makaron-cli 0.15.4 → 0.16.0

package/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "makaron-cli",
-  "version": "0.15.4",
+  "version": "0.16.0",
   "description": "Give Claude Code a creative agent. Pass complete creative requests and source media to Makaron Chat.",
   "author": {
     "name": "Versa AI",

package/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "makaron-cli",
-  "version": "0.15.4",
+  "version": "0.16.0",
   "description": "Give Codex a creative agent. Pass complete creative requests and source media to Makaron Chat.",
   "author": {
     "name": "Versa AI",

package/README.md

@@ -384,7 +384,7 @@ npx makaron-cli edit --image photo.jpg --out result.jpg "make it dramatic"
 npx makaron-cli edit --image-model gpt-image-2.5-flare --background transparent --out sticker.png "a magenta star sticker"
 ```
 
-Options: `--image`, `--image-model gemini|gemini-lite|qwen-spicy|openai|gpt-image-2.5-flare|gpt-image-2.5-sunburst|wan2.7-image`, `--ref <file>` (up to 3), `--aspect <ratio>`, `--background auto|opaque|transparent`, `--out <path>`. Qwen Spicy supports 0–3 input images; legacy `qwen` requests map to it. Pony and WAI are retired. Transparent output routes strictly to GPT Image 2.5 Flare and is returned only when the provider supplies real PNG/WebP alpha.
+Options: `--image`, `--image-model gemini|gemini-2.1|gemini-lite|qwen-spicy|openai|gpt-image-2.5-flare|gpt-image-2.5-sunburst|wan2.7-image`, `--ref <file>` (model-specific limit), `--aspect <ratio>`, `--background auto|opaque|transparent`, `--out <path>`. Nano Banana 2.1 (`gemini-2.1`) supports up to 14 total input images and `--image-resolution 1K|2K|4K` (default 1K); it uses OpenRouter and does not retry or switch models automatically. Qwen Spicy supports 0–3 input images; legacy `qwen` requests map to it. Pony and WAI are retired. Transparent output routes strictly to GPT Image 2.5 Flare and is returned only when the provider supplies real PNG/WebP alpha.
 
 `wan2.7-image` uses Alibaba international for fast, approximately 1K generation and editing (default 6 credits/image). Failed or timed-out Wan requests are not automatically retried or switched to another model. Face identity can change. Example: `makaron edit --image portrait.jpg --image-model wan2.7-image --aspect 16:9 --out stadium.jpg "Place this woman in a baseball stadium, preserving her face."`
 
@@ -429,7 +429,7 @@ Options for `video create`: `--script "..."`, `--script-file <path>`, `--image <
 
 fal H3 Turbo: use `--video-model minimax-h3-max` for faster-than-real-time generation. It supports exactly 5/10/15 seconds at 480p/768p and defaults to native 768p, with no image for T2V or exactly one `--image` for I2V. It does not accept reference video/audio or multiple images.
 
-Seedance 2.5: use `--video-model seedance-2.5` for 4-30 second output at 480p/720p. `--image` accepts local files or URLs (up to 30), while repeatable `--video` and `--audio` accept up to 10 each. Use `--video-operation generate|edit|extend`, `--extend-direction forward|backward`, `--output-format mp4|mov`, `--web-search`, `--generated-audio` / `--no-generated-audio`, and `--relaxed-content-filter`. Edit/extend require a video reference. The Evolink route does not currently expose 4K output.
+Seedance 2.5: use `--video-model seedance-2.5` or `seedance-2.5-eco` for 4-30 second videos generated at 480p then automatically upscaled to 720p, 1080p (default), 2K or 4K. Use `seedance-2.5-native` explicitly for native output at 480p (default) or 720p. `--image` accepts local files or URLs (up to 30), while repeatable `--video` and `--audio` accept up to 10 each. Use `--video-operation generate|edit|extend`, `--extend-direction forward|backward`, `--output-format mp4|mov`, `--web-search`, `--generated-audio` / `--no-generated-audio`, and `--relaxed-content-filter`. Edit/extend require a video reference. Eco delivers MP4; MOV is native-only. The native Evolink route does not expose 4K output.
 
 Wan 3.0: choose `--video-model wan-3.0` or the faster `--video-model wan-3.0-prime`. Both support 2-30 second generation, 480p/720p/1080p/2K/4K, and up to 10 images, 5 videos, and 5 audio references. Pass `--video-resolution 2k|4k` to use the matching FlashVSR/Pro endpoint automatically; Pro is not a separate model selector. Use generation mode with feature references; typed edit/extend and `--relaxed-content-filter` are not supported.
 

package/bin/makaron.mjs

@@ -1935,9 +1935,11 @@ Usage:
 
 Options:
   --image <file|url>        Base image to edit. Omit for text-to-image.
-  --ref <file|url>          Additional reference image. Repeatable, up to 3.
-  --image-model <id>        gemini, gemini-lite, qwen-spicy, openai, gpt-image-2.5-flare, gpt-image-2.5-sunburst, or wan2.7-image. Legacy qwen maps to qwen-spicy.
+  --ref <file|url>          Additional reference image. Repeatable; model-specific limit (Nano Banana 2.1: 14 total inputs).
+  --image-model <id>        gemini, gemini-2.1, gemini-lite, qwen-spicy, openai, gpt-image-2.5-flare, gpt-image-2.5-sunburst, or wan2.7-image. Legacy qwen maps to qwen-spicy.
   --skill <id>              enhance, creative, wild, or captions.
+  --image-resolution <id>  1K, 2K, or 4K; Auto chooses Nano Banana 2.1.
+  --nsfw                  Route NSFW directly to Qwen Spicy.
   --aspect <ratio>          Output aspect ratio, for example 1:1, 16:9, or 9:16.
   --background <mode>       auto, opaque, or transparent.
   --out <file>              Save the generated image to this path.
@@ -1992,7 +1994,8 @@ Recent model choices:
                   5/10/15s; native 768p default or 480p; no video/audio/multi-image references.
   wan-3.0-prime   Faster Wan 3.0 tier; 2-30s; 480p through 4k; multimodal refs.
   wan-3.0         Wan standard tier with the same public duration/resolution range.
-  seedance-2.5    4-30s; 480p/720p; generate/edit/extend and multimodal refs.
+  seedance-2.5    Defaults to Eco; 4-30s; 1080p default, 720p/2k/4k; multimodal refs.
+  seedance-2.5-eco  480p generation → ByteDance Fast; 1080p default, 2k/4k.
   minimax-h3      4-15s; 768p default or 2k; image/video/audio feature refs.
   grok            T2V/reference generation plus typed edit/extend.
   sync-lipsync-v3 Exactly one video plus one MP3/WAV replacement track.
@@ -3030,7 +3033,12 @@ if (!command || command === '--help' || command === '-h' || command === 'help')
       editArgs.referenceImages = editArgs.referenceImages || [];
       editArgs.referenceImages.push(imageToArg(args[++i]));
     }
+    else if (args[i] === '--image-resolution' && args[i + 1]) {
+      const resolution = args[++i];
+      editArgs.imageResolution = resolution;
+    }
     else if (args[i] === '--aspect' && args[i + 1]) editArgs.aspectRatio = args[++i];
+    else if (args[i] === '--nsfw') editArgs.isNsfw = true;
     else if (args[i] === '--background' && args[i + 1]) {
       const background = args[++i];
       if (!['auto', 'opaque', 'transparent'].includes(background)) {
@@ -3043,7 +3051,7 @@ if (!command || command === '--help' || command === '-h' || command === 'help')
     else promptParts.push(args[i]);
   }
   editArgs.editPrompt = promptParts.join(' ');
-  if (!editArgs.editPrompt) { console.error('Usage: makaron edit [--image <file|url>] [--image-model gemini|gemini-lite|qwen-spicy|openai|gpt-image-2.5-flare|gpt-image-2.5-sunburst|wan2.7-image] [--ref <file>] [--aspect <ratio>] [--background auto|opaque|transparent] [--out <file>] "prompt"'); process.exit(1); }
+  if (!editArgs.editPrompt) { console.error('Usage: makaron edit [--image <file|url>] [--image-model gemini|gemini-2.1|gemini-lite|qwen-spicy|openai|gpt-image-2.5-flare|gpt-image-2.5-sunburst|wan2.7-image] [--ref <file>] [--aspect <ratio>] [--background auto|opaque|transparent] [--out <file>] "prompt"'); process.exit(1); }
   process.stderr.write('🎨 Generating...\n');
   const result = await callMcpTool(baseUrl, headers, 'makaron_edit_image', editArgs);
   saveMcpImage(result, outputPath);
@@ -3123,7 +3131,7 @@ if (!command || command === '--help' || command === '-h' || command === 'help')
       : ['wan3-prime', 'wan3.0-prime', 'wan30-prime', 'wan-3-prime', 'w3.0-video-prime', 'w3.0-video-prime-pro', 'wan-3.0-prime-pro', 'prime'].includes(videoModel)
         ? 'wan-3.0-prime'
         : (videoModel || 'fal-h3-max');
-    const isSeedance25 = selectedVideoModel === 'seedance-2.5';
+    const isSeedance25 = ['seedance-2.5', 'seedance-2.5-eco'].includes(selectedVideoModel);
     const isWan30 = selectedVideoModel === 'wan-3.0' || selectedVideoModel === 'wan-3.0-prime';
     const isSeedanceModel = selectedVideoModel === 'seedance-fast' || selectedVideoModel === 'seedance-mini' || selectedVideoModel === 'seedance' || isSeedance25;
     const isMinimaxH3 = selectedVideoModel === 'minimax-h3';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "makaron-cli",
-  "version": "0.15.4",
+  "version": "0.16.0",
   "description": "Talk to Makaron Agent from the terminal — create projects, edit images, generate videos",
   "type": "module",
   "scripts": {

package/skills/makaron/SKILL.md

@@ -315,7 +315,7 @@ npx makaron-cli edit --image photo.jpg --out result.jpg "make it dramatic"
 npx makaron-cli edit --image-model gpt-image-2.5-flare --background transparent --out sticker.png "a magenta star sticker"
 ```
 
-Options: `--image`, `--image-model gemini|gemini-lite|qwen-spicy|openai|gpt-image-2.5-flare|gpt-image-2.5-sunburst|wan2.7-image`, `--ref <file>` (up to 3), `--aspect <ratio>`, `--background auto|opaque|transparent`, `--out <path>`. Qwen Spicy supports 0–3 input images; legacy `qwen` requests map to it. Pony and WAI are retired. Transparent output routes strictly to GPT Image 2.5 Flare and fails instead of returning an opaque fallback. Wan 2.7 Image is an explicit fast ~1K route; do not automatically retry failures/timeouts, and do not promise exact face preservation.
+Options: `--image`, `--image-model gemini|gemini-2.1|gemini-lite|qwen-spicy|openai|gpt-image-2.5-flare|gpt-image-2.5-sunburst|wan2.7-image`, `--ref <file>` (model-specific limit), `--aspect <ratio>`, `--image-resolution 1K|2K|4K`, `--background auto|opaque|transparent`, `--nsfw`, `--out <path>`. Image Model Capability prefers matching requirements, then relaxes output conflicts to deliver an image: GPT Image 2.5 Flare → Nano Banana 2.1 → Qwen Spicy. Omit the model for Auto: an 8:1 banner or an explicit resolution tier selects Nano Banana 2.1. Native ratio presets do not promise exact mathematical ratios or user pixel dimensions. For Agent-assessed NSFW, pass `--nsfw` to route directly to Spicy; never probe another provider first. Unknown IDs use Flare-first Auto. Output conflicts relax before submission: transparent 8:1 normally keeps 8:1 with opaque output; NSFW always uses Spicy regardless of unsupported dimensions. Preserve input content first and explain actual adjustments. See current MCP tool descriptions for the generated capability table. Classic Nano Banana 2 (`gemini`), Lite, Sunburst and Wan are explicit routes; `openai` means Flare and legacy `qwen` means Spicy. Pony/WAI are retired. Transparent output prefers Flare or explicitly selected Sunburst, and may relax alpha when other requirements conflict. Supplier failures never trigger automatic image resubmission or model switching.
 
 ### `video` — Standalone video tools (no project timeline)
 
@@ -355,7 +355,7 @@ Provider integration contract: every image passed to video generation is a featu
 
 fal H3 Turbo uses the `minimax-h3-max` selector, supports exactly 5/10/15 seconds at 480p/768p, and defaults to native 768p for faster-than-real-time T2V or one-start-image I2V.
 
-Seedance 2.5 uses `--video-model seedance-2.5` and supports 4-30s at 480p/720p, up to 30 images + 10 videos + 10 audios, repeatable local/URL references, `--video-operation generate|edit|extend`, `--extend-direction`, `--output-format mp4|mov`, and `--web-search`. The Evolink route does not currently expose 4K output.
+Seedance 2.5 uses `--video-model seedance-2.5` or `seedance-2.5-eco` for 4-30s videos: generate at 480p, then automatically upscale to 720p, 1080p (default), 2K or 4K in one task. This is upscaled output. Use `seedance-2.5-native` explicitly for unenhanced native output at 480p (default) or 720p; the native Evolink route does not expose 4K. All Seedance 2.5 routes accept up to 30 images + 10 videos + 10 audios, repeatable local/URL references, `--video-operation generate|edit|extend`, `--extend-direction`, and `--web-search`. Eco delivers MP4; native also supports `--output-format mov`. Use `makaron chat` to enhance an existing video without regeneration, or MCP `makaron_upscale_video` with one video URL and `resolution=720p|1080p|2k|4k` (up to 60s).
 
 Wan 3.0 exposes two model choices: `--video-model wan-3.0` and the faster `--video-model wan-3.0-prime`. Both support 2-30s generation, native audio, up to 10 images + 5 videos + 5 audios, and 480p/720p/1080p/2K/4K. Pass `--video-resolution 2k|4k` to select the matching FlashVSR/Pro endpoint automatically; Pro is not a separate model selector. Use generation mode with feature references; typed edit/extend and the relaxed content-filter flag are not supported.
 
@@ -514,3 +514,5 @@ send_message "All done!"
 FAL video models: **fal H3 Turbo** uses `minimax-h3-max` for single-start-frame I2V/T2V. **FAL H3 Max** uses the new selector `fal-h3-max`: native T2V or image/video/audio reference-to-video, default 768p, optional 480p/1080p, integer 5–15s; at most 9 images / 3 videos / 3 audios / 12 total. Reference video/audio each 2–15s and each modality totals at most 15s. Source-video modifications use generation with feature references, not typed edit/extend. Reference input tokens are billed in addition to output video; query current pricing.
 
 The legacy `openai` image-model parameter now resolves to GPT Image 2.5 Flare.
+
+Naming Seedance 2.5 in chat defaults to Seedance 2.5 Eco. Use `seedance-2.5-native` only when explicitly requesting native generation; Eco supports 720p/1080p/2K/4K final delivery.