@imgly/plugin-ai-image-generation-web 0.2.8 → 0.2.10

package/CHANGELOG.md

@@ -2,6 +2,49 @@
 
 ## [Unreleased]
 
+## [0.2.10] - 2025-10-22
+
+### New Features
+
+-   [video-generation] **Google Veo 3.1 Provider Suite**: Added comprehensive Google Veo 3.1 video generation providers via fal.ai, offering multiple models optimized for different use cases:
+    - **Text-to-Video Providers**:
+        - `Veo31TextToVideo`: Standard quality text-to-video with configurable aspect ratios (16:9, 9:16, 1:1), variable duration (4s, 6s, 8s), resolution options (720p, 1080p), and optional audio generation
+        - `Veo31FastTextToVideo`: Faster and more cost-effective text-to-video variant with same capabilities as standard version
+    - **Image-to-Video Providers**:
+        - `Veo31ImageToVideo`: Standard quality image-to-video with auto aspect ratio detection, multiple preset options (16:9, 9:16, 1:1), resolution options (720p, 1080p), fixed 8-second duration, and optional audio generation
+        - `Veo31FastImageToVideo`: Faster and more cost-effective image-to-video variant with same capabilities
+    - **First-Last Frame Providers** (experimental dual-image transformation):
+        - `Veo31FirstLastFrameToVideo`: Standard quality interpolation between two images (first and last frame)
+        - `Veo31FastFirstLastFrameToVideo`: Faster variant with dual image input UI, multiple aspect ratios (16:9, 9:16, 1:1, 4:3, 3:4), resolution options (480p, 720p, 1080p), adjustable duration (2-8 seconds), optional prompt guidance, and optional audio generation
+    - **Quick Action Support**: Added "Animate Between Images" quick action for creating smooth transitions between two selected images using the first-last-frame providers
+
+## [0.2.9] - 2025-10-16
+
+### New Features
+
+-   [image-generation] **GeminiFlash25 Provider**: Added Google Gemini Flash 2.5 text-to-image provider via fal.ai with fast generation times, multiple aspect ratios (1:1, 3:4, 4:3, 9:16, 16:9), custom dimensions support, and multiple output formats (JPEG, PNG, WEBP)
+-   [image-generation] **Gemini25FlashImageEdit Provider**: Added Google Gemini 2.5 Flash Image Edit provider via fal.ai for advanced image editing with multi-image support (1-10 images), comprehensive quick actions support (editImage, swapBackground, styleTransfer, artistTransfer, createVariant, combineImages, remixPage, remixPageWithPrompt), text-based editing instructions, and fast processing times
+
+### Improvements
+
+-   [generation-web] **Middleware preventDefault() API**: Added `options.preventDefault()` method to suppress default UI feedback (notifications, block states, console logging) when handling errors in custom middleware
+-   [all] **Internationalization Support**: All hardcoded strings across AI plugins have been removed and replaced with translation keys, enabling full localization support for plugin labels, actions, styles, and error messages
+-   [all] **Translation Keys Available**: Added comprehensive translation keys for:
+    -   Panel and dock labels (AI Image, AI Video, AI Sticker, AI Voice, Sound Generation)
+    -   Action labels (Generate Image, Generate Video, Generate Sticker)
+    -   Style transfer options (None, Anime, Cyberpunk, Kodak 400, Watercolor, Dark Fantasy, Vaporwave, Vector Flat, 3D Animation, Ukiyo-e, Surreal, Steampunk, Night Bokeh, Pop Art)
+    -   Error messages and UI elements
+-   [all] **Backwards Compatibility**: Translation system automatically detects CE.SDK version and gracefully falls back to English strings for CE.SDK versions < 1.59.0, ensuring no breaking changes for existing integrations
+
+### Fixed
+
+-   [generation-web] **Placeholder Block Error State**: Fixed placeholder blocks getting stuck in Pending state when generation fails or is aborted. Blocks are now properly destroyed when generation is aborted, or moved to Error state when generation fails, preventing perpetual loading spinners in the UI.
+-   [generation-web] **Middleware Block Targeting**: Fixed middleware to correctly receive block IDs for placeholder blocks and quick action targets. Previously, middleware would fall back to `findAllSelected()` which could target incorrect blocks if the selection changed during generation. Now placeholder blocks created during panel generation and target blocks from quick actions are explicitly passed to middleware, ensuring operations like pending state, locking, and highlighting affect the correct blocks.
+
+### Changed
+
+-   [generation-web] **BlockIds Type Refinement**: Removed unused `| null` type from `blockIds` parameter in `GenerationOptions` and `Generate` function signature. The `null` value was documented but never implemented or used. Use an empty array `[]` instead of `null` to explicitly target no blocks.
+
 ## [0.2.8] - 2025-09-29
 
 ### New Features

package/README.md

@@ -457,7 +457,62 @@ Key features:
 - Automatic prompt optimization
 - Custom headers support for API requests
 
-#### 4. GeminiFlashEdit (Image-to-Image)
+#### 4. GeminiFlash25 (Text-to-Image)
+
+A fast and efficient text-to-image model from Google's Gemini Flash 2.5 via fal.ai:
+
+```typescript
+text2image: FalAiImage.GeminiFlash25({
+  proxyUrl: 'http://your-proxy-server.com/api/proxy',
+  headers: {
+    'x-custom-header': 'value',
+    'x-client-version': '1.0.0'
+  },
+  // Optional: Configure default property values
+  properties: {
+    aspect_ratio: { default: '1:1' },  // Options: '1:1', '3:4', '4:3', '9:16', '16:9'
+    output_format: { default: 'jpeg' }  // Options: 'jpeg', 'png', 'webp'
+  }
+})
+```
+
+Key features:
+- Fast generation times with Google's Gemini Flash 2.5 model
+- Multiple aspect ratios: 1:1, 3:4, 4:3, 9:16, 16:9
+- Custom dimensions support
+- Multiple output formats (JPEG, PNG, WEBP)
+- Configurable number of images (1-4)
+- Custom headers support for API requests
+
+**Available Property Values:**
+
+```typescript
+// aspect_ratio options:
+'1:1'   // Square (1024×1024)
+'3:4'   // Portrait (768×1024)
+'4:3'   // Landscape (1024×768)
+'9:16'  // Tall Portrait (576×1024)
+'16:9'  // Wide Landscape (1024×576)
+
+// output_format options:
+'jpeg'  // JPEG format (default)
+'png'   // PNG format
+'webp'  // WebP format
+```
+
+**Custom Translations:**
+```typescript
+cesdk.i18n.setTranslations({
+  en: {
+    'ly.img.plugin-ai-image-generation-web.fal-ai/gemini-flash-2.5.property.prompt': 'Describe your image',
+    'ly.img.plugin-ai-image-generation-web.fal-ai/gemini-flash-2.5.property.aspect_ratio': 'Aspect Ratio',
+    'ly.img.plugin-ai-image-generation-web.fal-ai/gemini-flash-2.5.property.aspect_ratio.1:1': 'Square (1:1)',
+    'ly.img.plugin-ai-image-generation-web.fal-ai/gemini-flash-2.5.property.aspect_ratio.16:9': 'Wide (16:9)'
+  }
+});
+```
+
+#### 5. GeminiFlashEdit (Image-to-Image)
 
 An image modification model from fal.ai that transforms existing images:
 
@@ -479,9 +534,10 @@ image2image: FalAiImage.GeminiFlashEdit({
 
 Key features:
 - Transform existing images with text prompts
-- Available directly through canvas quick actions
+- Comprehensive quick actions support: edit image, swap background, style transfer, artist styles, create variants, upscale, outpaint, and combine images
 - Maintains original image dimensions
 - Includes style presets and artist-specific transformations
+- Supports multi-image inputs via combineImages quick action
 - Custom headers support for API requests
 
 **Custom Translations:**
@@ -921,24 +977,75 @@ Custom middleware functions follow this pattern:
 const customMiddleware = async (input, options, next) => {
   // Pre-processing logic
   console.log('Before generation:', input);
-  
+
   // Add custom fields or modify the input if needed
   const modifiedInput = {
     ...input,
     customField: 'custom value'
   };
-  
+
   // Call the next middleware or generation function
   const result = await next(modifiedInput, options);
-  
+
   // Post-processing logic
   console.log('After generation:', result);
-  
+
   // You can also modify the result before returning it
   return result;
 };
 ```
 
+#### Preventing Default Feedback
+
+Middleware can suppress default UI feedback behaviors using `options.preventDefault()`:
+
+```typescript
+const customErrorMiddleware = async (input, options, next) => {
+  try {
+    return await next(input, options);
+  } catch (error) {
+    // Prevent default error notification and block error state
+    options.preventDefault();
+
+    // When preventDefault() is called, you need to handle the block state yourself.
+    // Here we set the error state to mimic the default behavior:
+    options.blockIds?.forEach(blockId => {
+      if (options.engine.block.isValid(blockId)) {
+        options.engine.block.setState(blockId, { type: 'Error', error: 'Unknown' });
+      }
+    });
+    // Alternative: Delete the placeholder block instead
+    // options.blockIds?.forEach(blockId => {
+    //   if (options.engine.block.isValid(blockId)) {
+    //     options.engine.block.destroy(blockId);
+    //   }
+    // });
+
+    // Show custom error notification
+    options.cesdk?.ui.showNotification({
+      type: 'error',
+      message: `Custom error: ${error.message}`,
+      action: {
+        label: 'Contact Support',
+        onClick: () => window.open('mailto:support@example.com')
+      }
+    });
+
+    throw error;
+  }
+};
+```
+
+**What gets prevented:**
+- Error/success notifications
+- Block error state
+- Console error logging
+
+**What is NOT prevented:**
+- Pending → Ready transition (loading spinner always stops)
+
+For more details, see the [@imgly/plugin-ai-generation-web documentation](https://github.com/imgly/plugins/tree/main/packages/plugin-ai-generation-web#preventing-default-feedback).
+
 The middleware function signature is:
 
 ```typescript
@@ -1061,6 +1168,16 @@ FalAiImage.IdeogramV3Remix(config: {
 })
 ```
 
+#### GeminiFlash25
+
+```typescript
+FalAiImage.GeminiFlash25(config: {
+  proxyUrl: string;
+  headers?: Record<string, string>;
+  debug?: boolean;
+})
+```
+
 #### GeminiFlashEdit
 
 ```typescript
@@ -1071,6 +1188,16 @@ FalAiImage.GeminiFlashEdit(config: {
 })
 ```
 
+#### Gemini25FlashImageEdit
+
+```typescript
+FalAiImage.Gemini25FlashImageEdit(config: {
+  proxyUrl: string;
+  headers?: Record<string, string>;
+  debug?: boolean;
+})
+```
+
 #### QwenImageEdit
 
 ```typescript
@@ -1248,6 +1375,7 @@ const myImageProvider = {
   - Recraft20b: `ly.img.ai.fal-ai/recraft/v2/text-to-image`
   - IdeogramV3: `ly.img.ai.fal-ai/ideogram/v3`
   - IdeogramV3Remix: `ly.img.ai.fal-ai/ideogram/v3/remix`
+  - GeminiFlash25: `ly.img.ai.fal-ai/gemini-flash-2.5`
   - GeminiFlashEdit: `ly.img.ai.fal-ai/gemini-flash-edit`
   - QwenImageEdit: `ly.img.ai.fal-ai/qwen-image-edit`
   - GptImage1.Text2Image: `ly.img.ai.open-ai/gpt-image-1/text2image`
@@ -1266,6 +1394,7 @@ Generated images are automatically stored in asset sources with the following ID
 - Recraft20b: `fal-ai/recraft/v2/text-to-image.history`
 - IdeogramV3: `fal-ai/ideogram/v3.history`
 - IdeogramV3Remix: `fal-ai/ideogram/v3/remix.history`
+- GeminiFlash25: `fal-ai/gemini-flash-2.5.history`
 - GeminiFlashEdit: `fal-ai/gemini-flash-edit.history`
 - QwenImageEdit: `fal-ai/qwen-image-edit.history`
 - GptImage1.Text2Image: `open-ai/gpt-image-1/text2image.history`