ComfyUI
TODO[edit | edit source]
Nvidia_RTX_Nodes_ComfyUI - Faster than SeedVR2
Upscale Fix?
Tiled Diffusion?
IPAdapter
Outpainting - https://www.youtube.com/watch?v=j20P4hAZS1Q https://www.youtube.com/watch?v=qLZJ7iSq9tY
Misc[edit | edit source]
Regional prompting for AI Image Generation (Flux and ComfyUI) - Both DetailD (including the custom sampler stuff) and region based prompting
ComfyUI Tutorial, SEGS Picker: How to prompt for each individual face
EASY Outpainting in ComfyUI: 3 Simple Ways with Auto-Tagging (Joytag) | Creative Workflow Tutorial
https://www.youtube.com/@goshniiAI/videos - Upscaling
Generate Million-Megapixel Images with Z Image Turbo!
How to Train a Z-Image-Turbo LoRA with AI Toolkit
Stop Making Blurry AI Art! The "Halftone" Fix for Wan & Qwen (ComfyUI Workflow)
Insane ComfyUI Upgrade: 165 Custom Nodes That Actually Work
Different Region = Different LoRA! New ComfyUI Nodes for Area Composition
Comfyui 101: How to Create X/Y Plots for SDXL & Flux How to Use Flux Unet ( & GGUF) Models With TinyTerraNodes XY Plot
https://civitaiarchive.com https://civitasbay.org/
zer0int/CLIP-GmP-ViT-L-14/tree/main - what's the best version of Vit-L clip encoder? smooth and detail both r good
Updateing
git pull
1536×432 = 32:9
1536×864 = 16:9
Installing[edit | edit source]
Comfy CLI[edit | edit source]
pip install comfy-cli --break-system-packages comfy --install-completion
If Comfy cli isn't installed, install it from the AUR (check with "comfy --version"):
yay -S python-comfy-cli comfy --install-completion
comfy install --nvidia --cuda-version 13.0 #Figure out the newest version it supports with "comfy install --help", currently it's 13.0 /home/hegemon/comfy/ComfyUI/.venv/bin/python -m pip install matrix-nio # Enabled matrix sharing cd comfy/ComfyUI mv models models.stock ln -s ~/Data/image-models models # The models folder must be linked to the locally installed ones ln -s ~/Data/comfyui-workflows ./user/default/workflows # Link the user workspaces
To allow installing custom nodes in the manager, Edit ~/comfy/ComfyUI/user/__manager/config.ini and set
security_level = weak network_mode = personal_cloud
Install SageAttention 2.x[edit | edit source]
cd comfy/ComfyUI source .venv/bin/activate.fish git clone https://github.com/thu-ml/SageAttention.git export EXT_PARALLEL=4 NVCC_APPEND_FLAGS="--threads 8" MAX_JOBS=32 python setup.py install
Running[edit | edit source]
There is a systemd user service:
systemctl --user start comfyui.service
Otherwise it can be manually started with:
comfy launch -- --listen 0.0.0.0 --use-sage-attention
Custom Nodes[edit | edit source]
Install the following custom nodes:
comfy node install ComfyUI-Lora-Manager comfy node install ComfyUI-GGUF comfy node install ComfyUI-KJNodes comfy node install rgthree-comfy comfy node install comfyui-detail-daemon comfy node install ComfyUI-Pixaroma comfy node install seedvr2_videoupscaler
ComfyUI-Easy-Use
ComfyUI-Crystools - For resource monitor
ComfyUI-VideoHelperSuite
ComfyUI_UltimateSDUpscale
ComfyUI-AI-Pixel-Art-Enhancer
comfyui_controlnet_aux
UV install (old method)[edit | edit source]
uv pip install --upgrade torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu130
uv pip install sageattention==2.2.0 --no-build-isolation
pip install -r requirements.txt
Workflows[edit | edit source]
Inpainting[edit | edit source]
What is the current best way to do Inpainting... - Asking Gemini - Says to use ip-adapter. Also mentions 'grid enforcer' for pixel art (ComfyUI-AI-Pixel-Art-Enhancer)
Inpainting with Z-Image Turbo Explained (3 Different Options)
Z-image Advanced Tutorial: Controlnet, inpainting, 4K+ upscaling & more
Union 2.1 - Better Inpaint & Control for Z-Image-Turbo in ComfyUI!
The MOST reliable and simple Inpainting Qwen-Edit workflow! - Uses a 'reference' node.
Flux2 Klein 9B Just Nailed Inpainting | ComfyUI Tutorial - Used "Reference Latent"
Image-to-Image inpainting Explained — A Powerful Technique You Need to Know ｜ Draw Things Tutorial - Ernie Image Turbo - Not in ComfyUI. Not an inpainting model.
https://stable-diffusion-art.com/qwen-image-edit-inpaint/
Simple ComfyUI Inpainting Workflow | Image To Image AI Editing Step By Step - A simple workflow using cyberrealstic
| Model | |||
|---|---|---|---|
| Z-Image Turbo | Z-Image-Turbo-Fun-Controlnet-Union | ||
| Qwen-Image | Qwen-Image-InstantX-ControlNet-Inpainting | ||
LORA training[edit | edit source]
Training LoRA inside ComfyUI for character and style / detailed workflow guide with examples
How to Train a FLUX.2 LoRA with AI Toolkit - Trigger keyword+Differential Output Preservation. Cache Text Embeddings (Investigage Match Target Res)
Misc Workflows[edit | edit source]
FLUX 2 Klein 4B vs 9B Multi Camera Angles - One Click, 8 Camera Angles
ComfyUI With SAM 3.1 Segmentation Native Support! No Custom Nodes
Prompts[edit | edit source]
[X : Y : N] - Use X for N steps then Y. Not sure if this works in comfyui.
Prompting Guide - FLUX.2 [klein]
https://docs.bfl.ai/guides/prompting_guide_kontext_i2i
https://danbooru.donmai.us/wiki_pages/tag_groups
https://danbooru.donmai.us/related_tag
https://danbooru.donmai.us/posts?tags=list_of_style_parodies&z=2
https://betterwaifu.com/blog/ai-pose-prompts
Flux2 Klein 9B-kv[edit | edit source]
SAI-Enhance, SAI-Anime, SAI-Photographic, SAI-Digital art, SAI-Comic book, SAI-Fantasy art, SAI-Analog film, SAI-Neonpunk, SAI-Isometric, SAI-Lowpoly, SAI-Origami, SAI-Line art, SAI-Craft clay, SAI-Cinematic, SAI-3d-model, SAI-pixel art, SAI-Texture, photographic, deep-field, analog-film, cinematic, red, documentary, nature-photography, editorial, polaroid, 3d-model, low-poly, modeling-compound, diorama, miniatures, abstract, anime, comic-book, digital-art, fantasy-art, dark-fantasy, oil-painting, watercolor, charcoal, color-pencil, crayon, cross-stitch, felt, origami, scrapbook, isometric, line-art, vector-art, neon-punk, pixel-art, tile-texture, lithography, mosaic, woodblock-print, sticker, stained-glass, tattoo, statue, album-art, wes-anderson, vaporwave, clarendon, gingham, juno, lark, nonagon, kaleidoscope, haunting, glam, mecha, padam, collage, paper-mache, macrame, quilling, pottery, batik, felting, marquetry, wildstyle-graffiti, bubble-graffiti, stencil-graffiti, throw-up-graffiti, tag-graffiti
Anima Dog[edit | edit source]
ye-pop masterpiece, best quality, score_7, safe, 1animal, dog, solo, @iris scott, @leonid afremov. An expressive oil painting of a golden retriever dog swimming in dark, swirling water. The water ripples with intense, iridescent colors ranging from deep blues and greens to fiery oranges and purples. Warm light hits the dog's back, creating a glowing effect on its wet fur, rendered with thick, textured brushstrokes. animal focus, water, swimming, wet fur, high contrast, vibrant, colorful, iridescent, reflections, ripples, top-down perspective, traditional media, impasto, beautiful lighting
Artists[edit | edit source]
Deak Ferrand - Dune, Sandman
by Diane Pierce and Alena Aenami in the style of Dan Mumford,
Asaf hanuka
rodolphe wytsman impressionism
by Michal Karcz, by Arthur Sarnoff, by Marc Simonetti,
Greg Rutkowski
Syd Mead
Cyberpunk[edit | edit source]
Syd Mead - Bladerunner concept art
Josan Gonzalez
Sparth (Nicolas Bouvier)
Beeple (Mike Winkelmann) - Weird colours
Simon Stålenhag - Giant robots guy
Fantasy[edit | edit source]
Greg Rutkowski
Pascal Blanché
Boris Vallejo
Anima Artists[edit | edit source]
https://www.reddit.com/r/StableDiffusion/comments/1tnlz59/anima_style_explorer_anima_lora_explorer/
Benchmark[edit | edit source]
Video Upscale[edit | edit source]
| 2026-01-30 | 336.80s | GAN upscaler template in ColmfyUI | for a 3 second video using the | |
| 2026-01-30 | 19.04s | GAN upscaler template in ColmfyUI | for a second pass (Cached or load time?) | |
| 2026-01-30 | 343.10s | seedvr2_ema_3b_fp16 | ||
| 2026-01-30 | 351.37 | seedvr2_ema_3b_fp8 | ||
| 2026-01-30 | 354.54 | seedvr2_ema_3b-Q4_K_M | ||
| 2026-01-30 | 442.38 | seedvr2_ema_3b_fp8 | With torch compile and tile size of 768 (First run) | |
| 2026-01-30 | 305.02 | seedvr2_ema_3b_fp8 | With torch compile and tile size of 768 (Second run) | |
| 2026-01-30 | 408.12 | seedvr2_ema_3b_fp8 | With torch compile and tile size of 768, swap blocks 6 (First run) | |
| 2026-01-30 | 341.21 | seedvr2_ema_3b_fp8 | With torch compile and tile size of 768, swap blocks 6, batch size 21 | |
| 2026-01-30 | 375.13 | seedvr2_ema_3b_fp8 | With torch compile and encoding tile size of 1024, swap blocks 6, batch size 21 | |
| 2026-01-30 | 323.73 | seedvr2_ema_3b_fp8 | With torch compile and encoding tile size of 1024, swap blocks 9, batch size 21, sageattention2 | |
| 2026-01-30 | 405.08 | seedvr2_ema_7b_Q4_K_M | Default settings | |
Training LORAs[edit | edit source]
All the training programs just seem to be wrappers around hugging face scripts.
https://www.youtube.com/watch?v=zegMOOKy_6AAI ToolKit - Every Setting Explained
AI ToolKit - Every Setting Explained
| Last Updated | ⭐ | Z-Image | Flux.2 Klein | Qwen | |
|---|---|---|---|---|---|
| Kohya SS | 7 Months Ago | 11.9k | |||
| OneTrainer | 2 days ago. | 2.7k | ✔️ (unknown base/turbo) | ❌? | ✔️ |
| Ai Toolkit | 2 days ago. | 9.2k | |||
| Musubi | Yesterday. | 1.7k | ✔️ (base) | ✔️ | ✔️ (including layered) |
Upscaling[edit | edit source]
Seed VR2[edit | edit source]
SeedVR2 BEST SETTINGS for Every GPU! (6GB–32GB+) Nobody Told You This…
🤿 SeedVR2 v2.5 Video Upscaling: Official Guide from the ComfyUI Integration Team | AInVFX Nov 7 - Talks about out of memory at different phases.
NvidiaPiD[edit | edit source]
The BEST AI for 4K images. Free & fast
3D Modeling[edit | edit source]
Pixal3D on 6GB VRAM — Better Than Trellis 2! (Free & Local)
LTX-2[edit | edit source]
LTX 2 Q6 16GB on 8GB VRAM? 🤯 Free ComfyUI Workflow 2026
LTX-2 generate a 30s video in 310seconds - Has distilled GGUFs
LTX-2 Simplified Workflow 🔥 Distilled Checkpoints or Separated VAE & Transformer?
LTX-2 Fix: Yes, You Can Actually Use It Now (16-24gb VRAM) - Didn't work
This fixed my OOM issues with LTX-2 - Didn't work.
Ace Step 1.5[edit | edit source]
Civit AI Workflow - "Cover - Style transfer that preserves the semantic structure (rhythm, melody) while generating new audio with different characteristics"