Zengtudor/Wan2.1
1
0
Fork 0
mirror of https://github.com/Wan-Video/Wan2.1.git synced 2025-11-04 14:16:57 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

AccVideo support

Browse Source
This commit is contained in:
DeepBeepMeep 2025-06-06 00:50:30 +02:00
parent a669de3d82
commit db25325930
2 changed files with 11 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
wan/text2video.py
View File

@ -470,14 +470,14 @@ class WanT2V:
latent_noise_factor = t / 1000 latent_noise_factor = t / 1000
for zz, zz_r, ll in zip(z, z_reactive, [latents]): for zz, zz_r, ll in zip(z, z_reactive, [latents]):
pass pass
# zz[0:16, ref_images_count:overlapped_latents_size + ref_images_count] = zz_r[:, ref_images_count:] * (1.0 - overlap_noise_factor) + torch.randn_like(zz_r[:, ref_images_count:] ) * overlap_noise_factor zz[0:16, ref_images_count:overlapped_latents_size + ref_images_count] = zz_r[:, ref_images_count:] * (1.0 - overlap_noise_factor) + torch.randn_like(zz_r[:, ref_images_count:] ) * overlap_noise_factor
# ll[:, 0:overlapped_latents_size + ref_images_count] = zz_r * (1.0 - latent_noise_factor) + torch.randn_like(zz_r ) * latent_noise_factor ll[:, 0:overlapped_latents_size + ref_images_count] = zz_r * (1.0 - latent_noise_factor) + torch.randn_like(zz_r ) * latent_noise_factor
if conditioning_latents_size > 0 and overlap_noise > 0: if conditioning_latents_size > 0 and overlap_noise > 0:
pass pass
overlap_noise_factor = overlap_noise / 1000 overlap_noise_factor = overlap_noise / 1000
latents[:, conditioning_latents_size + ref_images_count:] = latents[:, conditioning_latents_size + ref_images_count:] * (1.0 - overlap_noise_factor) + torch.randn_like(latents[:, conditioning_latents_size + ref_images_count:]) * overlap_noise_factor # latents[:, conditioning_latents_size + ref_images_count:] = latents[:, conditioning_latents_size + ref_images_count:] * (1.0 - overlap_noise_factor) + torch.randn_like(latents[:, conditioning_latents_size + ref_images_count:]) * overlap_noise_factor
#timestep = [torch.tensor([t.item()] * (conditioning_latents_size + ref_images_count) + [t.item() - overlap_noise]*(len(timesteps) - conditioning_latents_size - ref_images_count))] # timestep = [torch.tensor([t.item()] * (conditioning_latents_size + ref_images_count) + [t.item() - overlap_noise]*(target_shape[1] - conditioning_latents_size - ref_images_count))]
if target_camera != None: if target_camera != None:
latent_model_input = torch.cat([latents, source_latents], dim=1) latent_model_input = torch.cat([latents, source_latents], dim=1)

12
wgp.py
View File

@ -42,8 +42,8 @@ global_queue_ref = []
AUTOSAVE_FILENAME = "queue.zip" AUTOSAVE_FILENAME = "queue.zip"
PROMPT_VARS_MAX = 10 PROMPT_VARS_MAX = 10
target_mmgp_version = "3.4.7" target_mmgp_version = "3.4.8"
WanGP_version = "5.4" WanGP_version = "5.41"
prompt_enhancer_image_caption_model, prompt_enhancer_image_caption_processor, prompt_enhancer_llm_model, prompt_enhancer_llm_tokenizer = None, None, None, None prompt_enhancer_image_caption_model, prompt_enhancer_image_caption_processor, prompt_enhancer_llm_model, prompt_enhancer_llm_tokenizer = None, None, None, None
from importlib.metadata import version from importlib.metadata import version
@ -3263,11 +3263,13 @@ def generate_video(
if exp > 0: if exp > 0:
from rife.inference import temporal_interpolation from rife.inference import temporal_interpolation
if sliding_window and window_no > 1: if sliding_window and window_no > 1:
sample = torch.cat([frames_already_processed[:, -2:-1], sample], dim=1) sample = torch.cat([previous_before_last_frame, sample], dim=1)
previous_before_last_frame = sample[:, -2:-1].clone()
sample = temporal_interpolation( os.path.join("ckpts", "flownet.pkl"), sample, exp, device=processing_device) sample = temporal_interpolation( os.path.join("ckpts", "flownet.pkl"), sample, exp, device=processing_device)
sample = sample[:, 1:] sample = sample[:, 1:]
else: else:
sample = temporal_interpolation( os.path.join("ckpts", "flownet.pkl"), sample, exp, device=processing_device) sample = temporal_interpolation( os.path.join("ckpts", "flownet.pkl"), sample, exp, device=processing_device)
previous_before_last_frame = sample[:, -2:-1].clone()
output_fps = output_fps * 2**exp output_fps = output_fps * 2**exp
@ -4843,8 +4845,8 @@ def generate_video_tab(update_form = False, state_dict = None, ui_defaults = Non
temporal_upsampling = gr.Dropdown( temporal_upsampling = gr.Dropdown(
choices=[ choices=[
("Disabled", ""), ("Disabled", ""),
("Rife x2 (32 frames/s)", "rife2"), ("Rife x2 frames/s", "rife2"),
("Rife x4 (64 frames/s)", "rife4"), ("Rife x4 frames/s", "rife4"),
], ],
value=ui_defaults.get("temporal_upsampling", ""), value=ui_defaults.get("temporal_upsampling", ""),
visible=True, visible=True,