
+          }
           addGeneratedResultLayer(generated, {
             frame: getGeneratingDialogPlaceholder(dialog),
+            assetKind: submissionPlan.result.assetKind,
+            title: submissionPlan.result.title,
             dialogId: canvasDialog?.id,
-            generationInputs: buildImageGenerationInputs(normalizedPrompt),
+            generationInputs: submissionPlan.result.generationInputs,
           });
         }
       } catch (error) {