The ultimate set of ComfyUI custom nodes to streamline your workflows. Happy Halloween! 🎃👻🕸️
Email info@flowstateengineering.com with questions.
** ❤️ Github Sponsor | 🪙 Ca$hApp | 💵 Venmo | 🐐 Stripe **
Everything is open and free, but if you like my work and wish to see updates and new features please consider sponsoring my projects.
- FlowState Unified Styler
- FlowState Unified Sampler
- FlowState Unified Model Loader
- FlowState Unified Prompt & Prompt Output
- FlowState Latent Chooser
- FlowState Unified Video Sampler
Video tutorials can be found on YouTube at FlowState YouTube Channel
The only way to keep the code open and free is by sponsoring its development. The more sponsorships, the more time I can dedicate to open source projects for everyone!
Please consider a Github Sponsorship or Patreon Membership, or one-time donations through CashApp, Venmo or Stripe.
Video Demo in my ComfyUI Playlist
Paste this link into your ComfyUI Manager > Install via Git URL: https://github.com/flowstateeng/FlowState-Nodes.git
OR
In your ComfyUI custom nodes folder...
git clone https://github.com/flowstateeng/FlowState-Nodes.gitOR
COMING SOON: Install through the ComfyUI Custom Node Manager!
- Introduction of orientation selector
- Fixed "unload_loras" bug
- Converted static model directories to dynamic, so no restart required when downloading new models or changing filenames
- Fixed issue with custom resolution inputs
- Video Sampler node is available but still experimental
- Styler node has LoRAs integrated, chain up to 3
- Resolution selection on the Latent Chooser
- Condensed model name fields on Model Loader
- Support for Flux Lite; select model_type = UNET
- CLIP Attention fields on Prompt node condensed to single preset drop down
- Minor bug fixes for wider compatibility
- Model Persistence (~1.5 days)
- Control Net support! (~1-3 days)
- Video Sampler Tuning & Presets (~2-5 days)
- Custom message instructions and presets for LLM prompt models, including for negative prompts (~3-7 days)
- Inpainting/Outpainting (~5-9 days)
- Image Composer (~7-13 days)
- ComfyUI Custom Node Manager Installation (under review now)
The FlowState Unified Styler brings together the powerful capabilities of LoRAs & Control Nets, unifying them in a single, efficient node. Now you can streamline style transfers, adaptive conditioning, and custom fine-tuning with ease, all while reducing workflow complexity. Perfect for creators who want versatility and control without the clutter!
**Control Nets Coming ~10/28
Use this powerful sampler with your Flux AND Stable Diffusion models. Includes support for testing a variety of parameters in your batches. Add parameters and/or prompts to your output images for easy identification. Adjust font size and number of lines to include exactly the text you want. Control your terminal/CMD prompt output. Includes Latent Multiply and VAE Decode functionality, so no need for extra nodes in your workflow.
Tighten up your workflow with the FlowState Unified Prompt & Prompt Output nodes. Load a .gguf model to generate sophisticated image prompts, or simply flip a switch to use your own. Handles the text encoding for you internally, so you don't need another Text Encode node off to the side. Outputs the original and generated text and conditioning, in case you want to pass it to another node. Also includes the Clip Attention Multiply functionality, thus eliminating another node from your workflow.
Simplify your model loading with the FlowState Unified Model Loader. Load your UNET and Checkpoint models in a single node. Including support for Flux NF4! (Flux .gguf support coming soon!) Includes global seed you can use for your entire workflow. No need to have three separate nodes for Stable Diffusion, Flux Dev/Schnell and Flux NF4. Do it all in a single node!
Expand your latent options and clean up your workflow with FlowState Latent Chooser. Easily select between an empty latent, input image as a latent, and imported image as a latent with a simple switch. Vary the denoise on your sampler to control the amount of style transfer you want from your image latent in your generated images.
Johnathan Chivington:
- PLEASE MESSAGE YOUR BUG REPORTS ON DISCORD, PATREON, OR SUBMIT AN ISSUE ON GITHUB. MERGE/PULL REQUESTS WILL BE DELETED.
Not currently accepting outside contributors, but feel free to use as you wish.
There is no license associated with this content.
