Extensions/ComfyUI-GlifNodes
ComfyUI Extension

ComfyUI-GlifNodes

Custom set of nodes used by glif.app. With glif you can build mini apps that are powered by custom comfy workflows.

By glifxyz·Created 3 years ago·Updated about a year ago· 65
glifxyz/ComfyUI-GlifNodes
Nodes11
On cloudLocal install
Categoryimage/postprocessing, sd
Stars65
Updatedabout a year ago

Nodes (11)

Film Grain Effect

Film grain for when the AI polish is too clean

image/postprocessing
Flux ReduxFloat Ramp

Ramping Flux Redux strength so style fades in instead of bulldozing

sd
Consistency VAE Decoder

DALL·E 3's VAE decoder in ComfyUI — interesting, and very 2023

latent
Patch Consistency VAE Decoder

Swap your SD 1.5 VAE decoder for DALL·E 3's, in tiles so it fits

sd
Glif Variable

A node that only does something inside glif's own editor

glif/variables
Load HF Embedding

Pull a textual-inversion embedding from HuggingFace without leaving ComfyUI

n/a
Load HF Lora

Load a LoRA straight from a HuggingFace repo, no manual download

loaders
Image Padding Advanced

Grow-only padding with three edge strategies (and a silent trap)

image
Image to Multiple of

The two-minute fix for 'latent must be divisible by 8' errors

image
Load Lora from URL

Paste a LoRA URL — Civitai, HuggingFace, or anywhere — and it just loads

loaders
Image to SDXL compatible WH

Get your image into an SDXL-native resolution without doing math

image
Readme

ComfyUI-GlifNodes

Custom nodes for comfyUI.

Install

pip install -r requirements.txt

Nodes

<details> <summary><b><code>GlifConsistencyDecoder</code></b> openai's consistency decoder from hf hub</summary> <img src="docs/consistency_vae.png" max-height="500px"/> </details> <details> <summary><b><code>🧪GlifPatchConsistencyDecoderTiled</code></b> decoder supporting tiled decoding</summary> <img src="docs/patch_vae.png" max-height="500px"/> </details> <details> <summary><b><code>SDXLAspectRatio</code></b> find the closest SDXL height and width for an image</summary> <img src="docs/sdxl_aspect_ratio.png" max-height="500px"/> </details> <details> <summary><b><code>ImageToMultipleOf</code></b> either crop or stretch an image to a multiple of a specific value</summary> <img src="docs/multiple_of.png" max-height="500px"/> </details> <details> <summary><b><code>HFHubLoraLoader</code></b> load LoRAs directly from Huggingface</summary> <img src="docs/load_hf_lora.png" max-height="500px"/> </details> <details> <summary><b><code>HFHubEmbeddingLoader</code></b> load embeddings directly from Huggingface</summary> <img src="docs/load_hf_embedding.png" max-height="500px"/> </details> <details> <summary><b><code>LoraLoaderFromURL</code></b> load LoRAs using a url.</summary> <img src="docs/load_lora_from_url.png" max-height="500px"/>

This will work with Huggingface, Civitai, and possibly others. Most models on Civitai will require an API key to download, which can be obtained on your Civitai account page. Add it to your environment variables as CIVITAI_API_KEY.

</details> <details> <summary><b><code>GlifVariable</code></b> easily use glif variables on the glif comfy editor</summary> <img src="docs/glif_variable.png" max-height="500px"/> </details> <br>

Acknowledgements

diffusers: https://github.com/huggingface/diffusers openai - consistencydecoder: https://github.com/openai/consistencydecoder reddit user Total-Resort-3120 for float ramp node

dev

Run tests:

python tests/test_image_padding_advanced.py