Update handler.py

handler.py

@@ -14,7 +14,7 @@ import numpy as np
 import torch
 from diffusers import LTXPipeline, LTXImageToVideoPipeline
 from PIL import Image
-
+from teacache import TeaCacheConfig, enable_teacache
 from varnish import Varnish
 
 # Configure logging
@@ -195,6 +195,10 @@ class GenerationConfig:
     # it should look the same or nearly the same as the input but it isn't technically lossless.
     # The range is exponential, so increasing the CRF value +6 results in roughly half the bitrate / file size, while -6 leads to roughly twice the bitrate.
     quality: int = 18
+
+    # TeaCache settings
+    enable_teacache: bool = True
+    teacache_threshold: float = 0.15
     
     def validate_and_adjust(self) -> 'GenerationConfig':
         """Validate and adjust parameters to meet constraints"""
@@ -347,6 +351,8 @@ class EndpointHandler:
                     - audio_prompt (optional, str): prompt to use for the audio generation (concepts to add)
                     - audio_negative_prompt (optional, str): nehative prompt to use for the audio generation (concepts to ignore)
                     - quality (optional, str, default to 18): The range of the CRF scale is 0–51, where 0 is lossless (for 8 bit only, for 10 bit use -qp 0), 23 is the default, and 51 is worst quality possible.
+                    - enable_teacache (optional, bool, default to True): Generate faster at the cost of a slight quality loss
+                    - teacache_threshold (optional, float, default to 0.15): Amount of cache
         Returns:
             Dictionary containing:
                 - video: Base64 encoded MP4 data URI
@@ -392,6 +398,10 @@ class EndpointHandler:
             audio_prompt=params.get("audio_prompt", GenerationConfig.audio_prompt),
             audio_negative_prompt=params.get("audio_negative_prompt", GenerationConfig.audio_negative_prompt),
             quality=params.get("quality", GenerationConfig.quality),
+            
+            # TeaCache settings
+            enable_teacache=params.get("enable_teacache", True),
+            teacache_threshold=params.get("teacache_threshold", 0.15)
         ).validate_and_adjust()
         
         #logger.debug(f"Global request settings:")