v5 release with triple architecture support and prompt enhancer

ltx_video/configs/ltxv-13b-0.9.7-dev.original.yaml

@@ -0,0 +1,41 @@
+    
+pipeline_type: multi-scale
+checkpoint_path: "ltxv-13b-0.9.7-dev.safetensors"
+downscale_factor: 0.6666666
+spatial_upscaler_model_path: "ltxv-spatial-upscaler-0.9.7.safetensors"
+stg_mode: "attention_values" # options: "attention_values", "attention_skip", "residual", "transformer_block"
+decode_timestep: 0.05
+decode_noise_scale: 0.025
+text_encoder_model_name_or_path: "PixArt-alpha/PixArt-XL-2-1024-MS"
+sampler: "from_checkpoint" # options: "uniform", "linear-quadratic", "from_checkpoint"
+prompt_enhancement_words_threshold: 120
+prompt_enhancer_image_caption_model_name_or_path: "MiaoshouAI/Florence-2-large-PromptGen-v2.0"
+prompt_enhancer_llm_model_name_or_path: "unsloth/Llama-3.2-3B-Instruct"
+stochastic_sampling: false
+
+
+first_pass:
+  #13b Dynamic
+  guidance_scale:  [1, 6, 8, 6, 1, 1]
+  stg_scale:  [0, 4, 4, 4, 2, 1]
+  rescaling_scale: [1, 0.5, 0.5, 1, 1, 1]
+  guidance_timesteps: [1.0, 0.9933, 0.9850, 0.9767, 0.9008, 0.6180]
+  skip_block_list:  [[11, 25, 35, 39], [22, 35, 39], [28], [28], [28], [28]]
+  num_inference_steps: 30 #default
+
+
+second_pass:
+  #13b Dynamic
+  guidance_scale:  [1, 6, 8, 6, 1, 1]
+  stg_scale:  [0, 4, 4, 4, 2, 1]
+  rescaling_scale: [1, 0.5, 0.5, 1, 1, 1]
+  guidance_timesteps: [1.0, 0.9933, 0.9850, 0.9767, 0.9008, 0.6180]
+  skip_block_list:  [[11, 25, 35, 39], [22, 35, 39], [28], [28], [28], [28]]
+  #13b Upscale
+  # guidance_scale:  [1, 1, 1, 1, 1, 1]
+  # stg_scale:  [1, 1, 1, 1, 1, 1]
+  # rescaling_scale: [1, 1, 1, 1, 1, 1]
+  # guidance_timesteps: [1.0, 0.9933, 0.9850, 0.9767, 0.9008, 0.6180]
+  # skip_block_list:  [[42], [42], [42], [42], [42], [42]]
+  num_inference_steps: 30 #default
+  strength: 0.85

ltx_video/configs/ltxv-13b-0.9.7-dev.yaml

@@ -0,0 +1,34 @@
+pipeline_type: multi-scale
+checkpoint_path: "ltxv-13b-0.9.7-dev.safetensors"
+downscale_factor: 0.6666666
+spatial_upscaler_model_path: "ltxv-spatial-upscaler-0.9.7.safetensors"
+stg_mode: "attention_values" # options: "attention_values", "attention_skip", "residual", "transformer_block"
+decode_timestep: 0.05
+decode_noise_scale: 0.025
+text_encoder_model_name_or_path: "PixArt-alpha/PixArt-XL-2-1024-MS"
+precision: "bfloat16"
+sampler: "from_checkpoint" # options: "uniform", "linear-quadratic", "from_checkpoint"
+prompt_enhancement_words_threshold: 120
+prompt_enhancer_image_caption_model_name_or_path: "MiaoshouAI/Florence-2-large-PromptGen-v2.0"
+prompt_enhancer_llm_model_name_or_path: "unsloth/Llama-3.2-3B-Instruct"
+stochastic_sampling: false
+
+first_pass:
+  guidance_scale: [1, 1, 6, 8, 6, 1, 1]
+  stg_scale: [0, 0, 4, 4, 4, 2, 1]
+  rescaling_scale: [1, 1, 0.5, 0.5, 1, 1, 1]
+  guidance_timesteps: [1.0, 0.996,  0.9933, 0.9850, 0.9767, 0.9008, 0.6180]
+  skip_block_list: [[], [11, 25, 35, 39], [22, 35, 39], [28], [28], [28], [28]]
+  num_inference_steps: 30
+  skip_final_inference_steps: 3
+  cfg_star_rescale: true
+
+second_pass:
+  guidance_scale: [1]
+  stg_scale: [1]
+  rescaling_scale: [1]
+  guidance_timesteps: [1.0]
+  skip_block_list: [27]
+  num_inference_steps: 30
+  skip_initial_inference_steps: 17
+  cfg_star_rescale: true

ltx_video/configs/ltxv-13b-0.9.7-distilled.yaml

@@ -0,0 +1,28 @@
+pipeline_type: multi-scale
+checkpoint_path: "ltxv-13b-0.9.7-distilled.safetensors"
+downscale_factor: 0.6666666
+spatial_upscaler_model_path: "ltxv-spatial-upscaler-0.9.7.safetensors"
+stg_mode: "attention_values" # options: "attention_values", "attention_skip", "residual", "transformer_block"
+decode_timestep: 0.05
+decode_noise_scale: 0.025
+text_encoder_model_name_or_path: "PixArt-alpha/PixArt-XL-2-1024-MS"
+precision: "bfloat16"
+sampler: "from_checkpoint" # options: "uniform", "linear-quadratic", "from_checkpoint"
+prompt_enhancement_words_threshold: 120
+prompt_enhancer_image_caption_model_name_or_path: "MiaoshouAI/Florence-2-large-PromptGen-v2.0"
+prompt_enhancer_llm_model_name_or_path: "unsloth/Llama-3.2-3B-Instruct"
+stochastic_sampling: false
+
+first_pass:
+  timesteps: [1.0000, 0.9937, 0.9875, 0.9812, 0.9750, 0.9094, 0.7250]
+  guidance_scale: 1
+  stg_scale: 0
+  rescaling_scale: 1
+  skip_block_list: [42]
+
+second_pass:
+  timesteps: [0.9094, 0.7250, 0.4219]
+  guidance_scale: 1
+  stg_scale: 0
+  rescaling_scale: 1
+  skip_block_list: [42]

ltx_video/configs/ltxv-2b-0.9.6-dev.yaml

@@ -0,0 +1,17 @@
+pipeline_type: base
+checkpoint_path: "ltxv-2b-0.9.6-dev-04-25.safetensors"
+guidance_scale: 3
+stg_scale: 1
+rescaling_scale: 0.7
+skip_block_list: [19]
+num_inference_steps: 40
+stg_mode: "attention_values" # options: "attention_values", "attention_skip", "residual", "transformer_block"
+decode_timestep: 0.05
+decode_noise_scale: 0.025
+text_encoder_model_name_or_path: "PixArt-alpha/PixArt-XL-2-1024-MS"
+precision: "bfloat16"
+sampler: "from_checkpoint" # options: "uniform", "linear-quadratic", "from_checkpoint"
+prompt_enhancement_words_threshold: 120
+prompt_enhancer_image_caption_model_name_or_path: "MiaoshouAI/Florence-2-large-PromptGen-v2.0"
+prompt_enhancer_llm_model_name_or_path: "unsloth/Llama-3.2-3B-Instruct"
+stochastic_sampling: false

ltx_video/configs/ltxv-2b-0.9.6-distilled.yaml

@@ -0,0 +1,17 @@
+pipeline_type: base
+checkpoint_path: "ltxv-2b-0.9.6-distilled-04-25.safetensors"
+guidance_scale: 3
+stg_scale: 1
+rescaling_scale: 0.7
+skip_block_list: [19]
+num_inference_steps: 8
+stg_mode: "attention_values" # options: "attention_values", "attention_skip", "residual", "transformer_block"
+decode_timestep: 0.05
+decode_noise_scale: 0.025
+text_encoder_model_name_or_path: "PixArt-alpha/PixArt-XL-2-1024-MS"
+precision: "bfloat16"
+sampler: "from_checkpoint" # options: "uniform", "linear-quadratic", "from_checkpoint"
+prompt_enhancement_words_threshold: 120
+prompt_enhancer_image_caption_model_name_or_path: "MiaoshouAI/Florence-2-large-PromptGen-v2.0"
+prompt_enhancer_llm_model_name_or_path: "unsloth/Llama-3.2-3B-Instruct"
+stochastic_sampling: true