完成视频部分的前后端

backend-java/src/main/java/com/nanri/aiimage/config/ImageVideoProperties.java

@@ -17,5 +17,5 @@ public class ImageVideoProperties {
     private String voiceCloneWorkflowId = "7652954356887961641";
     private String voiceSynthesisWorkflowId = "7652954297530105894";
     private int cozeConnectTimeoutMillis = 10000;
-    private int cozeReadTimeoutMillis = 600000;
+    private int cozeReadTimeoutMillis = 3600000;
 }

backend-java/src/main/java/com/nanri/aiimage/config/OssProperties.java

@@ -10,6 +10,7 @@ public class OssProperties {
     private String endpoint;
     private String publicEndpoint;
     private String bucket;
+    private String imageVideoBucket;
     private String accessKeyId;
     private String accessKeySecret;
 }

backend-java/src/main/java/com/nanri/aiimage/modules/file/service/oss/OssStorageService.java

@@ -38,10 +38,11 @@ public class OssStorageService {
 
     public UploadedResult uploadResultFileWithFreshDownloadUrl(File file, String moduleType) {
         String objectKey = String.format("result/%s/%s/%s", moduleType.toLowerCase(), UUID.randomUUID(), file.getName());
+        String bucket = resolveBucket(moduleType);
         OSS ossClient = buildClient();
         try {
-            ossClient.putObject(ossProperties.getBucket(), objectKey, file);
-            return new UploadedResult(objectKey, getPublicUrl(objectKey));
+            ossClient.putObject(bucket, objectKey, file);
+            return new UploadedResult(objectKey, getPublicUrl(objectKey, bucket));
         } finally {
             ossClient.shutdown();
         }
@@ -56,10 +57,11 @@ public class OssStorageService {
             objectName = file.getName();
         }
         String objectKey = String.format("upload/%s/%s/%s", normalizedModuleType, UUID.randomUUID(), objectName);
+        String bucket = resolveBucket(moduleType);
         OSS ossClient = buildClient();
         try {
-            ossClient.putObject(ossProperties.getBucket(), objectKey, file);
-            return new UploadedResult(objectKey, getPublicUrl(objectKey));
+            ossClient.putObject(bucket, objectKey, file);
+            return new UploadedResult(objectKey, getPublicUrl(objectKey, bucket));
         } finally {
             ossClient.shutdown();
         }
@@ -151,12 +153,16 @@ public class OssStorageService {
      * 获取公开（无签名）URL，格式：https://{bucket}.{endpoint}/{objectKey}
      */
     public String getPublicUrl(String value) {
+        return getPublicUrl(value, ossProperties.getBucket());
+    }
+
+    private String getPublicUrl(String value, String bucket) {
         if (value == null || value.isBlank()) {
             return null;
         }
         String objectKey = resolveObjectKey(value);
         String normalizedKey = objectKey.startsWith("/") ? objectKey.substring(1) : objectKey;
-        String host = String.format("%s.%s", ossProperties.getBucket(), publicEndpoint());
+        String host = String.format("%s.%s", bucket, publicEndpoint());
         try {
             return new URI("https", host, "/" + normalizedKey, null).toASCIIString();
         } catch (Exception ignored) {
@@ -210,6 +216,16 @@ public class OssStorageService {
         );
     }
 
+    private String resolveBucket(String moduleType) {
+        if (moduleType != null
+                && "IMAGE_VIDEO".equalsIgnoreCase(moduleType.trim())
+                && ossProperties.getImageVideoBucket() != null
+                && !ossProperties.getImageVideoBucket().isBlank()) {
+            return ossProperties.getImageVideoBucket().trim();
+        }
+        return ossProperties.getBucket();
+    }
+
     private String publicEndpoint() {
         String endpoint = ossProperties.getPublicEndpoint();
         if (endpoint == null || endpoint.isBlank()) {

backend-java/src/main/java/com/nanri/aiimage/modules/imagevideo/controller/ImageVideoController.java

@@ -53,7 +53,7 @@ public class ImageVideoController {
     @PostMapping("/douyin-copy")
     @Operation(summary = "抖音文案识别和仿写")
     public ApiResponse<DouyinCopyVo> douyinCopy(@Valid @RequestBody DouyinCopyRequest request) {
-        return ApiResponse.success(imageVideoCozeService.runDouyinCopy(request.getUserId(), request.getUrl()));
+        return ApiResponse.success(imageVideoCozeService.runDouyinCopy(request));
     }
 
     @PostMapping("/workflow/run")
@@ -77,7 +77,7 @@ public class ImageVideoController {
     @PostMapping("/voice/list")
     @Operation(summary = "查看音色")
     public ApiResponse<Map<String, Object>> listVoices(@Valid @RequestBody ImageVideoVoiceListRequest request) {
-        return ApiResponse.success(imageVideoCozeService.runVoiceList(request.getUserId()));
+        return ApiResponse.success(imageVideoCozeService.runVoiceList(request.getUserId(), request.getName()));
     }
 
     @PostMapping("/voice/delete")
@@ -89,7 +89,7 @@ public class ImageVideoController {
     @PostMapping("/voice/clone")
     @Operation(summary = "音色克隆")
     public ApiResponse<Map<String, Object>> cloneVoice(@Valid @RequestBody ImageVideoVoiceCloneRequest request) {
-        return ApiResponse.success(imageVideoCozeService.runVoiceClone(request.getUserId(), request.getAudioUrl(), request.getVideoUrl()));
+        return ApiResponse.success(imageVideoCozeService.runVoiceClone(request.getUserId(), request.getName(), request.getAudioUrl(), request.getVideoUrl()));
     }
 
     @PostMapping("/voice/synthesis")

backend-java/src/main/java/com/nanri/aiimage/modules/imagevideo/model/dto/DouyinCopyRequest.java

@@ -1,19 +1,58 @@
 package com.nanri.aiimage.modules.imagevideo.model.dto;
 
+import com.fasterxml.jackson.annotation.JsonAlias;
+import com.fasterxml.jackson.annotation.JsonProperty;
 import io.swagger.v3.oas.annotations.media.Schema;
 import jakarta.validation.constraints.NotBlank;
-import jakarta.validation.constraints.NotNull;
 import lombok.Data;
 
+import java.util.ArrayList;
+import java.util.List;
+
 @Data
 @Schema(description = "抖音文案解析和仿写请求")
 public class DouyinCopyRequest {
 
-    @NotNull(message = "userId 不能为空")
     @Schema(description = "用户ID", example = "1")
     private Long userId;
 
+    @Schema(description = "文案工作流业务 api_key")
+    @JsonProperty("api_key")
+    @JsonAlias("apiKey")
+    private String apiKey = "";
+
+    @Schema(description = "T8 平台密钥")
+    @JsonProperty("t8_key")
+    @JsonAlias("t8Key")
+    private String t8Key = "";
+
     @NotBlank(message = "url 不能为空")
     @Schema(description = "抖音分享口令、短链或视频链接", example = "2.38 复制打开抖音，看看")
     private String url;
+
+    @Schema(description = "目标视频时长，单位秒", example = "15")
+    private Integer duration;
+
+    @Schema(description = "产品信息")
+    @JsonProperty("proc_info")
+    @JsonAlias("procInfo")
+    private ProcInfo procInfo = new ProcInfo();
+
+    @Data
+    @Schema(description = "抖音文案仿写产品信息")
+    public static class ProcInfo {
+        @Schema(description = "产品类目", example = "连衣裙")
+        private String type = "";
+
+        @Schema(description = "产品名称", example = "亚麻连衣裙")
+        private String name = "";
+
+        @Schema(description = "产品图片 URL 列表")
+        @JsonProperty("proc_image")
+        @JsonAlias("procImage")
+        private List<String> procImage = new ArrayList<>();
+
+        @Schema(description = "核心卖点或产品属性", example = "优雅、亚麻、白色")
+        private String properties = "";
+    }
 }

backend-java/src/main/java/com/nanri/aiimage/modules/imagevideo/model/dto/ImageVideoVoiceCloneRequest.java

@@ -7,6 +7,7 @@ import lombok.Data;
 public class ImageVideoVoiceCloneRequest {
     @NotNull(message = "userId 不能为空")
     private Long userId;
+    private String name;
     private String audioUrl;
     private String videoUrl;
 }

backend-java/src/main/java/com/nanri/aiimage/modules/imagevideo/model/dto/ImageVideoVoiceListRequest.java

@@ -7,4 +7,6 @@ import lombok.Data;
 public class ImageVideoVoiceListRequest {
     @NotNull(message = "userId 不能为空")
     private Long userId;
+
+    private String name;
 }

backend-java/src/main/java/com/nanri/aiimage/modules/imagevideo/service/ImageVideoCozeService.java

@@ -6,6 +6,7 @@ import com.fasterxml.jackson.core.type.TypeReference;
 import com.nanri.aiimage.common.exception.BusinessException;
 import com.nanri.aiimage.config.ImageVideoProperties;
 import com.nanri.aiimage.modules.file.service.oss.OssStorageService;
+import com.nanri.aiimage.modules.imagevideo.model.dto.DouyinCopyRequest;
 import com.nanri.aiimage.modules.imagevideo.model.vo.ImageVideoMediaUploadVo;
 import com.nanri.aiimage.modules.imagevideo.model.vo.DouyinCopyVo;
 import lombok.RequiredArgsConstructor;
@@ -25,6 +26,7 @@ import java.util.ArrayList;
 import java.util.LinkedHashMap;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
 
 @Slf4j
 @Service
@@ -39,6 +41,9 @@ public class ImageVideoCozeService {
             "scriptDraft", "script_draft", "rewrittenContent", "rewritten_content",
             "fullContent", "full_content", "copywriting", "script", "output", "text", "result"
     );
+    private static final Set<String> IMAGE_VIDEO_MODELS = Set.of(
+            "sdols-2.0", "sdols-2.0-fast", "doubao-seedance-2.0-mini"
+    );
 
     private final ImageVideoProperties properties;
     private final ImageVideoSecretService secretService;
@@ -47,20 +52,23 @@ public class ImageVideoCozeService {
     private final ObjectMapper objectMapper;
     private volatile HttpClient sharedHttpClient;
 
-    public DouyinCopyVo runDouyinCopy(Long userId, String url) {
-        String normalizedUrl = url == null ? "" : url.trim();
+    public DouyinCopyVo runDouyinCopy(DouyinCopyRequest request) {
+        if (request == null) {
+            throw new BusinessException("请求参数不能为空");
+        }
+        String normalizedUrl = request.getUrl() == null ? "" : request.getUrl().trim();
         if (normalizedUrl.isBlank()) {
             throw new BusinessException("抖音口令或链接不能为空");
         }
         if (looksLikeJsonObject(normalizedUrl)) {
-            throw new BusinessException("请粘贴抖音分享口令或链接，不要粘贴接口返回内容");
-        }
-        if (!isLikelyDouyinSource(normalizedUrl)) {
-            throw new BusinessException("请粘贴包含抖音链接的分享口令");
+            throw new BusinessException("请粘贴视频链接或文案来源链接，不要粘贴接口返回内容");
         }
 
-        ImageVideoSecretService.ResolvedSecret secret = secretService.requireValid(userId);
-        String responseText = postWorkflow(normalizedUrl, secret.copyApiKey(), secret.t8Key(), secret.cozeToken(),
+        ImageVideoSecretService.ResolvedSecret secret = secretService.resolveForDouyinCopy(request.getUserId());
+        String responseText = postWorkflow(request, normalizedUrl,
+                firstNonBlank(request.getApiKey(), secret.copyApiKey()),
+                firstNonBlank(request.getT8Key(), secret.t8Key()),
+                secret.cozeToken(),
                 workflowConfigService.douyinCopyWorkflowId());
         return parseDouyinCopyResponse(responseText);
     }
@@ -119,6 +127,32 @@ public class ImageVideoCozeService {
             normalizedProcInfo.put("proc_image", normalizeStringList(normalizedProcInfo.get("proc_image")));
             parameters.put("proc_info", normalizedProcInfo);
         }
+
+        Object videoInfo = parameters.get("video_info");
+        if (videoInfo instanceof Map<?, ?>) {
+            Map<String, Object> normalizedVideoInfo = new LinkedHashMap<>();
+            ((Map<?, ?>) videoInfo).forEach((key, value) -> normalizedVideoInfo.put(String.valueOf(key), value));
+            String model = firstNonBlank(asText(normalizedVideoInfo.get("model")), "sdols-2.0-fast");
+            if (!IMAGE_VIDEO_MODELS.contains(model)) {
+                throw new BusinessException("不支持的视频模型: " + model);
+            }
+            normalizedVideoInfo.put("model", model);
+            normalizedVideoInfo.put("draft", !"false".equalsIgnoreCase(asText(normalizedVideoInfo.get("draft"))));
+            parameters.put("video_info", normalizedVideoInfo);
+        }
+
+        Object audioInfo = parameters.get("audio_info");
+        Map<String, Object> normalizedAudioInfo = new LinkedHashMap<>();
+        if (audioInfo instanceof Map<?, ?>) {
+            ((Map<?, ?>) audioInfo).forEach((key, value) -> normalizedAudioInfo.put(String.valueOf(key), value));
+        }
+        String audioType = firstNonBlank(asText(normalizedAudioInfo.get("type")), "1");
+        if (!"1".equals(audioType) && !"2".equals(audioType)) {
+            throw new BusinessException("不支持的音频类型: " + audioType);
+        }
+        normalizedAudioInfo.put("audio_url", firstNonBlank(asText(normalizedAudioInfo.get("audio_url"))));
+        normalizedAudioInfo.put("type", Integer.valueOf(audioType));
+        parameters.put("audio_info", normalizedAudioInfo);
     }
 
     private List<String> normalizeStringList(Object value) {
@@ -144,15 +178,6 @@ public class ImageVideoCozeService {
         return text.startsWith("{") && text.endsWith("}");
     }
 
-    private boolean isLikelyDouyinSource(String value) {
-        String text = firstNonBlank(value).toLowerCase();
-        return text.contains("douyin.com")
-                || text.contains("v.douyin")
-                || text.contains("iesdouyin")
-                || text.contains("dou音")
-                || text.contains("抖音");
-    }
-
     public Map<String, Object> getImageVideoWorkflowResult(Long userId, String executeId) {
         String resolvedExecuteId = firstNonBlank(executeId);
         if (!hasText(resolvedExecuteId)) {
@@ -210,10 +235,12 @@ public class ImageVideoCozeService {
         }
     }
 
-    public Map<String, Object> runVoiceList(Long userId) {
+    public Map<String, Object> runVoiceList(Long userId, String name) {
         ImageVideoSecretService.ResolvedSecret secret = secretService.requireValid(userId);
         Map<String, Object> parameters = new LinkedHashMap<>();
         parameters.put("api_key", secret.voiceApiKey());
+        parameters.put("name", firstNonBlank(name));
+        parameters.put("user_id", userId);
         return postWorkflowParameters("voice list", workflowConfigService.voiceListWorkflowId(), parameters, secret.cozeToken(), false);
     }
 
@@ -222,19 +249,22 @@ public class ImageVideoCozeService {
         Map<String, Object> parameters = new LinkedHashMap<>();
         parameters.put("api_key", secret.voiceApiKey());
         parameters.put("group_id", secret.voiceGroupId());
+        parameters.put("user_id", userId);
         parameters.put("voice_id", firstNonBlank(voiceId));
         return postWorkflowParameters("voice delete", workflowConfigService.voiceDeleteWorkflowId(), parameters, secret.cozeToken(), false);
     }
 
-    public Map<String, Object> runVoiceClone(Long userId, String audioUrl, String videoUrl) {
-        String resolvedAudioUrl = firstNonBlank(audioUrl);
+    public Map<String, Object> runVoiceClone(Long userId, String name, String audioUrl, String videoUrl) {
         String resolvedVideoUrl = firstNonBlank(videoUrl);
+        String resolvedAudioUrl = firstNonBlank(audioUrl, resolvedVideoUrl);
         if (!hasText(resolvedAudioUrl) && !hasText(resolvedVideoUrl)) {
             throw new BusinessException("请上传音频或填写视频链接");
         }
         ImageVideoSecretService.ResolvedSecret secret = secretService.requireValid(userId);
         Map<String, Object> parameters = new LinkedHashMap<>();
         parameters.put("api_key", secret.voiceApiKey());
+        parameters.put("name", firstNonBlank(name));
+        parameters.put("user_id", userId);
         parameters.put("audio_url", resolvedAudioUrl);
         parameters.put("video_url", resolvedVideoUrl);
         return postWorkflowParameters("voice clone", workflowConfigService.voiceCloneWorkflowId(), parameters, secret.cozeToken(), false);
@@ -250,7 +280,7 @@ public class ImageVideoCozeService {
         return postWorkflowParameters("voice synthesis", workflowConfigService.voiceSynthesisWorkflowId(), parameters, secret.cozeToken(), false);
     }
 
-    private String postWorkflow(String url, String apiKey, String t8Key, String token, String workflowId) {
+    private String postWorkflow(DouyinCopyRequest request, String url, String apiKey, String t8Key, String token, String workflowId) {
         String resolvedApiKey = firstNonBlank(apiKey);
         String resolvedT8Key = firstNonBlank(t8Key);
         String cozeToken = firstNonBlank(token);
@@ -268,6 +298,8 @@ public class ImageVideoCozeService {
         parameters.put("api_key", resolvedApiKey);
         parameters.put("t8_key", resolvedT8Key);
         parameters.put("url", url);
+        parameters.put("duration", normalizeDouyinCopyDuration(request.getDuration()));
+        parameters.put("proc_info", normalizeDouyinCopyProcInfo(request.getProcInfo()));
 
         Map<String, Object> body = new LinkedHashMap<>();
         String resolvedWorkflowId = firstNonBlank(workflowId, properties.getDouyinCopyWorkflowId());
@@ -282,6 +314,30 @@ public class ImageVideoCozeService {
         return postCozeJson("douyin copy", endpoint, body, cozeToken);
     }
 
+    private int normalizeDouyinCopyDuration(Integer duration) {
+        if (duration == null || duration < 0) {
+            return 0;
+        }
+        return duration;
+    }
+
+    private Map<String, Object> normalizeDouyinCopyProcInfo(DouyinCopyRequest.ProcInfo procInfo) {
+        Map<String, Object> normalized = new LinkedHashMap<>();
+        if (procInfo == null) {
+            normalized.put("properties", "");
+            normalized.put("type", "");
+            normalized.put("name", "");
+            normalized.put("proc_image", List.of());
+            return normalized;
+        }
+        normalized.put("properties", firstNonBlank(procInfo.getProperties()));
+        normalized.put("type", firstNonBlank(procInfo.getType()));
+        normalized.put("name", firstNonBlank(procInfo.getName()));
+        List<String> images = normalizeStringList(procInfo.getProcImage());
+        normalized.put("proc_image", images);
+        return normalized;
+    }
+
     private Map<String, Object> postWorkflowParameters(String label, String workflowId, Map<String, Object> parameters,
                                                        String token, boolean async) {
         String cozeToken = firstNonBlank(token);

backend-java/src/main/java/com/nanri/aiimage/modules/imagevideo/service/ImageVideoSecretService.java

@@ -103,6 +103,28 @@ public class ImageVideoSecretService {
         return new ResolvedSecret(cozeToken, copyApiKey, t8Key, voiceApiKey, voiceGroupId);
     }
 
+    public ResolvedSecret resolveForDouyinCopy(Long userId) {
+        String cozeToken = normalize(properties.getCozeToken());
+        if (!hasText(cozeToken)) {
+            throw new BusinessException("图生视频 Coze 访问令牌未配置");
+        }
+        if (userId == null || userId <= 0) {
+            return new ResolvedSecret(cozeToken, "", "", "", "");
+        }
+
+        ImageVideoSecretEntity row = selectByUserId(userId);
+        if (row == null || row.getExpiresAt() == null || !row.getExpiresAt().isAfter(LocalDateTime.now())) {
+            return new ResolvedSecret(cozeToken, "", "", "", "");
+        }
+        return new ResolvedSecret(
+                cozeToken,
+                decrypt(row.getCopyApiKey()),
+                decrypt(row.getT8Key()),
+                decrypt(row.getVoiceApiKey()),
+                decrypt(row.getVoiceGroupId())
+        );
+    }
+
     private ImageVideoSecretEntity selectByUserId(Long userId) {
         return secretMapper.selectOne(new LambdaQueryWrapper<ImageVideoSecretEntity>()
                 .eq(ImageVideoSecretEntity::getUserId, userId)

backend-java/src/main/resources/application.yml

@@ -88,6 +88,7 @@ aiimage:
     endpoint: ${AIIMAGE_OSS_ENDPOINT:oss-cn-hangzhou.aliyuncs.com}
     public-endpoint: ${AIIMAGE_OSS_PUBLIC_ENDPOINT:oss-cn-hangzhou.aliyuncs.com}
     bucket: ${AIIMAGE_OSS_BUCKET:nanri-ai-images}
+    image-video-bucket: ${AIIMAGE_IMAGE_VIDEO_OSS_BUCKET:shufu-video}
     access-key-id: ${AIIMAGE_OSS_ACCESS_KEY_ID:LTAI5tNpyvzMNz9f2dHarsm8}
     access-key-secret: ${AIIMAGE_OSS_ACCESS_KEY_SECRET:bQSZnFH455i8tzyOgeahJmUzwmhynz}
   transient-storage:
@@ -232,10 +233,10 @@ aiimage:
     coze-result-buffer-enabled: ${AIIMAGE_SIMILAR_ASIN_COZE_RESULT_BUFFER_ENABLED:true}
   image-video:
     coze-base-url: ${AIIMAGE_IMAGE_VIDEO_COZE_BASE_URL:https://api.coze.cn}
-    coze-token: ${AIIMAGE_IMAGE_VIDEO_COZE_TOKEN:}
+    coze-token: ${AIIMAGE_IMAGE_VIDEO_COZE_TOKEN:sat_Ws4VB1caOPasDivpKIvtOySYx3lhKgQ95H3crIh0tBwiNYtPTyi6bqe0pBaRzpVu}
     coze-workflow-path: ${AIIMAGE_IMAGE_VIDEO_COZE_WORKFLOW_PATH:/v1/workflow/run}
     coze-connect-timeout-millis: ${AIIMAGE_IMAGE_VIDEO_COZE_CONNECT_TIMEOUT_MILLIS:10000}
-    coze-read-timeout-millis: ${AIIMAGE_IMAGE_VIDEO_COZE_READ_TIMEOUT_MILLIS:600000}
+    coze-read-timeout-millis: ${AIIMAGE_IMAGE_VIDEO_COZE_READ_TIMEOUT_MILLIS:3600000}
   security:
     shop-credential-key: ${AIIMAGE_SHOP_CREDENTIAL_KEY:change-me-shop-credential-key}
     internal-token: ${AIIMAGE_INTERNAL_TOKEN:}

frontend-vue/src/pages/image-video/components/DeliveryVideoWorkspace.vue

@@ -14,11 +14,12 @@
           <AiSectionCard v-if="activeTab === 'remake'" step="01" title="参考视频" description="本地视频或抖音链接，作为复刻依据。"
             class="delivery-card">
             <AiAssetDropzone title="参考视频" description="上传后会先转为 OSS 地址。" placeholder="上传本地视频"
-              helper="支持 MP4 / MOV，也可在下方粘贴抖音/视频链接" accept="video/*" :file-name="currentWorkspace.referenceAsset.fileName"
+              helper="支持 MP4 / MOV，也可在下方粘贴抖音/视频链接" accept="video/*"
+              :file-name="currentWorkspace.referenceAsset.fileName"
               :preview-url="currentWorkspace.referenceAsset.previewUrl"
               :preview-kind="currentWorkspace.referenceAsset.previewKind"
-              :source-url="currentWorkspace.referenceAsset.sourceUrl" :loading="currentWorkspace.referenceAsset.uploading"
-              compact action-label="上传参考视频" selected-label="参考视频"
+              :source-url="currentWorkspace.referenceAsset.sourceUrl"
+              :loading="currentWorkspace.referenceAsset.uploading" compact action-label="上传参考视频" selected-label="参考视频"
               @select="(file) => setAsset(activeTab, 'referenceAsset', file)"
               @clear="clearAsset(activeTab, 'referenceAsset')" />
             <div class="script-learning-panel__toolbar">
@@ -43,6 +44,10 @@
                 <span class="delivery-field__label">输出分辨率</span>
                 <AiChoicePills v-model="currentWorkspace.resolution" :options="resolutionOptions" />
               </div>
+              <div class="delivery-field">
+                <span class="delivery-field__label">模型</span>
+                <AiChoicePills v-model="currentWorkspace.model" :options="modelOptions" />
+              </div>
               <div class="delivery-field">
                 <span class="delivery-field__label">视频时长</span>
                 <div class="duration-control">
@@ -77,15 +82,14 @@
                 placeholder="例如：干净明亮的现代厨房背景，温暖阳光照入。" />
             </div>
             <AiAssetDropzone v-if="currentWorkspace.sceneMode === '上传场景图替换'" title="场景图" description="上传后会先转为 OSS 地址。"
-              placeholder="上传场景图" helper="支持 JPG / PNG" accept="image/*" :file-name="currentWorkspace.sceneAsset.fileName"
-              :preview-url="currentWorkspace.sceneAsset.previewUrl" :preview-kind="currentWorkspace.sceneAsset.previewKind"
-              :source-url="currentWorkspace.sceneAsset.sourceUrl" :loading="currentWorkspace.sceneAsset.uploading" allow-url
-              compact action-label="上传场景图" selected-label="场景图" url-label="场景图链接" url-placeholder="也可以粘贴场景图链接"
+              placeholder="上传场景图" helper="支持 JPG / PNG" accept="image/*"
+              :file-name="currentWorkspace.sceneAsset.fileName" :preview-url="currentWorkspace.sceneAsset.previewUrl"
+              :preview-kind="currentWorkspace.sceneAsset.previewKind"
+              :source-url="currentWorkspace.sceneAsset.sourceUrl" :loading="currentWorkspace.sceneAsset.uploading"
+              allow-url compact action-label="上传场景图" selected-label="场景图" url-label="场景图链接" url-placeholder="也可以粘贴场景图链接"
               @select="(file) => setAsset(activeTab, 'sceneAsset', file)"
-              @url-change="(url) => setAssetUrl(activeTab, 'sceneAsset', url, 'image/*')"
-              preview-clickable
-              @preview="openImagePreview"
-              @clear="clearAsset(activeTab, 'sceneAsset')" />
+              @url-change="(url) => setAssetUrl(activeTab, 'sceneAsset', url, 'image/*')" preview-clickable
+              @preview="openImagePreview" @clear="clearAsset(activeTab, 'sceneAsset')" />
           </AiSectionCard>
         </div>
 
@@ -101,6 +105,10 @@
                 <span class="delivery-field__label">输出分辨率</span>
                 <AiChoicePills v-model="currentWorkspace.resolution" :options="resolutionOptions" />
               </div>
+              <div class="delivery-field">
+                <span class="delivery-field__label">模型</span>
+                <AiChoicePills v-model="currentWorkspace.model" :options="modelOptions" />
+              </div>
               <div class="delivery-field">
                 <span class="delivery-field__label">视频时长</span>
                 <div class="duration-control">
@@ -138,17 +146,18 @@
               <el-input v-model="currentWorkspace.productFocus" type="textarea" :rows="4" resize="none"
                 placeholder="例如：防滑耐磨、复古网面透气设计。" />
             </div>
-            <AiAssetDropzone title="产品图" description="上传后会先转为 OSS 地址，提交时按接口数组格式传入。" placeholder="上传产品图" helper="支持 JPG / PNG"
-              accept="image/*" :file-name="currentWorkspace.productAsset.fileName"
-              :preview-url="currentWorkspace.productAsset.previewUrl"
-              :preview-kind="currentWorkspace.productAsset.previewKind"
-              :source-url="currentWorkspace.productAsset.sourceUrl" :loading="currentWorkspace.productAsset.uploading"
-              allow-url compact preview-clickable action-label="上传产品图" selected-label="产品图" url-label="产品图链接"
-              url-placeholder="也可以粘贴产品图链接"
-              @select="(file) => setAsset(activeTab, 'productAsset', file)"
-              @url-change="(url) => setAssetUrl(activeTab, 'productAsset', url, 'image/*')"
-              @preview="openImagePreview"
-              @clear="clearAsset(activeTab, 'productAsset')" />
+            <div class="product-assets">
+              <AiAssetDropzone v-for="(asset, index) in currentWorkspace.productAssets" :key="index" :title="`产品图 ${index + 1}`"
+                :description="index === 0 ? '最多上传 5 张，提交时按接口数组格式传入。' : ''" placeholder="上传产品图"
+                helper="支持 JPG / PNG" accept="image/*" :file-name="asset.fileName" :preview-url="asset.previewUrl"
+                :preview-kind="asset.previewKind" :source-url="asset.sourceUrl" :loading="asset.uploading" allow-url compact
+                preview-clickable action-label="上传产品图" selected-label="产品图" url-label="产品图链接"
+                url-placeholder="也可以粘贴产品图链接" @select="(file) => setProductAsset(activeTab, index, file)"
+                @url-change="(url) => setProductAssetUrl(activeTab, index, url)" @preview="openImagePreview"
+                @clear="clearProductAsset(activeTab, index)" />
+              <button v-if="currentWorkspace.productAssets.length < 5" type="button" class="script-action-btn product-assets__add"
+                @click="addProductAsset(activeTab)">添加产品图（{{ currentWorkspace.productAssets.length }}/5）</button>
+            </div>
           </AiSectionCard>
 
           <AiSectionCard :step="activeTab === 'remake' ? '05' : '04'" title="人脸" description="性别、模特图或人物类型三选一。"
@@ -163,16 +172,17 @@
             </div>
             <div v-if="currentWorkspace.faceInputType === '人物类型'" class="delivery-field">
               <span class="delivery-field__label">人物类型</span>
-              <el-input v-model="currentWorkspace.modelFigure" clearable placeholder="输入人物类型，例如：甜妹 / 御姐 / 成熟女性 / 职场女性" />
+              <el-input v-model="currentWorkspace.modelFigure" clearable
+                placeholder="输入人物类型，例如：甜妹 / 御姐 / 成熟女性 / 职场女性" />
             </div>
             <AiAssetDropzone v-if="currentWorkspace.faceInputType === '上传模特图'" title="模特图 / 人脸图"
               description="上传后会先转为 OSS 地址。" placeholder="上传清晰正脸图" helper="建议 1080px 以上，单人正面" accept="image/*"
               :file-name="currentWorkspace.modelAsset.fileName" :preview-url="currentWorkspace.modelAsset.previewUrl"
-              :preview-kind="currentWorkspace.modelAsset.previewKind" :source-url="currentWorkspace.modelAsset.sourceUrl"
-              :loading="currentWorkspace.modelAsset.uploading" allow-url compact preview-clickable action-label="上传模特图" selected-label="模特图"
-              url-label="模特图链接" url-placeholder="也可以粘贴模特图链接" @select="(file) => setAsset(activeTab, 'modelAsset', file)"
-              @url-change="(url) => setAssetUrl(activeTab, 'modelAsset', url, 'image/*')"
-              @preview="openImagePreview"
+              :preview-kind="currentWorkspace.modelAsset.previewKind"
+              :source-url="currentWorkspace.modelAsset.sourceUrl" :loading="currentWorkspace.modelAsset.uploading"
+              allow-url compact preview-clickable action-label="上传模特图" selected-label="模特图" url-label="模特图链接"
+              url-placeholder="也可以粘贴模特图链接" @select="(file) => setAsset(activeTab, 'modelAsset', file)"
+              @url-change="(url) => setAssetUrl(activeTab, 'modelAsset', url, 'image/*')" @preview="openImagePreview"
               @clear="clearAsset(activeTab, 'modelAsset')" />
           </AiSectionCard>
 
@@ -194,21 +204,22 @@
               <el-input v-model="currentWorkspace.productFocus" type="textarea" :rows="4" resize="none"
                 placeholder="例如：防滑耐磨、复古网面透气设计。" />
             </div>
-            <AiAssetDropzone title="产品图" description="上传后会先转为 OSS 地址，提交时按接口数组格式传入。" placeholder="上传产品图" helper="支持 JPG / PNG"
-              accept="image/*" :file-name="currentWorkspace.productAsset.fileName"
-              :preview-url="currentWorkspace.productAsset.previewUrl"
-              :preview-kind="currentWorkspace.productAsset.previewKind"
-              :source-url="currentWorkspace.productAsset.sourceUrl" :loading="currentWorkspace.productAsset.uploading"
-              allow-url compact preview-clickable action-label="上传产品图" selected-label="产品图" url-label="产品图链接"
-              url-placeholder="也可以粘贴产品图链接"
-              @select="(file) => setAsset(activeTab, 'productAsset', file)"
-              @url-change="(url) => setAssetUrl(activeTab, 'productAsset', url, 'image/*')"
-              @preview="openImagePreview"
-              @clear="clearAsset(activeTab, 'productAsset')" />
+            <div class="product-assets">
+              <AiAssetDropzone v-for="(asset, index) in currentWorkspace.productAssets" :key="index" :title="`产品图 ${index + 1}`"
+                :description="index === 0 ? '最多上传 5 张，提交时按接口数组格式传入。' : ''" placeholder="上传产品图"
+                helper="支持 JPG / PNG" accept="image/*" :file-name="asset.fileName" :preview-url="asset.previewUrl"
+                :preview-kind="asset.previewKind" :source-url="asset.sourceUrl" :loading="asset.uploading" allow-url compact
+                preview-clickable action-label="上传产品图" selected-label="产品图" url-label="产品图链接"
+                url-placeholder="也可以粘贴产品图链接" @select="(file) => setProductAsset(activeTab, index, file)"
+                @url-change="(url) => setProductAssetUrl(activeTab, index, url)" @preview="openImagePreview"
+                @clear="clearProductAsset(activeTab, index)" />
+              <button v-if="currentWorkspace.productAssets.length < 5" type="button" class="script-action-btn product-assets__add"
+                @click="addProductAsset(activeTab)">添加产品图（{{ currentWorkspace.productAssets.length }}/5）</button>
+            </div>
           </AiSectionCard>
 
-          <AiSectionCard :step="activeTab === 'remake' ? '06' : '05'" title="话术 / 语音"
-            description="编辑口播台词，选择最终配音方式。" class="delivery-card">
+          <AiSectionCard :step="activeTab === 'remake' ? '06' : '05'" title="话术 / 语音" description="编辑口播台词，选择最终配音方式。"
+            class="delivery-card">
             <div class="voice-section">
               <div class="delivery-field">
                 <span class="delivery-field__label">语言</span>
@@ -219,6 +230,11 @@
               <button type="button" class="script-action-btn voice-manage-btn" @click="openVoiceDialog">音色管理</button>
             </div>
 
+            <div class="delivery-field">
+              <span class="delivery-field__label">音频类型</span>
+              <AiChoicePills v-model="currentWorkspace.audioType" :options="audioTypeOptions" />
+            </div>
+
             <div class="delivery-field">
               <span class="delivery-field__label">口播台词</span>
               <el-input v-model="currentWorkspace.scriptText" type="textarea" :rows="6" resize="none"
@@ -247,9 +263,10 @@
               <AiAssetDropzone title="最终配音文件" description="对应 text_info.file_url；只放最终用于视频配音的音频。" placeholder="上传音频文件"
                 helper="支持 MP3 / WAV，也可粘贴可访问的音频链接" accept="audio/*" :file-name="currentWorkspace.speechAsset.fileName"
                 :preview-url="currentWorkspace.speechAsset.previewUrl"
-                :preview-kind="currentWorkspace.speechAsset.previewKind" :source-url="currentWorkspace.speechAsset.sourceUrl"
-                :loading="currentWorkspace.speechAsset.uploading" allow-url compact action-label="上传音频" selected-label="最终配音"
-                url-label="最终配音文件链接" @select="(file) => setAsset(activeTab, 'speechAsset', file)"
+                :preview-kind="currentWorkspace.speechAsset.previewKind"
+                :source-url="currentWorkspace.speechAsset.sourceUrl" :loading="currentWorkspace.speechAsset.uploading"
+                allow-url compact action-label="上传音频" selected-label="最终配音" url-label="最终配音文件链接"
+                @select="(file) => setAsset(activeTab, 'speechAsset', file)"
                 @url-change="(url) => setAssetUrl(activeTab, 'speechAsset', url, 'audio/*')"
                 @clear="clearAsset(activeTab, 'speechAsset')" />
             </template>
@@ -290,6 +307,14 @@
 
       <AiSectionCard :step="activeTab === 'remake' ? '07' : '06'" title="视频结果" description="提交后在这里等待、预览和下载成片。"
         class="delivery-card delivery-card--preview">
+        <el-tabs v-model="currentWorkspace.generationMode" class="generation-mode-tabs">
+          <el-tab-pane label="样片模式" name="draft">
+            <p>快速生成预览视频，用于验证场景结构、镜头调度、主体动作和提示词意图，消耗更少 token。</p>
+          </el-tab-pane>
+          <el-tab-pane label="正常生成" name="standard">
+            <p>关闭样片模式，按正常规格生成视频。</p>
+          </el-tab-pane>
+        </el-tabs>
         <div class="preview-panel__meta">
           <div class="preview-row">
             <span class="preview-row__label">模式</span>
@@ -310,8 +335,8 @@
             :disabled="assemblyBusy || assemblyDownloading" @click="handleAssemblyPrimaryAction">
             {{ assemblyPrimaryActionText }}
           </el-button>
-          <el-button v-if="canResetWorkspace" class="delivery-secondary-btn" :disabled="assemblyBusy || assemblyDownloading"
-            @click="resetCurrentWorkspace">
+          <el-button v-if="canResetWorkspace" class="delivery-secondary-btn"
+            :disabled="assemblyBusy || assemblyDownloading" @click="resetCurrentWorkspace">
             重置内容
           </el-button>
         </div>
@@ -319,7 +344,7 @@
           :class="{ 'assembly-waiting--done': !assemblyBusy && assemblyExecuteId }">
           <div class="assembly-waiting__spinner" aria-hidden="true"></div>
           <div class="assembly-waiting__content">
-            <strong>{{ assemblyPolling ? 'Coze 正在生成视频，请保持页面打开' : '工作流已提交' }}</strong>
+            <strong>{{ assemblyPolling ? '正在生成视频，请保持页面打开' : '工作流已提交' }}</strong>
             <span v-if="assemblyExecuteId">执行 ID：{{ assemblyExecuteId }}</span>
             <span v-if="assemblyPolling">已查询 {{ assemblyPollCount }} 次，系统会每 5 秒自动刷新结果。</span>
           </div>
@@ -372,7 +397,7 @@
         <el-form-item label="音色分组 ID">
           <el-input v-model="secretForm.voiceGroupId" type="password" show-password placeholder="不填写则保留当前有效值" />
           <div v-if="secretStatus?.voiceGroupIdMasked" class="secret-form__hint">当前：{{ secretStatus.voiceGroupIdMasked
-            }}</div>
+          }}</div>
         </el-form-item>
         <el-form-item label="有效期">
           <el-select v-model="secretForm.expireDays" class="delivery-select">
@@ -398,7 +423,8 @@
               <strong>{{ filteredVoiceListItems.length }}</strong>
               <span>/ {{ voiceListItems.length }} 个音色</span>
             </div>
-            <el-input v-model="voiceSearchKeyword" class="voice-search-input" clearable placeholder="搜索音色名称或 voice_id" />
+            <el-input v-model="voiceSearchKeyword" class="voice-search-input" clearable
+              placeholder="搜索音色名称或 voice_id" @keyup.enter="listVoices" @clear="listVoices" />
             <button type="button" class="script-primary-action-btn" :disabled="voiceLoading" @click="listVoices">
               刷新音色
             </button>
@@ -412,18 +438,24 @@
               </div>
               <span class="voice-list__source">{{ item.sourceLabel }}</span>
               <button type="button" class="script-action-btn" @click="useVoice(item)">使用此音色</button>
+              <button v-if="item.sourceKey !== 'system_voice'" type="button" class="script-action-btn script-action-btn--danger"
+                :disabled="voiceLoading" @click="deleteVoice(item)">删除音色</button>
             </div>
           </div>
           <el-empty v-else description="暂无可展示音色，请确认接口已返回音色数据" />
         </el-tab-pane>
 
         <el-tab-pane label="克隆音色" name="clone">
+          <div class="delivery-field">
+            <span class="delivery-field__label">音色名称</span>
+            <el-input v-model="voiceCloneName" clearable placeholder="输入克隆后的音色名称" />
+          </div>
           <AiAssetDropzone title="克隆素材" description="仅用于音色克隆，不会写入 text_info.file_url。" placeholder="上传音频或视频"
             helper="支持 MP3 / WAV / MP4，或粘贴素材链接" accept="audio/*,video/*" :file-name="voiceCloneAsset.fileName"
             :preview-url="voiceCloneAsset.previewUrl" :preview-kind="voiceCloneAsset.previewKind"
             :source-url="voiceCloneAsset.sourceUrl" :loading="voiceCloneAsset.uploading" allow-url compact
-            action-label="上传克隆素材" selected-label="克隆素材" url-label="克隆素材链接"
-            @select="setVoiceCloneAsset" @url-change="setVoiceCloneUrl" @clear="clearVoiceCloneAsset" />
+            action-label="上传克隆素材" selected-label="克隆素材" url-label="克隆素材链接" @select="setVoiceCloneAsset"
+            @url-change="setVoiceCloneUrl" @clear="clearVoiceCloneAsset" />
           <div class="voice-dialog__toolbar">
             <button type="button" class="script-primary-action-btn" :disabled="voiceLoading" @click="cloneVoice">
               开始克隆
@@ -431,16 +463,6 @@
           </div>
         </el-tab-pane>
 
-        <el-tab-pane label="删除音色" name="delete">
-          <div class="delivery-field">
-            <span class="delivery-field__label">待删除 voice_id</span>
-            <el-input v-model="deleteVoiceId" clearable placeholder="输入 voice_id，或从音色列表选择后再删除" />
-          </div>
-          <div class="voice-dialog__toolbar">
-            <button type="button" class="script-action-btn script-action-btn--danger" :disabled="voiceLoading"
-              @click="deleteVoice">确认删除音色</button>
-          </div>
-        </el-tab-pane>
       </el-tabs>
     </el-dialog>
 
@@ -476,7 +498,7 @@ import {
 type WorkspaceTab = 'remake' | 'imageToVideo'
 type PreviewKind = 'image' | 'video' | 'file'
 type MediaType = 'image' | 'video' | 'audio' | 'file'
-type AssetKey = 'referenceAsset' | 'productAsset' | 'sceneAsset' | 'modelAsset' | 'speechAsset'
+type AssetKey = 'referenceAsset' | 'sceneAsset' | 'modelAsset' | 'speechAsset'
 type VoiceMode = 'textOnly' | 'uploadAudio' | 'synthesize'
 type VoiceStatusType = 'info' | 'success' | 'error'
 
@@ -503,12 +525,14 @@ type AssemblyState = {
 
 type WorkspaceState = {
   referenceAsset: AssetState
-  productAsset: AssetState
+  productAssets: AssetState[]
   sceneAsset: AssetState
   modelAsset: AssetState
   speechAsset: AssetState
   ratio: string
   resolution: string
+  model: string
+  generationMode: 'draft' | 'standard'
   duration: string
   durationSeconds: number
   videoPrompt: string
@@ -518,6 +542,7 @@ type WorkspaceState = {
   gender: string
   modelFigure: string
   language: string
+  audioType: string
   voiceMode: VoiceMode
   voiceId: string
   voiceName: string
@@ -546,12 +571,12 @@ const secretDialogVisible = ref(false)
 const secretSaving = ref(false)
 const secretStatus = ref<ImageVideoSecretStatusVo | null>(null)
 const voiceDialogVisible = ref(false)
-const voiceDialogTab = ref<'list' | 'clone' | 'delete'>('list')
+const voiceDialogTab = ref<'list' | 'clone'>('list')
 const voiceStatusMessage = ref('')
 const voiceStatusType = ref<VoiceStatusType>('info')
 const voiceListItems = ref<VoiceListItem[]>([])
 const voiceSearchKeyword = ref('')
-const deleteVoiceId = ref('')
+const voiceCloneName = ref('')
 const imagePreviewVisible = ref(false)
 const imagePreviewUrl = ref('')
 const currentAssembly = computed(() => currentWorkspace.value.assembly)
@@ -572,9 +597,9 @@ const assemblyBusy = computed(() => currentAssembly.value.loading || currentAsse
 const canResetWorkspace = computed(() =>
   Boolean(
     currentAssembly.value.executeId
-      || currentAssembly.value.resultText
-      || currentAssembly.value.videoUrl
-      || currentAssembly.value.debugUrl,
+    || currentAssembly.value.resultText
+    || currentAssembly.value.videoUrl
+    || currentAssembly.value.debugUrl,
   ),
 )
 const assemblyPrimaryActionText = computed(() => {
@@ -594,11 +619,20 @@ const secretForm = reactive({
 
 const ratioOptions = ['9:16', '16:9', '3:4', '4:3', '1:1']
 const resolutionOptions = ['480p', '720p', '1080p']
+const modelOptions = [
+  { label: 'seedance-2.0-mini', value: 'doubao-seedance-2.0-mini' },
+  { label: 'seedance-2.0-fast', value: 'sdols-2.0-fast' },
+  { label: 'seedance-2.0', value: 'sdols-2.0' },
+]
 const durationOptions = ['15', '30', '45', '60']
 const sceneModeOptions = ['提示词微调', '上传场景图替换']
 const faceInputTypeOptions = ['性别选项', '上传模特图', '人物类型']
 const genderOptions = ['女', '男']
 const languageOptions = ['中文', '英语', '韩语', '粤语']
+const audioTypeOptions = [
+  { label: '人物口播', value: '1' },
+  { label: '背景旁白', value: '2' },
+]
 const voiceModeOptions = [
   { label: '只提交文案', value: 'textOnly' },
   { label: '上传/粘贴已有音频', value: 'uploadAudio' },
@@ -634,12 +668,14 @@ function createAssemblyState(): AssemblyState {
 function createWorkspaceState(): WorkspaceState {
   return {
     referenceAsset: createAssetState(),
-    productAsset: createAssetState(),
+    productAssets: [createAssetState()],
     sceneAsset: createAssetState(),
     modelAsset: createAssetState(),
     speechAsset: createAssetState(),
     ratio: '9:16',
     resolution: '720p',
+    model: 'sdols-2.0-fast',
+    generationMode: 'draft',
     duration: '15',
     durationSeconds: 15,
     videoPrompt: '',
@@ -649,6 +685,7 @@ function createWorkspaceState(): WorkspaceState {
     gender: '女',
     modelFigure: '御姐',
     language: '中文',
+    audioType: '1',
     voiceMode: 'textOnly',
     voiceId: '',
     voiceName: '',
@@ -682,8 +719,8 @@ const secretStatusText = computed(() => {
 
 function normalizeDuration(value: number | string) {
   const nextValue = Number(value)
-  if (Number.isFinite(nextValue) && nextValue > 0) return String(Math.round(nextValue))
-  return '15'
+  if (Number.isFinite(nextValue) && nextValue > 0) return Math.round(nextValue)
+  return 15
 }
 
 function setDurationPreset(value: string | string[]) {
@@ -774,6 +811,54 @@ function clearAsset(tab: WorkspaceTab, assetKey: AssetKey) {
   scheduleDeliveryGridLayout()
 }
 
+function addProductAsset(tab: WorkspaceTab) {
+  const productAssets = workspaces[tab].productAssets
+  if (productAssets.length < 5) productAssets.push(createAssetState())
+  scheduleDeliveryGridLayout()
+}
+
+function setProductAssetUrl(tab: WorkspaceTab, index: number, url: string) {
+  const target = workspaces[tab].productAssets[index]
+  if (!target) return
+  const normalizedUrl = (url || '').trim()
+  target.sourceUrl = normalizedUrl
+  target.previewUrl = normalizedUrl
+  target.fileName = normalizedUrl ? '在线素材' : ''
+  target.mediaType = resolveMediaTypeFromUrl(normalizedUrl, 'image/*')
+  target.previewKind = resolvePreviewKindFromMediaType(target.mediaType)
+  scheduleDeliveryGridLayout()
+}
+
+async function setProductAsset(tab: WorkspaceTab, index: number, file: File) {
+  const target = workspaces[tab].productAssets[index]
+  if (!target) return
+  target.uploading = true
+  try {
+    const result = await uploadImageVideoMedia(file)
+    target.fileName = result.originalFilename || file.name
+    target.sourceUrl = result.url
+    target.previewUrl = result.url
+    target.mediaType = (result.mediaType || 'file') as MediaType
+    target.previewKind = resolvePreviewKindFromMediaType(target.mediaType)
+    ElMessage.success('产品图上传成功')
+  } catch (error) {
+    ElMessage.error(error instanceof Error ? error.message : '产品图上传失败')
+  } finally {
+    target.uploading = false
+    scheduleDeliveryGridLayout()
+  }
+}
+
+function clearProductAsset(tab: WorkspaceTab, index: number) {
+  const productAssets = workspaces[tab].productAssets
+  if (productAssets.length === 1) {
+    Object.assign(productAssets[0], createAssetState())
+  } else {
+    productAssets.splice(index, 1)
+  }
+  scheduleDeliveryGridLayout()
+}
+
 function openImagePreview(url: string) {
   const normalizedUrl = (url || '').trim()
   if (!normalizedUrl) return
@@ -867,17 +952,22 @@ async function rewriteScriptFromSource() {
     return
   }
   if (looksLikeJsonObject(url)) {
-    ElMessage.warning('请粘贴抖音分享口令或链接，不要粘贴接口返回内容')
-    return
-  }
-  if (!isLikelyDouyinSource(url)) {
-    ElMessage.warning('请粘贴包含抖音链接的分享口令')
+    ElMessage.warning('请粘贴视频链接或文案来源链接，不要粘贴接口返回内容')
     return
   }
   if (activeTab.value === 'remake') currentWorkspace.value.scriptSourceUrl = url
   copyLearningLoading.value = true
   try {
-    const result = await runImageVideoDouyinCopy(url)
+    const result = await runImageVideoDouyinCopy({
+      url,
+      duration: Math.max(0, Math.round(Number(currentWorkspace.value.durationSeconds) || 0)),
+      proc_info: {
+        type: currentWorkspace.value.categoryName.trim(),
+        name: currentWorkspace.value.productName.trim(),
+        proc_image: resolveProductImageUrls(currentWorkspace.value),
+        properties: currentWorkspace.value.productFocus.trim(),
+      },
+    })
     const text = cleanCopyText(result.scriptDraft || result.recognizedContent || '')
     currentWorkspace.value.scriptText = text
     ElMessage.success('文案已提取')
@@ -897,15 +987,6 @@ function looksLikeJsonObject(value: string) {
   return text.startsWith('{') && text.endsWith('}')
 }
 
-function isLikelyDouyinSource(value: string) {
-  const text = value.trim().toLowerCase()
-  return text.includes('douyin.com')
-    || text.includes('v.douyin')
-    || text.includes('iesdouyin')
-    || text.includes('dou音')
-    || text.includes('抖音')
-}
-
 function cleanCopyText(value: string) {
   return value
     .replace(/^\s*https?:\/\/\S+\s*$/gm, '')
@@ -916,8 +997,7 @@ function cleanCopyText(value: string) {
 }
 
 function resolveProductImageUrls(workspace: WorkspaceState) {
-  const productImageUrl = resolveAssetUrl(workspace.productAsset)
-  return productImageUrl ? [productImageUrl] : []
+  return workspace.productAssets.map(resolveAssetUrl).filter(Boolean).slice(0, 5)
 }
 
 function resolveSpeechFileUrl(workspace: WorkspaceState) {
@@ -954,27 +1034,8 @@ function resolveFaceInfo(workspace: WorkspaceState): ImageVideoWorkflowParameter
   return { type: 1, model_figure: workspace.gender, model_image: [] }
 }
 
-function validateWorkspace(workspace: WorkspaceState) {
-  const productImages = resolveProductImageUrls(workspace)
-  const referenceVideo = resolveAssetUrl(workspace.referenceAsset)
-  if (activeTab.value === 'remake' && !referenceVideo) return '请先上传或填写参考视频'
-  if (!productImages.length) return '请先上传或填写产品图'
-  if (!workspace.categoryName.trim()) return '请先输入产品类目'
-  if (!workspace.productName.trim()) return '请先输入产品名称'
-  if (!workspace.productFocus.trim()) return '请先输入核心卖点/产品属性'
-  if (workspace.sceneMode === '上传场景图替换' && !resolveAssetUrl(workspace.sceneAsset)) return '请先上传或填写场景图'
-  if (workspace.faceInputType === '上传模特图' && !resolveAssetUrl(workspace.modelAsset)) return '请先上传或填写模特图'
-  return ''
-}
-
 function buildImageVideoWorkflowParameters(): ImageVideoWorkflowParameters | null {
   const workspace = currentWorkspace.value
-  const error = validateWorkspace(workspace)
-  if (error) {
-    ElMessage.warning(error)
-    return null
-  }
-
   const referenceVideoInfo = resolveReferenceVideoInfo(workspace)
   const productImageUrls = resolveProductImageUrls(workspace)
   return {
@@ -1000,12 +1061,17 @@ function buildImageVideoWorkflowParameters(): ImageVideoWorkflowParameters | nul
       text: workspace.scriptText.trim(),
       file_url: resolveSpeechFileUrl(workspace),
     },
+    audio_info: {
+      audio_url: resolveSpeechFileUrl(workspace),
+      type: Number(workspace.audioType) || 1,
+    },
     video_info: {
       video_url: referenceVideoInfo.videoUrl,
       share_url: referenceVideoInfo.shareUrl,
       ref_video_mode: referenceVideoInfo.refVideoMode,
       mode: activeTab.value === 'imageToVideo' ? '1' : '2',
-      model: 'sdols-2.0-fast',
+      draft: workspace.generationMode === 'draft',
+      model: workspace.model,
       prompt: workspace.videoPrompt.trim(),
       ratio: workspace.ratio,
       resolution: workspace.resolution,
@@ -1140,7 +1206,7 @@ async function pollAssemblyResult(tab: WorkspaceTab, executeId: string) {
     return
   }
 
-  if (assembly.pollCount >= 120) {
+  if (assembly.pollCount >= 720) {
     assembly.polling = false
     ElMessage.warning('Coze 执行结果查询超时，请稍后通过 execute_id 查看')
     return
@@ -1213,10 +1279,10 @@ function collectVoiceArrays(value: unknown): unknown[] {
       return
     }
     const record = node as Record<string, unknown>
-    ;['system_voice', 'voice_cloning', 'voice_generation', 'music_generation'].forEach((key) => {
-      const item = record[key]
-      if (Array.isArray(item)) arrays.push(...item)
-    })
+      ;['system_voice', 'voice_cloning', 'voice_generation', 'music_generation'].forEach((key) => {
+        const item = record[key]
+        if (Array.isArray(item)) arrays.push(...item)
+      })
     Object.values(record).forEach(collect)
   }
 
@@ -1456,9 +1522,9 @@ function extractVoiceItems(value: unknown): VoiceListItem[] {
     sourceConfigs.forEach((config) => {
       const direct = record[config.key]
       const nested = getRecordValue(record, ['data', config.key])
-      ;[direct, nested].forEach((source) => {
-        if (Array.isArray(source)) source.forEach((item) => addVoice(item, config.key, config.label))
-      })
+        ;[direct, nested].forEach((source) => {
+          if (Array.isArray(source)) source.forEach((item) => addVoice(item, config.key, config.label))
+        })
     })
     Object.values(record).forEach(collectFromKnownArrays)
   }
@@ -1482,7 +1548,7 @@ function firstString(...values: unknown[]) {
 async function listVoices() {
   voiceLoading.value = true
   try {
-    const result = await listImageVideoVoices()
+    const result = await listImageVideoVoices(voiceSearchKeyword.value.trim())
     voiceListItems.value = extractVoiceItems(result)
     setVoiceStatus(
       voiceListItems.value.length ? `已获取 ${voiceListItems.value.length} 个音色` : '音色接口已返回，未解析到列表项',
@@ -1497,12 +1563,12 @@ async function listVoices() {
   }
 }
 
-async function deleteVoice() {
-  const voiceId = deleteVoiceId.value.trim() || currentWorkspace.value.voiceId.trim()
-  if (!voiceId) {
-    ElMessage.warning('请先输入 voice_id')
+async function deleteVoice(item: VoiceListItem) {
+  if (item.sourceKey === 'system_voice') {
+    ElMessage.warning('系统音色不支持删除')
     return
   }
+  const voiceId = item.voiceId
   try {
     await ElMessageBox.confirm(`确定删除音色 ${voiceId} 吗？`, '删除音色', {
       confirmButtonText: '删除',
@@ -1528,6 +1594,11 @@ async function deleteVoice() {
 }
 
 async function cloneVoice() {
+  const name = voiceCloneName.value.trim()
+  if (!name) {
+    ElMessage.warning('请输入音色名称')
+    return
+  }
   const sourceUrl = resolveAssetUrl(voiceCloneAsset)
   if (!sourceUrl) {
     ElMessage.warning('请先上传或填写音频/视频链接')
@@ -1536,10 +1607,11 @@ async function cloneVoice() {
   voiceLoading.value = true
   try {
     const clonePayload = {
-      audioUrl: voiceCloneAsset.mediaType === 'video' ? '' : sourceUrl,
+      audioUrl: sourceUrl,
       videoUrl: voiceCloneAsset.mediaType === 'video' ? sourceUrl : '',
     }
     const result = await cloneImageVideoVoice({
+      name,
       audioUrl: clonePayload.audioUrl,
       videoUrl: clonePayload.videoUrl,
     })
@@ -1598,7 +1670,6 @@ async function synthesizeVoice() {
 function useVoice(item: VoiceListItem) {
   currentWorkspace.value.voiceId = item.voiceId
   currentWorkspace.value.voiceName = item.name || item.voiceId
-  deleteVoiceId.value = item.voiceId
   currentWorkspace.value.voiceMode = 'synthesize'
   setVoiceStatus(`已选择音色：${currentWorkspace.value.voiceName}（${item.voiceId}）`, 'success')
   voiceDialogVisible.value = false
@@ -1856,6 +1927,15 @@ onBeforeUnmount(() => {
   gap: 8px;
 }
 
+.product-assets {
+  display: grid;
+  gap: 12px;
+}
+
+.product-assets__add {
+  justify-self: start;
+}
+
 .scene-face-grid {
   display: grid;
   grid-template-columns: repeat(2, minmax(0, 1fr));
@@ -1956,6 +2036,10 @@ onBeforeUnmount(() => {
   padding-right: 4px;
 }
 
+.voice-dialog__tabs :deep(.el-tabs__item:not(.is-active)) {
+  color: #fff;
+}
+
 .voice-dialog__toolbar {
   display: flex;
   justify-content: flex-end;
@@ -1998,7 +2082,7 @@ onBeforeUnmount(() => {
 
 .voice-list__item {
   display: grid;
-  grid-template-columns: minmax(0, 1fr) auto auto;
+  grid-template-columns: minmax(0, 1fr) auto auto auto;
   gap: 10px;
   align-items: center;
   padding: 10px 12px;
@@ -2049,6 +2133,29 @@ onBeforeUnmount(() => {
   background: #141414;
 }
 
+.generation-mode-tabs :deep(.el-tabs__header) {
+  margin-bottom: 10px;
+}
+
+.generation-mode-tabs :deep(.el-tabs__item) {
+  color: #f2f2f2;
+  font-weight: 700;
+}
+
+.generation-mode-tabs :deep(.el-tabs__active-bar) {
+  background-color: #b99455;
+}
+
+.generation-mode-tabs :deep(.el-tabs__content) {
+  color: #a8a8a8;
+  font-size: 12px;
+  line-height: 1.5;
+}
+
+.generation-mode-tabs p {
+  margin: 0;
+}
+
 .preview-row {
   display: flex;
   justify-content: space-between;

frontend-vue/src/shared/api/java-modules.ts

@@ -2483,6 +2483,19 @@ export interface ImageVideoDouyinCopyVo {
   debugUrl?: string;
 }
 
+export interface ImageVideoDouyinCopyPayload {
+  url: string;
+  api_key?: string;
+  t8_key?: string;
+  duration?: number;
+  proc_info?: {
+    type: string;
+    name: string;
+    proc_image: string[];
+    properties: string;
+  };
+}
+
 export interface ImageVideoSecretStatusVo {
   userId?: number;
   configured?: boolean;
@@ -2535,16 +2548,21 @@ export interface ImageVideoWorkflowParameters {
     text: string;
     file_url: string;
   };
+  audio_info: {
+    audio_url: string;
+    type: number;
+  };
   video_info: {
     video_url: string;
     share_url: string;
     ref_video_mode: string;
     mode: string;
+    draft: boolean;
     model: string;
     prompt: string;
     ratio: string;
     resolution: string;
-    duration: string;
+    duration: number;
   };
 }
 
@@ -2601,11 +2619,11 @@ export function saveImageVideoSecrets(payload: ImageVideoSecretSavePayload) {
   );
 }
 
-export function runImageVideoDouyinCopy(url: string) {
+export function runImageVideoDouyinCopy(payload: ImageVideoDouyinCopyPayload) {
   return unwrapJavaResponse(
-    post<JavaApiResponse<ImageVideoDouyinCopyVo>, { userId: number; url: string }>(
+    post<JavaApiResponse<ImageVideoDouyinCopyVo>, ImageVideoDouyinCopyPayload & { userId: number }>(
       `${JAVA_API_PREFIX}/image-video/douyin-copy`,
-      { userId: getCurrentUserId(), url },
+      { userId: getCurrentUserId(), ...payload },
       { timeout: 180000 },
     ),
   );
@@ -2616,7 +2634,7 @@ export function runImageVideoWorkflow(parameters: ImageVideoWorkflowParameters)
     post<JavaApiResponse<ImageVideoWorkflowResponse>, ImageVideoWorkflowRunRequest>(
       `${JAVA_API_PREFIX}/image-video/workflow/run`,
       { userId: getCurrentUserId(), parameters },
-      { timeout: 120000 },
+      { timeout: 3600000 },
     ),
   );
 }
@@ -2626,7 +2644,7 @@ export function getImageVideoWorkflowResult(executeId: string) {
     post<JavaApiResponse<ImageVideoWorkflowResponse>, ImageVideoWorkflowResultRequest>(
       `${JAVA_API_PREFIX}/image-video/workflow/result`,
       { userId: getCurrentUserId(), executeId },
-      { timeout: 600000 },
+      { timeout: 3600000 },
     ),
   );
 }
@@ -2660,11 +2678,11 @@ function resolveImageVideoMediaType(file: File) {
   return "file";
 }
 
-export function listImageVideoVoices() {
+export function listImageVideoVoices(name = "") {
   return unwrapJavaResponse(
-    post<JavaApiResponse<ImageVideoVoiceWorkflowResponse>, { userId: number }>(
+    post<JavaApiResponse<ImageVideoVoiceWorkflowResponse>, { userId: number; name: string }>(
       `${JAVA_API_PREFIX}/image-video/voice/list`,
-      { userId: getCurrentUserId() },
+      { userId: getCurrentUserId(), name },
       { timeout: 180000 },
     ),
   );
@@ -2680,9 +2698,9 @@ export function deleteImageVideoVoice(voiceId: string) {
   );
 }
 
-export function cloneImageVideoVoice(payload: { audioUrl?: string; videoUrl?: string }) {
+export function cloneImageVideoVoice(payload: { name?: string; audioUrl?: string; videoUrl?: string }) {
   return unwrapJavaResponse(
-    post<JavaApiResponse<ImageVideoVoiceWorkflowResponse>, { userId: number; audioUrl?: string; videoUrl?: string }>(
+    post<JavaApiResponse<ImageVideoVoiceWorkflowResponse>, { userId: number; name?: string; audioUrl?: string; videoUrl?: string }>(
       `${JAVA_API_PREFIX}/image-video/voice/clone`,
       { userId: getCurrentUserId(), ...payload },
       { timeout: 180000 },