Home Artists Posts Import Register

Downloads

Content

sdxl requires xformers, so please use venv install guide for local Windows install. Please use path with no whitespaces.

Guide: https://github.com/Sxela/WarpFusion#local-installation-guide-for-windows-venv

An intermediary release with some controlnet logic cleanup and QoL improvements, before diving into sdxl controlnets.

Changelog:

v0.19.2, 13.08.2023:

  • add extra per-controlnet settings: source, mode, resolution, preprocess
  • add global and per-controlnet settings to gui
  • add beep sounds by Infinitevibes
  • add normalize controlnet weights toggle

v0.19.3, 14.08.2023:

  • fix beep error
  • bring back init scale, fix deflicker init scale error thanks to #rebirthai

v0.19.4, 14.08.2023:

  • make cc_masked_diffusion a schedule

v0.19.5, 14.08.2023:

  • fix control_source error for sdxl

v0.19.6, 18.08.2023:

  • fix cc_masked_diffusion not loaded from gui

19.08.2023: sent to L tier

v0.19.7, 24.08.2023:

  • fix ModuleNotFound: safetensors error

v0.19.8:

  • fix cv2.error ssize.empty error for face controlnet

11.09.2023: 

  • move to M tier

Per-controlnet Settings

gui -> controlnet

Allows finer control over controlnet settings. Note, that some controlnets don't have preprocessors or resolution settings as they use raw images as input and render output resolution.

mode:
balanced - mode used in previous versions, balanced between prompt and controlnet
controlnet - pay more attention to controlnet (seems to be buggy at the moment)
prompt - pay more attention to prompt
global - use global settings

source:
stylized - use current frame init (stylized warped previous frame)
cond_video - use cond_video frame
raw_frame - use raw init frame
color_video - use color_video frame
custom source - use frames from custom path
global - use global settings from cond_image_src variable
for inpaint model the default is stylized

resolution:
sets controlnet annotator resolution
-1 = use global settings

preprocess:
apply annotator to controlnet source.
for example, you need to disable this for depth controlnet if you have a custom depth video.

Controlnet Global Settings

Manages global controlnet settings. The same logic applies, except source is limited to init (raw_frame) stylized and cond video.

Normalize Controlnet Weights

gui -> controlnet -> normalize_cn_weights

By default, controlnet weights are normalized to add up to 1. You can disable this to have higher controlnet weights.

Local install guide:
https://github.com/Sxela/WarpFusion/blob/main/README.md

Tutorials:
https://youtu.be/HkM-7wxtkGA
https://www.youtube.com/watch?v=FxRTEILPCQQ
https://www.youtube.com/watch?v=wqXy_r_9qw8
https://www.youtube.com/watch?v=VMF7L0czyIg
https://www.youtube.com/watch?v=m8xaPnaooyg

Youtube playlist with settings:
https://www.youtube.com/watch?v=wvvcWm4Snmc&list=PL2cEnissQhlCUgjnGrdvYMwUaDkGemLGq

For tech support and other questions please join our discord server:
https://discord.gg/YrpJRgVcax

Discord is the preferred method because it is nearly impossible to provide any decent help or tech support via Patreon due to its limited text formatting and inability to add screenshots or videos to comments or DMs.
Error reports in comments will be deleted and reposted in Discord.

Comments

Anonymous

Hello my 1st frame looks really beautiful but after that it's become very chaotic... how can I achieve consistent frames.? help me out please...

sxela

you can start here -https://discord.com/channels/973802253204996116/1117654771470442587 also check my settings on youtube

Anonymous

hello, can you help me with this error? in () 409 410 gui_misc = { --> 411 "user_comment": Textarea(value=user_comment,layout=Layout(width=f'80%'), description = 'user_comment:', description_tooltip = 'Enter a comment to differentiate between save files.'), 412 "blend_json_schedules": Checkbox(value=blend_json_schedules, description='blend_json_schedules',indent=True, description_tooltip = 'Smooth values between keyframes.', tooltip = 'Smooth values between keyframes.'), 413 "VERBOSE": Checkbox(value=VERBOSE,description='VERBOSE',indent=True, description_tooltip = 'Print all logs'), NameError: name 'user_comment' is not defined

Kim

Hi, thank you for the update and I really love to play around! Just one minor question. I used to use v0.17 and recently turn to v0.19. Everything went well running until finished ‘create video’. However, when I use the same vm and continue to run another diffusion, I noticed that the model_path turn to ‘realesr-animevideov3.pth’ not my chosen model. Is it intended?

sxela

no, probably due to upscale video model chosen variable having the same name

Iulian Dovletiu

can you please help me with this? the images are in videoframes forlder. Total frames: 0 --------------------------------------------------------------------------- IndexError Traceback (most recent call last) in () 202 frames_in = sorted(glob(batchFolder+f"/{folder}({run})_*.png")) 203 --> 204 frame0 = Image.open(frames_in[0]) 205 if use_background_mask_video: 206 frame0 = apply_mask(frame0, 0, background_video, background_source_video, invert_mask_video) IndexError: list index out of range.

sxela

you have no frames rendered in the selected run, make sure you have correct settings in video export cell