MariusKM 02c8214e73 Fix keyframe repetition to properly use frame_blend_width parameter
- Keyframes are now repeated across the entire blend_width region
- Previously only unmasked as many frames as provided in input images
- Now single-frame keyframes are extended to fill the blend region
- Start frame: repeated from frame 0 to frame_blend_width
- Middle frame: repeated across frame_blend_width region centered at middle_frame_position
- End frame: repeated across last frame_blend_width frames
- This provides much stronger anchoring and reduces blurry inpainted regions
- Fixes issue where most of video was masked with single-frame keyframes
2025-10-14 16:01:00 +02:00
2024-05-31 12:15:30 +02:00
2024-05-31 12:11:23 +02:00

Badman Utiliity Nodes

A small collection of Utility and quality of Life Nodes developed to aid me in my workflows. Nothing special and nothing crazy, just extending certain functionalities of base nodes.

Files

BadmanCLIPTextEncodeSDXLRegion : SDXL conditioning Node intended for use with CutOff, sadly SDXL does not do too well with CutOff.

ImageBlend(Badman) : Extended Image Blend node with some extra blend functions.

Int Math (Badman) : Integer Math node with some basic Math functions

HexGenerator(Badman) : Node that generates Hex Colors from linear RGB Values

String (Badman) : Simple String Type node

Concat String (Badman) : Simple String Concat Node with a new line option, ideal for combining Tokens from BLIP or CLIP interrogation

Print (Badman) : Prints String input to console

IO Config (Badman) : IO configurator that sets up paths dynamically to be stored in setter nodes

Select String from List (Badman): Selects a specific String from a String List output and forwards this to the output. Can be used to target a specific String when loading prompts from files or from a multi image BLIP interrogator.

Inject Latent Noise Masked (Badman): Injects noise into latent space with mask control. High mask values receive more noise, low values receive less. Supports multiple blend modes (replace, add, multiply) and mask inversion.

WAN Three Frame To Video: Generates video from 3 keyframes (start, middle, end) with proper masking for smooth transitions. Supports adjustable middle frame positioning, configurable frame blend width for smooth transitions, and CLIP vision output concatenation for enhanced conditioning.

TODO

  • IO Config (Badman) : Refactor to allow for renaming specific I/O paths, currently fixed path names that I tend to use.
S
Description
No description provided
Readme MIT
105 KiB
Languages
Python 91.6%
JavaScript 8.4%