feat: persist photos and enrich conversation history

backend/services/pipecat/pipeline.py

@@ -36,6 +36,7 @@ from services.vision import (
     config_with_main_llm_as_vision,
     config_with_vision_resource,
     image_data_uri,
+    image_jpeg_bytes,
 )
 from services.workflow_engine import WorkflowEngine
 
@@ -58,6 +59,7 @@ from pipecat.processors.aggregators.llm_response_universal import (
 )
 from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
 from pipecat.services.llm_service import FunctionCallParams
+from pipecat.utils.time import time_now_iso8601
 from services.pipecat.turn_config import (
     ConfigurableLLMUserAggregator,
     create_user_turn_strategies,
@@ -763,6 +765,14 @@ async def run_pipeline(
         except asyncio.TimeoutError as exc:
             raise ValueError("等待摄像头视频帧超时") from exc
 
+        if recorder:
+            image_bytes = await asyncio.to_thread(image_jpeg_bytes, image_frame)
+            recorder.record_image_later(
+                image_bytes,
+                input_id=value.input_id,
+                timestamp=time_now_iso8601(),
+            )
+
         if native_vision:
             input_frame = await asyncio.to_thread(
                 _multimodal_user_input_frame,