configs/vlog_read_script_sample.yaml

# path:
ckpt: "pretrained/ShowMaker.pt"
pretrained_model_path: "pretrained/stable-diffusion-v1-4/"
image_encoder_path: "pretrained/OpenCLIP-ViT-H-14"
save_path: "results/vlog/teddy_travel/video"

# script path
reference_image_path: ["results/vlog/teddy_travel/ref_img/teddy.jpg"]
script_file_path: "results/vlog/teddy_travel/script/video_prompts.txt"
zh_script_file_path: "results/vlog/teddy_travel/script/zh_video_prompts.txt"
protagonist_file_path: "results/vlog/teddy_travel/script/protagonists_places.txt"
reference_file_path: "results/vlog/teddy_travel/script/protagonist_place_reference.txt"
time_file_path: "results/vlog/teddy_travel/script/time_scripts.txt"
video_transition: False

# model config:
model: UNet
num_frames: 16
image_size: [320, 512]
negative_prompt: "white background"

# sample config:
ref_cfg_scale: 0.3
seed: 3407
guidance_scale: 7.5
cfg_scale: 8.0
sample_method: 'ddim'
num_sampling_steps: 100
researve_frame: 3
mask_type: "first3"
use_mask: True
use_fp16: True
enable_xformers_memory_efficient_attention: True
do_classifier_free_guidance: True
fps: 8
sample_num: 

# model speedup
use_compile: False