ComfyUI Workflows
Browse real AI workflows from Flux, SDXL, Pony, ControlNet, video generation and commercial AI creators.
LTX 2.3 Workflows (RuneXX workflows, remade by Stefan Falkok) + NSFW (Base I2V, First-Last frame, Controlnet, Edit, Add Audio + Lipsync, Foley, Extended Video)
all base workflows for ltx 2.3, remade by me, creator is RuneXXMy ComfyUI Build - https://huggingface.co/datasets/StefanFalkok/ComfyUI_portable_torch_2.11.0_cu130_cp313_sageattention_triton/tree/mainMy NSFW TG Channel - https://t.me/stefanfalkokmothMy TG Channel - https://t.me/StefanFalkokAIMy TG Chat - https://t.me/+y4R5JybDZcFjMjFi
LTX-2.3, simple workflow. T2V, I2V, I2V external audio, FFLF
Pretty simple workflow for LTXV with 3 modes so far and quite an early version. There is no prompt enchancer; in my opinion, it's useless garbage as a tool, and text LLM work mostly works very slow in ComfyUI. There is no latent upscaling. Almost the same useless garbage.Setting the full resolution will be faster and better than half resolution and latent upscaling with an interpolated result, but it requires resources like second generation. If your system is not powerful enough for the full resolution you set, almost certainly the server will run out of memory on the latent upscale stage. And that's even worse in IMG2video cases with the pixel upscaler itself.The workflow works fine on my 8GB VRAM videocard at 720p; I haven't tried higher resolution yet on FP8 dev model with distilled lora. There is a reason why VAE are separated into loaders and files, I haven't finished that part fully. Controlling it is the same as any other my workflow. If you try even one of my workflows, it will be similar.
![[LTX 2.3] 4-Image Reference to Video (Manual Control + OmniNFT LoRA)](https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/2911664d-7470-4be9-b81a-6a4bc744f189/original=true/132770096.jpeg)
[LTX 2.3] 4-Image Reference to Video (Manual Control + OmniNFT LoRA)
๐ Run this Workflow for FREE on Runninghub: https://www.runninghub.ai/post/2061355424439431169/?inviteCode=rh-v1497 (๐ Tip: Sign up via the link above to get 1,000 free credits + 100 daily login points!) ๐ฅ Watch the Full Video Tutorial & Review: ๐ OverviewReady to push LTX 2.3 to its absolute breaking point? This workflow is designed to handle 4 reference images, attempting to mimic the advanced multi-image capabilities of closed-source giants like Grok.Because pushing 4 images significantly increases the risk of consistency breakdown, this workflow uses Manual Nodes for the reference inputs. This gives you absolute granular control over how the global prompts and segmented prompts interact.๐ Key FeaturesManual prompt relay Node Architecture: Built for power users who want strict control over 4 separate image indexes and text prompts.OmniNFT LoRA Included: Essential for maintaining character body proportions when feeding the model too much visual data. It helps keep the aesthetics clean even under heavy load.
LTX 2.3 3-Reference Image-to-Video
๐ Run this Workflow for FREE on Runninghub: https://www.runninghub.ai/post/2061372745287553025/?inviteCode=rh-v1497(๐ Tip: Sign up via the link above to get 1,000 free credits + 100 daily login points!) ๐ฅ Watch the Full Video Tutorial & Review: ๐ OverviewMissing the multi-image referencing features from closed-source models like Seedance and Grok? This experimental workflow pushes the open-source LTX 2.3 model to handle 3 reference images simultaneously.By utilizing the smart nodes from Prompt Relay, we assign equal weights (1.0) and sequential index frames to smoothly blend multiple visual concepts into a single video generation.๐ Key FeaturesSmart Prompt Relay: Uses the intuitive Prompt Relay smart node for clean and easy 3-image setup.Open-Source Alternative: A great starting point to mimic commercial multi-image video models.
Image to Image Video With Alpha Layer (WEBM format, wan2.2)
The Git Repo is here: https://github.com/therealkove-wq/img2imgVideoTransparency# img2img Transparent Video โ ComfyUI WorkflowGenerates videos with a transparent background (WEBM + alpha channel) from a start and end frame using Wan 2.2 I2V with LightX2V LoRAs and rembg AI background removal.---## How it works1. You provide a start frame and an end frame (PNG images)2. The model generates the in-between frames3. rembg AI strips the background from every frame4. Output is a WEBM file with a real alpha channel, ready for compositing---## Required ModelsAll models are from [Comfy-Org/Wan_2.2_ComfyUI_Repackaged](https://huggingface.co/Comfy-Org/Wan_2.2_ComfyUI_Repackaged) on HuggingFace.| File | Destination folder ||------|--------------------|| wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors | ComfyUI/models/diffusion_models/ || wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors | ComfyUI/models/diffusion_models/ || wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors | ComfyUI/models/loras/wan/ || wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors | ComfyUI/models/loras/wan/ || umt5_xxl_fp8_e4m3fn_scaled.safetensors | ComfyUI/models/text_encoders/ || wan_2.1_vae.safetensors | ComfyUI/models/vae/ |> The workflow will show download buttons for missing models directly in ComfyUI.---## Custom Node SetupTwo custom node packages are required.### 1. wan_alpha_nodes (this repo)Provides the SaveWebmWithAlpha node used to export the final video.Copy the two files from this repo into ComfyUI:```ComfyUI/custom_nodes/wan_alpha_nodes/โโโ wan_alpha_nodes.pyโโโ init.py```### 2. ComfyUI-rembgProvides the RembgBackgroundRemover node used for AI background removal. Install it via ComfyUI Manager or clone it into ComfyUI/custom_nodes/.---## Using the Workflow1. Open ComfyUI and load kove_img_to_img_transparent.json2. In the Start Frame node, upload your starting image3. In the End Frame node, upload your ending image4. Edit the Positive Prompt (see Prompting section below)5. In the WAN I2V Conditioning node, set width, height, and length (frame count)6. In the Save node, set your output filename and frame rate7. Click QueueThe output WEBM file will appear in ComfyUI/output/.---## Key Parameters| Parameter | Node | Notes ||-----------|------|-------|| Width / Height | WAN I2V Conditioning | Resolution of the output video. Default: 1024ร1024 || Length | WAN I2V Conditioning | Number of frames. 61 frames @ 30 FPS โ 2 seconds || Filename | Save | Output file name. Existing files with the same name are overwritten without warning. || Frame Rate | Save | Playback FPS of the output WEBM || Codec | Save | VP8 (faster) or VP9 (better compression) || Quality | Save | 0โ100. Default 20 is a good starting point |---## PromptingAlways include this phrase in the positive prompt:> "The character is isolated against a transparent background."Without it, rembg has less context and background removal quality drops.The negative prompt is pre-filled with blurry, distorted, low quality, static, no motion โ leave it as-is unless you have a specific reason to change it.### Example promptsIdle character animation```The character is isolated against a transparent background.The lighting is constant and even throughout the sequence.The character breathes gently and sways slightly from side to side.Hair and clothing move subtly with the sway.The character does not speak. Slow movements. Idle animation.```Floating object```The character is isolated against a transparent background.A glowing orb floats gently up and down. Soft pulsing light. Smooth looping motion.```Abstract effect```The character is isolated against a transparent background.Colorful particles swirl and drift slowly. Ethereal. Cinematic quality.```---## TroubleshootingNodes missing after loading the workflow- Confirm both custom node packages are installed and ComfyUI has been restarted- Check the ComfyUI console for import errorsBackground not fully removed- Make sure the transparent background phrase is in your positive prompt- Try a different rembg model in the RembgBackgroundRemover node (e.g. isnet-anime for illustrated characters)CUDA out of memory- Reduce width/height or frame count- The two diffusion models are large (14B fp8); 16GB+ VRAM recommendedOutput file is overwritten unexpectedly- Change the filename in the Save node before each runWEBM won't play in browser/editor- Switch codec to VP9 in the Save node for broader compatibility
LTX-2.3 DEV/DIST - IMAGE to Video and TEXT to Video with Ollama/RTX VSR/LTX-Director
V2.6 LTX-2.3 DEV & Distilled Video with Audio + LTX Director workflowMinor update to LTX2.3 Image to Video with Ollama workflow:sampling Preview implementedNAG to allow negative prompt for Distilled model (CFG=1)Included a Workflow supporting LTX Director node, which is awesome, it supports:First, Mid, Last or whatever frame, basically any frame is a keyframePrompt Relay which allows to generate frames prior to Input Imagemore precise editing: What shall happen when in the clip?Audio Import to create lipsynced clipsImage or Text to Video, even both in one processmore Info: https://github.com/WhatDreamsCost/WhatDreamsCost-ComfyUIrecommend to watch the creatorยดs youtube video in above link for how to.=> added a longer example video at the end.V2.5 LTX-2.3 DEV & Distilled Video with AudioImage to Video and a Text to Video workflow, both can use own Prompts or Ollama generated/enhanced prompts.works with latest LTX 2.3 Distilled model (8 steps, CFG=1) or Dev model (20 steps, CFG=3)Updated the processing for DISTILLED and DEV model, select the DIST or DEV model in loader node and switch to dedicated DIST or DEV processing pipeline, so each model has its own processing.DIST model pipeline: Standard Guider and Basic Scheduler, follows the manual sigmas issued by LightricksDEV model pipeline: MultiModal Guider and LTX Scheduler + Distilled Lora on latent upscalerIncluded a workflow version with "RTX Video Super Resolution" node, which upscales videos in highspeed.Tip: With latest Comfy and LTX updates, the processing got faster for me, so I can increase the scale_by in sampler node from 0.5 to 0.6 or higher to have crisper videos with minor impact on render time.V2.3 LTX-2.3 DEV & Distilled Video with AudioDownloads for LTX 2.3:update : April 14th 2026 : Lightricks has updated their LTX 2.3 distilled model to 1.1 (and Lora):Model (1.1 fp8 _scaled by Kijai): https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/diffusion_modelsdist. Lora 1.1 : https://huggingface.co/Lightricks/LTX-2.3/tree/mainLTX-2.3 Distilled & Dev Models (fp8_scaled): https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/diffusion_modelsTextencoder1: (fp8_e4m3fn, same as LTX-2): https://huggingface.co/GitMylo/LTX-2-comfy_gemma_fp8_e4m3fn/tree/mainTextencoder2: (projection_bf16): https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/text_encodersVideo & Audio Vae: https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/vaePreview VAE (taeltx2_3): https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/vaeLoras:Spartial upscaler (x2-1.1): https://huggingface.co/Lightricks/LTX-2.3/tree/mainDistilled Lora for upscaler (lora.384): https://huggingface.co/Lightricks/LTX-2.3/tree/mainSmaller, alternative Desitilled Lora by Kijai: https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/lorasDetailer Lora (same as LTX-2): https://huggingface.co/Lightricks/LTX-2-19b-IC-LoRA-Detailer/tree/mainOllama Model (prompt only, fast): https://ollama.com/mirage335/Llama-3-NeuralDaredevil-8B-abliterated-virtuosoalternative model with Vision (reads input image+prompt, slower): https://ollama.com/huihui_ai/qwen3-vl-abliteratedother model with Vision (great for I2V): https://ollama.com/huihui_ai/qwen3.5-abliteratedsmaller LTX 2.3 GGUF Dev or Dist. models work as well. (replace Checkpoint loader node with Unet loader node from this custom node: https://github.com/city96/ComfyUI-GGUF ):models: https://huggingface.co/unsloth/LTX-2.3-GGUF/tree/mainsave to models/unet/V1.5 LTX-2 DEV Video with Audio including latest ๐ ๐ ฃ๐ ง Multimodal GuiderImage to Video and a Text to Video workflow, both can use own Prompts or Ollama generated/enhanced prompts.Replaced the Guider node with latest Multimodal Guider node, see more details in WF notes or here: https://ltx.io/model/model-blog/ltx-2-better-control-for-real-workflows Before we had 1 CFG parameter for audio and video. With multimodal guider, we now can tweak audio and video seperately with even more parameters...added a Power Lora Loader node to inject further Lorasuse Image to Video Adapter Lora to improve motion for I2V: https://huggingface.co/MachineDelusions/LTX-2_Image2Video_Adapter_LoRa/tree/mainreplaced a node to no longer require comfymath custom nodesV1.0 LTX-2 DEV Video with Audio:Image to Video and a Text to Video workflow with own Prompts or Ollama generated/enhanced prompts.setup for the LTX2 Dev model.uses Detailer Lora for better quality and LTX tiled VAE to avoid OOM and visual grids2 pass rendering (motion+upscale). Upscale process uses distilled and spatial upscale Lorasetup with latest LTXVNormalizingSampler to increase video & audio quality.Text to Video can use dynamic prompts with wildcards.Download LTX-2 Files: (Workflow V1.0 and V1.5 only)Find Model/Lora Loader nodes within Sampler Subgraph node.- LTX2 Dev Model (dev_Fp8): https://huggingface.co/Lightricks/LTX-2/tree/main- Detailer Lora: https://huggingface.co/Lightricks/LTX-2-19b-IC-LoRA-Detailer/tree/main- Distilled (lora-384) & Spatial upscaler Lora: https://huggingface.co/Lightricks/LTX-2/tree/main- VAE (already included in above dev_FP8 model, but needed if you go for GGUF models): https://huggingface.co/Lightricks/LTX-2/tree/main/vae- Textencoder (fp8_e4m3fn): https://huggingface.co/GitMylo/LTX-2-comfy_gemma_fp8_e4m3fn/tree/main- Image to Video Adapter Lora (more motion with I2V): https://huggingface.co/MachineDelusions/LTX-2_Image2Video_Adapter_LoRa/tree/mainSave Location:๐ ComfyUI/โโโ ๐ models/โ โโโ ๐ checkpoints/โ โ โโโ ltx-2-19b-dev-fp8.safetensorsโ โโโ ๐ text_encoders/โ โ โโโ gemma_3_12B_it_fp8_e4m3fn.safetensorsโ โโโ ๐ loras/โ โ โโโ ltx-2-19b-distilled-lora-384.safetensorsโ โโโ ๐ latent_upscale_models/โ โโโ ltx-2-spatial-upscaler-x2-1.0.safetensorsโ โโโ ๐ Clip/โ โโโ ltx-2.3_text_projection_bf16.safetensorsCustom Nodes used:https://github.com/Lightricks/ComfyUI-LTXVideohttps://github.com/rgthree/rgthree-comfyhttps://github.com/yolain/ComfyUI-Easy-Usehttps://github.com/stavsap/comfyui-ollamahttps://github.com/kijai/ComfyUI-KJNodeshttps://github.com/Comfy-Org/Nvidia_RTX_Nodes_ComfyUI (RTX VSR Version)Text 2 Video only: https://github.com/KoinnAI/ComfyUI-DynPromptSimplifiedLTX Director only: https://github.com/WhatDreamsCost/WhatDreamsCost-ComfyUIOllama help:Install Ollama from https://ollama.com/download a model: Go to a model page, chose a model , then hit the copy button, i.e. https://ollama.com/huihui_ai/qwen3-vl-abliteratedopen terminal and paste the model name, i.e.: ollama run huihui_ai/qwen3-vl-abliteratedmodel will be downloaded and can be selected in green comfy node "Ollama Connectivity". Hit "Reconnect" to refresh.Example longer Video
ใWAN2.1ใStartEndFrames
โจ WAN2.1 โ StartEndFrames โ Simple WorkflowA clean, all-in-one WAN image-to-video workflow built entirely with the UmeAiRT Toolkit for ComfyUI.Only 12 nodes. No spaghetti wires. Just load your model, write your prompt, and hit generate.โ ๏ธ IMPORTANT โ Nodes 2.0 RequiredThis workflow is built for the Nodes 2.0 (Vue) interface of ComfyUI. If you don't enable it, the workflow may have display problems.How to activate Nodes 2.0:Open ComfyUIGo to Settings (โ๏ธ icon, bottom-left)Find "Use Nodes V2 (Vue)" and toggle it ONRefresh the pageLoad the workflowIf you prefer the classic interface, check out my Legacy version of this workflow instead (link).๐ฏ FeaturesText-to-video generationAutomatic download of models in auto versionBuilt-in SeedVR2 upscaler โ high-quality tiled upscaling (toggleable on/off) Slower than a classic upscaler, but significantly better qualityFull metadata embedding โ your images are saved with all generation parameters, ready for online publishing and remixing3 LoRA slots โ with individual on/off toggles and strength control and you can connect as many other lora modules to each other for as many LoRA as you want.๐ฆ Custom Node RequiredOnly one custom node to install:๐ ComfyUI-UmeAiRT-ToolkitInstall via ComfyUI Manager (search "UmeAiRT") or use the UmeAiRT Auto-Installer.The Toolkit packages everything internally โ upscaler, face detailer, metadata saver. No other custom nodes needed.๐ Files you need (in manual version)Recommendation :>24 gb Vram: base or Q8_016 gb Vram: Q5_K_S<12 gb Vram: Q4_K_SFor base versionVACE Model: wan2.1_vace_14B_fp8_e4m3fn.safetensors or wan2.1_vace_1.3B_fp16.safetensorsIn models/diffusion_modelsCLIP: umt5_xxl_fp8_e4m3fn_scaled.safetensorsin models/clipFor GGUF versionVACE Quant Model: Wan2.1-VACE-14B-QX_0.ggufIn models/diffusion_modelsQuant CLIP: umt5-xxl-encoder-QX.ggufin models/clipVAE: wan_2.1_vae.safetensorsin models/vaeANY upscale model (depreciated):Realistic : RealESRGAN_x4plus.pthAnime : RealESRGAN_x4plus_anime_6B.pthin models/upscale_models
ใWAN2.2ใTXT to VIDEO
โจ WAN2.2 โ Text to video โ Simple WorkflowA clean, all-in-one WAN text-to-video workflow built entirely with the UmeAiRT Toolkit for ComfyUI.Only 12 nodes. No spaghetti wires. Just load your model, write your prompt, and hit generate.โ ๏ธ IMPORTANT โ Nodes 2.0 RequiredThis workflow is built for the Nodes 2.0 (Vue) interface of ComfyUI. If you don't enable it, the workflow may have display problems.How to activate Nodes 2.0:Open ComfyUIGo to Settings (โ๏ธ icon, bottom-left)Find "Use Nodes V2 (Vue)" and toggle it ONRefresh the pageLoad the workflowIf you prefer the classic interface, check out my Legacy version of this workflow instead (link).๐ฏ FeaturesText-to-Image generationAutomatic download of models in auto versionBuilt-in SeedVR2 upscaler โ high-quality tiled upscaling (toggleable on/off) Slower than a classic upscaler, but significantly better qualityFull metadata embedding โ your images are saved with all generation parameters, ready for online publishing and remixing3 LoRA slots โ with individual on/off toggles and strength control and you can connect as many other lora modules to each other for as many LoRA as you want.๐ฆ Custom Node RequiredOnly one custom node to install:๐ ComfyUI-UmeAiRT-ToolkitInstall via ComfyUI Manager (search "UmeAiRT") or use the UmeAiRT Auto-Installer.The Toolkit packages everything internally โ upscaler, face detailer, metadata saver. No other custom nodes needed.๐ Files you need (in manual version)For base versionT2V Model : wan2.2_t2v_high_noise_14B_fp8_scaled.safetensors and wan2.2_t2v_low_noise_14B_fp8_scaled.safetensorsIn models/diffusion_modelsFor GGUF versionT2V Quant Model : wan2.2_t2v_high_noise_14B_QX.gguf and wan2.2_t2v_low_noise_14B_QX.ggufIn models/unetFor lightx2V versionlightx2V LoRA : Wan2.2-Lightning_T2V-A14B-4steps-lora_HIGH_fp16.safetensors and Wan2.2-Lightning_T2V-A14B-4steps-lora_LOW_fp16.safetensorsCLIP: umt5_xxl_fp8_e4m3fn_scaled.safetensorsin models/clipVAE: wan_2.1_vae.safetensorsin models/vaeANY upscale model:Realistic : RealESRGAN_x4plus.pthAnime : RealESRGAN_x4plus_anime_6B.pthin models/upscale_models
ใWAN2.2ใIMG to VIDEO
โจ WAN2.2 โ Image to video โ Simple WorkflowA clean, all-in-one WAN image-to-video workflow built entirely with the UmeAiRT Toolkit for ComfyUI.Only 12 nodes. No spaghetti wires. Just load your model, write your prompt, and hit generate.โ ๏ธ IMPORTANT โ Nodes 2.0 RequiredThis workflow is built for the Nodes 2.0 (Vue) interface of ComfyUI. If you don't enable it, the workflow may have display problems.How to activate Nodes 2.0:Open ComfyUIGo to Settings (โ๏ธ icon, bottom-left)Find "Use Nodes V2 (Vue)" and toggle it ONRefresh the pageLoad the workflowIf you prefer the classic interface, check out my Legacy version of this workflow instead (link).๐ฏ FeaturesText-to-Image generationAutomatic download of models in auto versionBuilt-in SeedVR2 upscaler โ high-quality tiled upscaling (toggleable on/off) Slower than a classic upscaler, but significantly better qualityFull metadata embedding โ your images are saved with all generation parameters, ready for online publishing and remixing3 LoRA slots โ with individual on/off toggles and strength control and you can connect as many other lora modules to each other for as many LoRA as you want.๐ฆ Custom Node RequiredOnly one custom node to install:๐ ComfyUI-UmeAiRT-ToolkitInstall via ComfyUI Manager (search "UmeAiRT") or use the UmeAiRT Auto-Installer.The Toolkit packages everything internally โ upscaler, face detailer, metadata saver. No other custom nodes needed.๐ Files you need (in manual version)For base versionI2V Model : wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors and wan2.2_i2v_low_noise_14B_fp8_scaled.safetensorsIn models/diffusion_modelsCLIP: umt5_xxl_fp8_e4m3fn_scaled.safetensorsin models/clipFor GGUF versionI2V Quant Model : wan2.2_i2v_high_noise_14B_QX.gguf and wan2.2_i2v_low_noise_14B_QX.ggufIn models/unetQuant CLIP: umt5-xxl-encoder-QX.ggufin models/clipFor speed versionlightning LoRA : Wan2.2-Lightning_I2V-A14B-4steps-lora_HIGH_fp16.safetensors and Wan2.2-Lightning_I2V-A14B-4steps-lora_LOW_fp16.safetensorsFor AIO versionI2V FAST AIO Model : wan2.2-i2v-rapid-aio.safetensorsIn models/checkpointVAE: wan_2.1_vae.safetensorsin models/vaeANY upscale model:Realistic : RealESRGAN_x4plus.pthAnime : RealESRGAN_x4plus_anime_6B.pthin models/upscale_models
ใWAN2.1ใIMG to VIDEO
โจ WAN2.1 โ Image to video โ Simple WorkflowA clean, all-in-one WAN image-to-video workflow built entirely with the UmeAiRT Toolkit for ComfyUI.Only 12 nodes. No spaghetti wires. Just load your model, write your prompt, and hit generate.โ ๏ธ IMPORTANT โ Nodes 2.0 RequiredThis workflow is built for the Nodes 2.0 (Vue) interface of ComfyUI. If you don't enable it, the workflow may have display problems.How to activate Nodes 2.0:Open ComfyUIGo to Settings (โ๏ธ icon, bottom-left)Find "Use Nodes V2 (Vue)" and toggle it ONRefresh the pageLoad the workflowIf you prefer the classic interface, check out my Legacy version of this workflow instead (link).๐ฏ FeaturesText-to-video generationAutomatic download of models in auto versionBuilt-in SeedVR2 upscaler โ high-quality tiled upscaling (toggleable on/off) Slower than a classic upscaler, but significantly better qualityFull metadata embedding โ your images are saved with all generation parameters, ready for online publishing and remixing3 LoRA slots โ with individual on/off toggles and strength control and you can connect as many other lora modules to each other for as many LoRA as you want.๐ฆ Custom Node RequiredOnly one custom node to install:๐ ComfyUI-UmeAiRT-ToolkitInstall via ComfyUI Manager (search "UmeAiRT") or use the UmeAiRT Auto-Installer.The Toolkit packages everything internally โ upscaler, face detailer, metadata saver. No other custom nodes needed.๐ Files you need (in manual version)For base versionI2V Model : 480p or 720pIn models/diffusion_modelsFor GGUF versionI2V Quant Model :- 720p : Q8, Q5, Q3- 480p : Q8, Q5, Q3In models/unetCommon files :CLIP: umt5_xxl_fp8_e4m3fn_scaled.safetensorsin models/clipCLIP-VISION: clip_vision_h.safetensorsin models/clip_visionVAE: wan_2.1_vae.safetensorsin models/vaeSpeed LoRA: 480p, 720pin models/lorasANY upscale model:Realistic : RealESRGAN_x4plus.pthAnime : RealESRGAN_x4plus_anime_6B.pthin models/upscale_models
LipSync (Extend Loop)
ComfyUI WorkflowMaking this workflow possible required adding several new nodes, so you'll need Eclipse 3.5.17 or later. Please update Eclipse before loading the workflow.In its current setup, the workflow uses the audio duration you've configured to automatically calculate the required number of loops. To enter the loop_count manually in the Smart Folder node, simply switch the dual-switch from 2 to 1 (behind the Smart Folder Node).
Simple Reactor Video Faceswap
Fast video faceswapping, and possibly a base for more advanced workflows.On my Ryzen 4600G,16G RAM, RTX 4060 8GB:A 5 second, 608x1080 video @ 30fps, takes about 5 minutes. Versions used:ComfyUI Desktop 0.21.1Reactor 0.7.0-a2May work on other versions,but not guaranteed.Also, could break if something changes in ComfyUI or Reactor.You can always try to extend it with other nodes to generate faces, upscale,etc.
Full-Auto AI Video Translator: Multi-Speaker & BGM Separation
๐ Run this Workflow for FREE on Runninghub: https://www.runninghub.ai/post/2056650178861559810/?inviteCode=rh-v1497(๐ Tip: New users get 1000 points via the link above + 100 daily login points!)๐ฅ Watch the Full Video Tutorial (Beginner to Pro):๐ OverviewWant a complete hands-free solution to translate foreign videos? this workflow allows you to upload a raw audio or video file and automatically handles transcribing, translating, voice cloning, and audio splicing in one single run.It is perfect for bulk translating content or quickly processing dialogue with multiple distinct speakers.๐ Key Features & ArchitectureAll-in-One Processing: Simply upload your file, select your options, and let it cook. It splits the workflow into transcription, LLM translation (via Gemini API), and audio generation.80+ Languages Supported: Powered by FishAudio 2 for high-fidelity cross-lingual voice synthesis.BGM & Dialogue Separation: Its true secret weapon is keeping the original background music/audio track intact while automatically cloning and placing the translated dialogue.2-Speaker Detection: Capable of directly processing conversational audio with 2 speakers while attempting to match individual voice profiles.๐ฌ Join the Community๐ [https://discord.gg/NmBhk9Vk8b]
Post Processing IMG/VIDEO Any
simple workflow to do the post processing for mainly videos.since i get many OOM. in some parts of the video generations. i split the post processing to avoid this problem.any content of this workflows dont require checkpoint models. only his own models like rife49 for interpolation. etc.etc.etc.v1.05 - requires now my own custom nodes to solve some issues from now onshirosaki33/shiroaudio: custom nodes for comfy, related to audiov1.055 - experimental toolsshirosaki33/shirotools: many utility tools made for different situationsthe image process is just a extra.you can do by single file or batch path in both.
LTXV2.3 GGUF I2V/T2V/FFLF+IDaudioLora 12gb Vram sageattn
A simple, color-coded GGUF workflow for running LTXV-2 in ComfyUI.Designed to be straightforward and lightweight.Common IssuesError: LTXVEmptyLatentAudio โ AttributeErrorUpdate Kijai Node Pack to v1.2.8Checkerboard / patterned noise outputUpdate all nodes + ComfyUI(Outdated GGUF loader or ComfyUI build)Required Node PacksComfyUI-GGUFhttps://github.com/city96/ComfyUI-GGUFComfyUI-KJNodeshttps://github.com/kijai/ComfyUI-KJNodes (install & update it to nightly)additional install via manager๐ฆ Downloads๐งฉ Kijaiโs Repo (ALL files except diffusion models โ choose ONE LoRA)https://huggingface.co/Kijai/LTX2.3_comfy/tree/mainIncludes:Distilled LoRA (REQUIRED)Text Projection (embedding connector)๐ง Diffusion Models (GGUF)https://huggingface.co/unsloth/LTX-2.3-GGUF/tree/mainChoose based on VRAM:Q4_K_M โ lower VRAM (recommended)Q8_0 โ higher quality, more VRAMโ ๏ธ Very low quants = basically unusableIncludes:VAE (audio + video)GGUF modelsโ๏ธ Text Encoders (REQUIRED)Option 1 โ Gemma (recommended for low VRAM)https://huggingface.co/Comfy-Org/ltx-2/tree/main/split_files/text_encodersOption 2 โ Full version (higher VRAM, better quality)โก๏ธ You still ALSO need:Text Projection (from Kijaiโs repo)๐ Latent Upscalerhttps://huggingface.co/Lightricks/LTX-2/tree/main/Files:ltx-2-spatial-upscaler-x2-1.0.safetensorsltx-2-temporal-upscaler-x2-1.0.safetensorsโ ๏ธ Important NotesUse DEV GGUF diffusion models (no distilled base model yet)Distilled LoRA is still REQUIREDMost issues = outdated nodes or ComfyUI.
Wan 2.2 - All in Wan I2V (I2V, F2LF, SVI with optional F2LF, NAG, LTX for V2A, Pulse of Motion, Lora Optimizer, CFG-Ctrl, 4 modes, ...)
If you are having trouble finding some nodes, here are some of the repos of the nodes people are having trouble finding:Pulse of Motion: https://github.com/akashzeno/ComfyUI-PulseOfMotion/tree/mainThe fpsframedrop is one I made to solve the problem of potentially having different FPS when combining 2 videos to keep the same speed, I added it to comfy registry but apparently that doesn't work (yet?), here it is:https://github.com/dailydoseofaiart/ComfyUI-FPSFrameDrop/blob/main/README.mdSVI with FLF:https://github.com/Well-Made/ComfyUI-Wan-SVI2Pro-FLF(only relevant for v1:) PrismAudio: https://github.com/richservo/ComfyUI-PrismAudio.gitIt was a bit of a pain to install for me, I had AI help me.v2.1 Release:Some important bug fixes, including:The workflow no longer cuts off 4 frames at the beginning of every video (only SVI now, as it should)Fixed an issue causing the same video to be resized to different sizes, preventing generation in SVIv2 Release:A complete visual and functional overhaul, the features now include:- Regular I2V- F2LF (I2V but with a chosen end frame)- Video extension with SVI 2.0 Pro- Adding audio to an existing video or a just-generated video using LTX-2.3- NAG (Negative Attention Guidance) to be able to use a negative prompt even with CFG 1.0 (though it also works with higher CFG)- CFG-Ctrl/SMC-CFG (A node for potentially better prompt-following)- Pulse of Motion to automatically adjust the speed of the video to a natural-looking one (but also manual FPS control)- LoRA optimizer node for better combination of LoRAs (separate from regular LoRA loader so you can choose whatever works better)v1:This workflow is configured so that you can use it like a normal I2V workflow, indicated by red nodes. Additional features are indicated with cyan nodes.I also added notes everywhere to explain everything. I always dislike big workflows where I don't know where to even look and tweak or not, so I tried to make it intuitive with the coloring and explaining it all.The highlights of this workflow that make it different from the standard Wan 2.2 I2V workflow are these:- Ability to crop images easily in here and stitch 2 images together easily in case you need that- By default: Uses "Pulse of Motion" to automatically determine the best framerate for a natural real time video speed. You can override it if it doesn't work well for a video. The workflow also interpolates the video at that speed.- Audio Generation using PrismAudio. Uses the video prompt by default, but I recommend overriding it with a custom prompt for the audio as described in the note by the node.- Lora optimizer node. If you have multiple loras, this node optimizes them to work well together. More on that in the notes in the subgraph. Turned off by default.- Cfg-ctrl node. This improves prompt adherence when you use cfg. Turned off by default. You shouldn't use the lightx2v on the high noise when activating this.I'm thinking of including SVG and NAG in the next version as well. Let me know if you want that. Maybe I will just do NAG because I don't really know how to integrate SVG without making it a completely separate section, in which case I might as well just make a different workflow.
videoLab - LTX 2.3
*When copy+pasting the workflow Audio VAEs disconnect and Math Expression #74 has to be reconnected from it's port b to Frame Rate*Check out my other workflows for swap and upscale techniquesLTX 2.3 Super speaks โ but now it speaks in your voice.Your Custom VoiceLTX audio merged inLip sync stabilityThis workflow combines voice cloning with LTX 2.3's native audio generation. One prompt produces two synchronized audio tracks โ cloned dialogue and AI-generated background sound โ merged and conditioned into the video latent.Your voice. Your script. LTX's world. One prompt.LTX 2.3 โ AI Video Workflow (part 4 of Lab series)Four inputs. One cinematic video.Drop in an image, a voice sample, your scene description, and your character's dialog โ the workflow takes it from there.โ๏ธ Script-Driven Dialog Write exactly what you want your character to say in the Dialog Script node. Your words, your story โ the workflow handles all the technical execution automatically.๐ฌ Scene Description Describe the scene, the mood, the environment. The AI reads this alongside your dialog and builds everything else โ no prompt engineering knowledge needed.๐ค Fully Automatic Prompting Two local AI enhancers run silently in the background. One writes a complete structured video generation prompt from your scene and dialog. The other crafts precise voice delivery direction โ pace, tone, emotion, texture โ fed directly into the voice engine. You never write a prompt manually.๐๏ธ VoX Voice Cloning Provide any short voice sample and VoxCPM2 clones it. Your character speaks your script in that voice, with the AI-generated delivery direction shaping how it's performed โ not just what is said, but how.๐ Dual Voice Reinforcement The video and audio pipelines are locked together. The video prompt instructs the model on lipsync behavior while the audio direction drives matching emotional delivery through VoX โ a unified performance from both sides simultaneously.โ๏ธ GGUF & Safetensors Compatible Run whichever model format your hardware prefers โ GGUF quantized or full safetensors, both supported out of the box.
Simple Video Combiner Workflow For Extending Videos
Just a simple tool i use to make longer generations without having a super complicated workflow. I got tired of dowloading all the big giant crazy workflows, having them oom, not work, or explode after 45 minutes of genrating. and just wanted a super simple solution to making longer videos.. So.. boom... this!!!basically Just take your existing video generation workflow.after the Vae Decode (before video combine) add a get last frame node. from the get last frame node attach a save image node. use that very last frame to generate your next video.. Then Open this tool. upload video 1upload video 2push run.. Boom longer smooth video with no degradation!(instructions included in workflow)
Frame Interpolation workflow - smoother video playback
Adds frames to video, increasing the frames per second (fps), for smoother playback.Audio from the original can be piped into the final video or extracted and saved separately for further post processing.Auto calculates output frame_rate based on original frame rate and frame multiplier.
LTX2.3 Sulphur Multi Frame Video Gen(NSFW)
A workflow uses the latest LTX2.3 Sulphur and Prompt Relay and First and last multi- frames to generate high-quality videos.Other model details are within the workflow.Try the workflow online for free: https://www.runninghub.ai/post/2051993332863254530Please check the video tutorial before you use this workflow:ย https://rumble.com/v79ibo0-comfyui-ltx2.3-sulphur-video-generation-detailed-guige.html
LTX-2.3, simple workflow. T2V, I2V, I2V external audio, FFLF
Pretty simple workflow for LTXV with 3 modes so far and quite an early version. There is no prompt enchancer; in my opinion, it's useless garbage as a tool, and text LLM work mostly works very slow in ComfyUI. There is no latent upscaling. Almost the same useless garbage.Setting the full resolution will be faster and better than half resolution and latent upscaling with an interpolated result, but it requires resources like second generation. If your system is not powerful enough for the full resolution you set, almost certainly the server will run out of memory on the latent upscale stage. And that's even worse in IMG2video cases with the pixel upscaler itself.The workflow works fine on my 8GB VRAM videocard at 720p; I haven't tried higher resolution yet on FP8 dev model with distilled lora. There is a reason why VAE are separated into loaders and files, I haven't finished that part fully. Controlling it is the same as any other my workflow. If you try even one of my workflows, it will be similar.
LTX Video 2.3 version 1.1 GGUF Image to Video
A GGUF version of the Image to Video workflow of the newest LTX Video version 2.3 in subversion 1.1.ย HowtoDrag an image into the image nodes, adjust the prompt, and press queue. The rest should fit.DescriptionAn image to video ComfyUI workflow with LTX Video 2.3 subversion 1.1. The nodes are wired and visible in the traditional way and grouped together. No set and get nodes, no subgraph. I like it simple. It produces two videos, one with audio, one without.TimeThe example video in a resolution of 1024 x 576 was generated in 240 seconds at Windows 11 at an AMD Radeon AI PRO R9700. I have yet to test it on Linux, which is usually faster by a third.RequirementsThis workflow was generated with 32 gb vram. There is a ram saver node involved, which you can turn on. So you should be able to go lower. The bottleneck that even made my rig freeze is the VAE decode.
Wan Vace 2.1 Flow Archive
This is an archive of my vace flows, most are over a year old. It will take time for me to get them all in working order for release but this is where they will end up.Archive Flow 1simplified wan 2.1 vaceThis workflow was designed to simplify video to video copies with wan 2.1 vaceThis shows a few tricks many may not knowArchive Flow 2 (every single video i released for vace 2.1 used this flow to fix faces)This can fix any faces, from any video, not just vace.Face fixing and enhance flowThis was a tutorial flow i made long ago to show how to fix ai mush faces with vace.Vace face fixing is not as hard as many think and can be done with 1.3b model and bbox masks. Its the best face fixing still to this day for videos. even with 1.3b.there is one main trick in this flow you will find that enables near perfect face fixing that i dont know anyone else that uses.This crops the face and does not process the whole videoThis has extend abilities to do the vace flows batches or to break larger videos into batches to process.I cleaned it up and replaced many nodes that were outdatedNotes should explain It was a tutorial flow.Archive Flow 3 Full vaceVace work flow.This is the main flow i used for over a year. I cleaned it up A LOT. I delt with a mess here for a long time LOL. This can do any of the videos i have done. But you will need to learn it. I spent over a year learning and this flow is about as strait forward as i get when showing you how its done. I tried automating many things but each video i made needed me to adjust. It needed me to direct it. And it took a lot of time to do.This is not my most advanced flow for vace as it does not include inpaint but i never used inpainting. I make new videos not change them. I used this more than any other flow.I dont know if this is worth the time for many. LTX can do some crazy stuff very fast. This takes time. This can still do more than LTX when it comes to action and i think thats where people will get use of this. I cant replicate what i did here with LTX or any other local model, what i do with LTX is very different.I will answer any questions in the comments if you have them. I probably dont explain things well enough.If anything is wrong, errors. Post and i will see to fix them.I am including a few copies i made of videos. I cant share real videos on civitai.These copies are some of my best. They can be copied themselves.Good luck everyone.
Wan 2.2 T2V, I2V, FFLF, lora manager. WanVideoWrapped extension based.
Pretty simple workflow for Wan2.2, mostly about the 14B model (I didnโt try the 5B). Nothing special, just finally completed the workflow for myself, cleared up all the mess, with some potential for upgrades.
Consistent AI Portrait Animator (AnimateDiff + IPAdapter Plus)
Professional AI Pipeline: Identity-Preserved Temporal AnimationThis is a custom ComfyUI workflow engineered for high-end digital character creation. It provides a robust solution for the "identity drift" problem in AI video, ensuring 100% facial consistency across dynamic frames.Technical Architecture & Features:Neural Identity Mapping: Implements a precise identity-locking mechanism using IPAdapter Plus to maintain the user's unique facial features.Temporal Coherence Engine: Utilizes a customized AnimateDiff-Evolved setup to generate fluid, flicker-free movement.Modular Phase Design: The workflow is logically structured into three distinct phasesโData Ingestion, Motion Synthesis, and High-Res Refinementโallowing for granular control over the output.Production-Ready Output: Optimized for generating high-fidelity video sequences (H.264/MP4) suitable for professional digital portfolios.Developer's Note: This project serves as a technical demonstration of advanced AI pipeline integration, focusing on the intersection of portrait photography and temporal neural synthesis.
Vace video copy
This was my attempt to do the opposite of what i normally do. I nomally GIVE options, many options.. and those options confuse people. I noticed most dont want to be a AI video artist. They want a simple way to make the video they want and dont want to play with options that they dont fully understand. I get it, its not me... but i get it. I have lived inside comfy every day for 3+ years and sometimes forget not everyone wants to understand it all like me.https://civitai.red/models/2570937/wan-vace-21-flow-archiveThis takes any choice that you can make and does what i would do for 80% of videos without asking you.There are still options but as few as i could, i decided for you for many of the things that would normally be a choice.This means it will make your video, or it wont... if it wont you have very few options you can play with to get it working.This will fail hard on flips. upside down is VERY HARD and usually requires a custom depth map to follow properly.This will also fail hard on backgrounds with too many strait lines. you will know it when you see it, as it combines the pidnet control with the video making seriously werid results. Increasing the res of the control can help with this but if this happens its the video and probably wont copyI dont think this copies actual porn. I tried at one point and it becomes a mess, porn is best done with loras i think and i also think wan 2.2 image to vid works best for this purpose.Contrast shifting of batches is a vace thing. Its there and not going away trust me i have tried everything. LCM and eular a tend to the least but its still there, color matching only works sometimes, and the strength of the distilled lora and rank effect this. The main problem is static backgrounds. They get worse over time as vace just cant leave well enough alone and tries to "enhance" the background each batch making them darker as it goes. I found no way around this even with complex masking controls. Moving backgrounds can solve this even just small movements like a shaky cam wont drift nearly as much. Lowering the distilled lora strength for second batch can help but there is no real rules here each video acts differently.This has little memory of previous batches. The problem many dont see is when camera moves or the subject covers a background object for the guide frames... it wont know what was behind them. the new batch will generate something different because it has no idea what the previous batch had there. No getting around this one in this flow as its a big option that needs the full vace flow to work properly. And yes that means inserting frames memories(single frame with info needed) into vace can be done at the start of the guide frames allowing it to remember things it never could before for consistancy across batches. This is something i did not start doing until very late because it was never something i noticed until i started doing much more complex motions.That being said... this still does most things that the others do, it just asks you very few questions and just does it.most small videos should work. Dancing, skipping, weightlifting, workouts, etc barring the backgrounds being too busy.Because i cant share real people videos understandablyI am giving you an image and video of my ai muse to test and copy, dont fuck with her. I worked hard on her image as sexy but not nsfw. Its the only thing i will fight someone on. Get your own muse please.The example uses frame count of 93 with 8 guide frames. the last batch(4) the guide frames change to 7.All batches that come out should merge with the previous one without much issues or be loaded into my face fixing flow.
LtxMTV - 2-Minutes Music Video Generator
LtxMTV is a APP MODE 2-Minutes Music Video Generator(for best results use 50FPS and be patient :) )using Ace1.5 / 4b and Ltx2.3 Distilled FP8both downloadable directly from Comfyui.It's using KJNodes as Custom Nodes.Options:- Auto-Image Resize to 480p or 544p- FPS- BPM- Lyrics- Musical Style- Language- Time SignatureWrite your lyrics, pick your music/singer styleChoose 2 Images, Each scene duration is 20 seconds for a total 6 scenes with alternating between those 2 images as a start frame.Making a good quality music video with minimal effort.for more options, go in Work Flow mode.Tested on a 14900k / nvidia 5080 with 64go Ram
LTX 2.3 - Clear Learning Workflow
This workflow is meticulously designed to demystify the LTX 2.3 generation process. Instead of providing a "black box" solution, this setup prioritizes architectural clarity, making it the perfect tool for creators who want to understand, learn, and eventually replicate high-end AI video generation.Key Features & Educational ValueLogical Modular Structure: Every stage of the processโfrom model loading to final video renderingโis laid out in a linear, easy-to-follow path. It separates the Video and Audio VAE processes, allowing you to see exactly how the multimodal generation is constructed.Sage Attention Integration: Features the PathchSageAttentionKJ node, demonstrating how to optimize memory and performance within the LTX architecture.Advanced Conditioning: Gain full control over the output using the dedicated LTXVConditioning and LTX2_NAG (Negative Adversarial Guidance) nodes. This section shows how to balance CFG and guidance scales to achieve professional results.Two-Stage Upscaling: Learn how to use the spatial-upscaler model effectively. The workflow includes a clear path from the initial latent generation to the spatial upsampling stage, ensuring high-fidelity 5-second clips.Native Audio Management: Includes a dedicated audio latent path, showing you how to generate synchronized ambient sound or dialogue without external tools.Precise Temporal Control: The workflow uses custom calculators to manage FPS and frame duration (e.g., 5s @ 24fps), teaching you the math behind video latency and frame consistency.Why use this Workflow?Most workflows prioritize speed over understanding. This one is different. It is built as a blueprint: a clean, "read-only" style layout that teaches you the importance of each connection. Whether you are a beginner looking for a solid starting point or an expert seeking a clean template for LTX 2.3, this workflow provides the precision and transparency needed for professional-grade AI cinematography.
LTX2.3 Video Gen with Low VRAM(NSFW)
A workflow uses the latest LTX2.3 Dasiwa to generate high-quality videos with low VRAM and speed. Thanks to the author.https://civitai.red/models/2543443/dasiwa-ltx-23-or-lightspeed-or-safetensorsOther model details are within the workflow.Try the workflow online for free: https://www.runninghub.ai/post/2045438512022949889Please check the video tutorial before you use this workflow:ย https://rumble.com/v78pszc-comfyui-ltx2.3-dasiwa-video-generationunlimited.html
LTX2 Img and Audio to Video
A comprehensive production-grade pipeline designed for the LTX-2 model. It specializes in generating high-fidelity video by combining a source image and an audio track to create synchronized content, such as music videos with lip-syncing and dance.Key Features & ArchitectureThe workflow is organized into distinct logical stages using subgraphs to manage complexity and optimize hardware resources:Multimodal Input Processing:Image Handling: Uses ImageResizeKJv2 to prepare a source image, which acts as the visual foundation for the video.Audio Integration: Employs a VHS_LoadAudioUpload node to bring in external audio files, which guide the timing and motion of the generation.Dual-Stage Sampling Pipeline:Stage 1 (Initial Generation): Focuses on establishing the core motion and structure.Stage 2 (Refinement): A secondary pass that refines the video and audio latents for higher quality.VRAM Optimization:Gemma API Text Encode: Instead of loading the massive Gemma-3 12B model locally, this workflow uses an API-based text encoder. This significantly reduces local VRAM requirements, allowing the workflow to run on GPUs with as little as 12GB to 16GB.Creative Controls:Camera LoRAs: Includes dedicated slots for LTX-2 Camera Control LoRAs (e.g., Dolly Left), allowing for precise cinematic movement.Latent Upscaling: Incorporates a spatial upscaler to enhance the resolution of the final output.
LTX 2.3 FLF2V
The LTX 2.3 KJ Stripped 1.1 workflow is a specialized audio-visual (AV) generation pipeline designed to interpolate between two static images. It utilizes a "Guided" architecture to ensure the generated video adheres strictly to a designated start and end point.Core Architecture & ProcessingModel Loading: The workflow employs the LTX-Video 2.3 22B Distilled transformer model via a specialized KJ loader node.Dual-Modality Processing: It initializes both an EmptyLTXVLatentVideo and an LTXVEmptyLatentAudio space.AV Concatenation: Visual and audio latents are merged into a single latent stream using the LTXVConcatAVLatent node, allowing the sampler to process both simultaneously.Frame Guidance SystemStart Frame Anchor: An image is loaded and preprocessed to serve as the reference for frame_idx: 0.End Frame Anchor: A second image is loaded and preprocessed to serve as the reference for the final frame (frame_idx: -1).Guide Application: The LTXVAddGuide nodes apply these images to the latent space with a configurable strength (set to 0.7 in this version) to dictate the videoโs trajectory.Sampling & DecodingKSampler Configuration: The workflow uses a high-speed 8-step Euler sampler with a simple scheduler.Separation & Decoding: After sampling, the LTXVSeparateAVLatent node splits the results back into distinct video and audio streams.Tiled Decoding: Visuals are decoded using VAEDecodeTiled to manage high-resolution output (1024x1024) efficiently.Enhancement & OutputNVIDIA RTX Integration: The workflow includes an RTXVideoSuperResolution node that provides a 2x hardware-accelerated upscale on the final image sequence.Final Assembly: The VHS_VideoCombine nodes produce two versions of the videoโa base generation and a super-resolved versionโboth featuring synchronized, AI-generated audio.
LTX 2.3 โ Image + Audio + Video (IC-LoRA - Union Control / Detailer) to Video
This workflow uses the LTX IC-LoRA for LTX-2.3.ControlNet (Union Control):Load an image and an audio file (either your own or the original audio from the source video), or alternatively use LTX Audioโthe audio is used for lip synchronization. Then load the target video to track and transfer its movements.Refine and Upscale (Detailer):You can also refine and upscale an existing video by setting ControlNet to "Off", Image Bypass to "True" and loading the IC-LoRA file for the detailer "ltx-2-19b-ic-lora-detailer.safetensors" instead of the ControlNet model "ltx-2.3-22b-ic-lora-union-control-ref0.5.safetensors".Info:The length of the output video is determined by the number of frames in the input video, not by the duration of the audio file.For upscaling, I use RTX Video Super Resolution.Tips:If you experience issues with lip sync, try lowering the IC-LoRA Strength and IC-LoRA Guidance Strength values. A value of around 0.7 is a good starting point. If you notice issues with output quality, try lowering the IC-LoRA Strength as well.
LTX 2.3 simple
Working on a simplified version to help you all out. Will include any improvements i have made and a few custom sigmas that may help.I use DPMpp_SDE for first stage, and eular_ancestral_cfg for upscalesFew custom sigmas that work (bong tangent custom is the same as if you downloaded the node that offers it)8 step/7..Bong tangent sigmas - 1, .971, .929, .863, .744, .500, .213, .07, 0 - Stable and my most used sigma...My sigmas, they drop off a cliff at step 2-3, keeping more of the motion of loras and prompts.My custom sigma - 1, .97, .9, .62, .48, .42, .38, .34My custom sigma more movement/stronger loras - 1, .857, .5, .42, .38, .28, .22, .18My custom may help get more motion, but will be less stable, use only if nothing else is allowing the motion you want.LTX cooks videos in about 2-3 steps, using high denoise beyond that nullifies motion and can limit lora strengths. Using the custom uses more of the motion from loras or from prompt but can be harder to control. The second and third sigma numbers are key to affecting this.SGM Uniform is high denoise throughout, makes stable videos but can limit the motion you get.Upscale sigmas - This is how i upscale. 1 step is good enough most of the time, 2 steps helps if teeth are an issue, and 3 step is only needed if your base video is bad and needs to actually be fixed. I also use RTX upscale instead of the latent upscale. Both work but RTX seems to hold teeth better than latent. Latent if viewed directly after latent upscale you can see why it has issue with teeth as the mouth gets very wonky in the latent upscale.Sigmas1 step - .65, 02 step - .65,.44, 03 step - .85, .52, .22, 0This is fairly fast as well.Second stage upscale can be used as detailer without upscaling more, using 19b detailer lora at 0.3-0.4. It can help clear faces and mouth if needed. Using 1 step sigmas above.LTX 2.3V1.1No changes to the main flowAdded 3 example flows with example files and instructions2 frame interpolation exampleImg2vid with motion transfer exampleremake video adding sync or sound exampleAll have their own json flow with everything set up for the task and instructions.All have the needed files, video ref, image refs, audio files, etx.This should show you how to use it for the most part.1-3 stages1. Basic image to video2. First/last frame interpolation3. Multi frame interpolation4. Video motion transfer with guide frames5. Video insertion as a mid video ref.6. Video reproduction adding sound and syncAudio1. Custom voice ref to guide the generated voice2. Custom audio can be syncedAll options are optional, if no video or image ref is used it does regular text to video.All options can be used together when needed.Included the images i used for interpolation for you to test with. She is an ai girl.Notes everywhere, read. Enjoy :)
LTX 2.3 Precision Redraw (Point Editor Control)
๐ Run this workflow on Runninghub (Cloud GPU):https://www.runninghub.ai/post/2043609028336558081/?inviteCode=rh-v1497(Claim 1000 FREE credits upon registration. Recommended to use the 48G VRAM Plus version for maximum precision!)๐บ Watch the Full Tutorial: Want to see how I replaced the monsters in Stranger Things using this workflow? Check out the breakdown here: ๐ []Introduction: When text prompts aren't enough, you need surgical precision. This professional-grade workflow utilizes a Point Editor to give you total control over complex scenes (like selecting one specific animal in a group).Key Features:Precision Targeting: Use Shift+Left Click (Green) to select edit areas and Shift+Right Click (Red) to protect original pixels.High-Quality Stack: Powered by the 22B-dev model and a custom 10,000-step inpainting LoRA for superior detail and consistency.Customizable Masks: Adjustable block_size to fit the silhouette of your target perfectly, whether you're replacing a small prop or a giant monster.
ZRune Motion
ZRune Motion ๆฏไธๅผ ๅบไบWan2.2 I2V็ๅทฅไฝๆต๏ผไธป่ฆ็จไบ่ง่ฒๅจไฝๅฑ็คบใ้ๅคดๆจ่ฟๅไธๆฎตๅผ่ง้ข็ๆใๅๅธ็ๅ ๅฎนไป ไพๅไบซใๅญฆไน ๏ผไธๆฏ้ถไพ่ตๅฏๅค็ฐๆดๅๅ ใ้็จๆนๅ๏ผ- ่ง่ฒๅจไฝ่ง้ข- ไธๆฎตๅผ้ๅคด็ป็ป- I2V ไบบ็ฉๆผ็คบ- ้ฃๆ ผๅๅจไฝๅฑ็คบๆ ธๅฟ็น็น๏ผ- ๅบไบ Wan2.2 I2V high / low ๅๅๆฏ- ้ๅ่ง่ฒๅจไฝๅ้ๅคด่ฟ็ปญๆง่กจ่พพ- ้ๅๅจ ComfyUI ไธญ็ปง็ปญๆฟๆข่พๅ ฅๅพใๆ็คบ่ฏๅ LoRA ่ฟ่กไบๆฌกๅไฝๆ ธๅฟไพ่ต๏ผ- Wan2.2 I2V A14B high/low ๅบๅบง- wan2.2_i2v_high_noise_14B_Q8_0.gguf- wan2.2_i2v_low_noise_14B_Q8_0.gguf- Wan2_1_VAE_bf16.safetensors- umt5_xxl_fp8_e4m3fn_scaled.safetensors- SVI_v2_PRO_Wan2.2-I2V-A14B_HIGH_lora_rank_128_fp16.safetensors- SVI_v2_PRO_Wan2.2-I2V-A14B_LOW_lora_rank_128_fp16.safetensors- wan2.2_i2v_A14b_high_noise_lora_rank64_lightx2v_4step_1022.safetensors- wan2.2_i2v_A14b_low_noise_lora_rank64_lightx2v_4step_1022.safetensors- wan2.2_wugong_i2v_high.safetensors- wan2.2_wugong_i2v_low.safetensors- zr_cine_high_s600.safetensors- zr_cine_low_s600.safetensors่ชๅฎไน่็นไพ่ต๏ผ- rgthree-comfy- comfyui-kjnodes- ComfyUI-GGUF- ComfyUI-Custom-Scripts- ComfyUI-Frame-Interpolationๆณจๆไบ้กน๏ผ- ่ฟๆฏ Wan2.2 I2V ๅทฅไฝๆต๏ผไธ้้ T2V ๆ T2I LoRA- ๆฌๆฌกๅไบซไธๅ ๅซzr_cine_high_s600.safetensorsๅzr_cine_low_s600.safetensorsไธคไธช็นๅฎไบบ็ฉๅฝข่ฑก้ฃๆ ผ็LoRAๆจกๅ๏ผๅฏผๅ ฅๅทฅไฝๆตๅ่ฏทๅ ็ฆ็จๆๆฟๆข่ฟไธคไธช LoRA- ๆไฝ้ ็ฝฎ่ฆๆฑไธบ24Gๆพๅญใ
LtxAdd - V2V ADD AUDIO AND EXTEND ANY VIDEO (ltx2.3)
LtxAdd - V2V ADD AUDIO AND EXTEND ANY VIDEO (ltx2.3)This is another App Mode I know I will use a LOT.It will ADD AUDIO AND EXTEND any video!I modified the original experimental following workflow from RUNEXX for my own needsLTX-2.3_-_V2V_Foley_Add_Sound_To_Any_VideoChanged the models to the ones used by the Vanilla LTX2.3 Comfyui Templateso if you already have those original models it's an out of the box experience :)Options:Toggle between adding 10 or 20 seconds.Toggle ON/OFF for Enhanced PromptToggle to Add Custom LoRA
Wan 2.1 Video Restyle | Transform Video with First Frame for Consistent Stylization
Transform your video style by applying the restyled first frame using Wan 2.1 video restyle workflow.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewUpdated 6/16/2025: ComfyUI version updated to v0.3.39 for improved stability and compatibility, with SageAttn installed for 30% faster generation. Wan 2.1 video restyle is a powerful video-to-video workflow designed for Wan 2.1 models. It allows users to apply the look of a stylized first frame to an entire video sequence, achieving high-quality transformations without complex manual adjustments. Ideal for animations, cinematic scenes, and creative projects, this workflow ensures seamless and coherent stylization across all frames.Important nodes:Green:Purple:NotesWan 2.1 Video Restyle | Transform Video with First Frame for Consistent Stylization โ see RunComfy page for the latest node requirements.
Wan 2.1 Fun Motion Control | AI Photo-to-Video Animation
Design motion paths to animate still photos into videos.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewWan 2.1 Trajectory Control is a specialized workflow within the Wan 2.1 Fun family that enables motion-controlled video generation from a single input image. By drawing or defining a trajectory path, users can guide the camera or subject movement across frames to produce cinematic effects, animations, or dynamic storytelling scenes. This method unlocks powerful control over motion planning while keeping the artistic direction intact. Built for the Wan 2.1 Fun model ecosystem, this workflow makes high-quality AI video synthesis accessible and intuitive.Important nodes:Positive Prompt:Negative Prompt:NotesWan 2.1 Fun Motion Control | AI Photo-to-Video Animation โ see RunComfy page for the latest node requirements.
Wan 2.1 Fun | Image-to-Video and Text-to-Video AI Generation
Empower your AI videos with Wan 2.1 Fun.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewUpdated 6/16/2025: ComfyUI version updated to v0.3.39 for improved stability and compatibility. This Wan 2.1 Fun workflow supports both image-to-video and text-to-video generation modes, offering a versatile creative toolkit for AI-powered motion content. From transforming static images into dynamic scenes to crafting imaginative sequences from pure text prompts, this setup uses the Wan 2.1 Fun modelโs capabilities to deliver coherent, cinematic results. It supports prompt-based transitions, frame prediction, and flexible durations, making it perfect for digital storytelling, AI animation, or experimental media art.Important nodes:Positive Prompt:Negative Prompt:NotesWan 2.1 Fun | Image-to-Video and Text-to-Video AI Generation โ see RunComfy page for the latest node requirements.
SkyReels-A2 Element-to-Video Workflow in ComfyUI | E2V Technology
Combine multi elements into dynamic videos with precision.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewSkyReels-A2 is a powerful elements-to-video (E2V) workflow that allows you to combine multiple reference images into dynamic, realistic videos. This solution preserves the visual integrity of each element while ensuring natural composition and motion. Perfect for creating virtual product demonstrations, character animations, or scenes with multiple subjects, this workflow gives you precise control over video generation through both visual references and text prompts.NotesSkyReels-A2 Element-to-Video Workflow in ComfyUI | E2V Technology โ see RunComfy page for the latest node requirements.
FramePack Wrapper for ComfyUI | Long Video Generation with Low Memory
Create stable, 60s+ long videos with minimal cloud resources.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewUpdated 6/16/2025: ComfyUI version updated to v0.3.39 for improved stability and compatibility, with SageAttn installed for 30% faster generation. The FramePack workflow for ComfyUI brings Stanford University's revolutionary video generation technology to RunComfy. FramePack enables you to create professional-quality videos up to 60+ seconds at full 30 FPS with consistent quality from start to finish. It solves two critical video generation problems: 'forgetting' (fading memory of earlier content) and 'drifting' (quality degradation over time) through its unique bi-directional sampling methods. Generate thousands of frames without the common quality drop issues - all while using minimal cloud resources!Important nodes:Resize ImageNotesFramePack Wrapper for ComfyUI | Long Video Generation with Low Memory โ see RunComfy page for the latest node requirements.
ComfyUI HunyuanCustom Dual-Subject Workflow | Video Generation
Create dual-subject videos with exceptional identity preservation.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewThis ComfyUI HunyuanCustom Multi-Subject workflow integrates Tencent's advanced model into the ComfyUI environment, specifically optimized for dual-subject generation. Built upon the HunyuanVideo framework and enhanced by Kijai, this workflow excels at maintaining consistent identity features of both subjects throughout generated videos. With FreSca technology and specialized nodes, it achieves superior quality for virtual demonstrations, product showcases, and natural subject interactions with approximately 30% faster processing speeds.Important nodes:ResolutionFramesCFG ScaleFlow ShiftEnabledWeightTemporal ConcatStrengthNotesComfyUI HunyuanCustom Dual-Subject Workflow | Video Generation โ see RunComfy page for the latest node requirements.
Overtli LLM Studio Suite
A ComfyUI suite of nodes for Pollinations, LM Studio, Copilot CLI, and OpenAI-compatible generation with prompt enhancing, image gen, video gen, speech/audio gen, for local/cloud multi-engine workflows.You can download a sample workflow with it and Flux2Klein 9b GGUF (can always switch model loader).Access the github here for comfyui install:github.com/OvertliDS/overtli-studio-suite## ๐ Installation & Quick Start ### Installation 1. Navigate to your ComfyUI custom nodes directory and clone this repository: ```powershell cd ComfyUI/custom_nodes/ git clone https://github.com/OvertliDS/overtli-studio-suite.git ``` 2. Navigate into the cloned folder and install the required dependencies using your ComfyUI Python environment: ```powershell cd overtli-studio-suite pip install -r requirements.txt ``` 3. Restart ComfyUI. ### Quick Start Workflow *You can find the nodes by simply searching for "Overtli".* 1. Add the **`GZ_ProviderSettings`** node and save your keys or URLs once. 2. Add the **`GZ_AdvancedTextEnhancer`** or **`GZ_LLMTextEnhancer`** node. 3. Choose your `provider` and `active_engine`. 4. Connect optional `IMAGE` or `AUDIO` inputs when needed. 5. Execute and use the native output directly in your workflow. ### Use With Existing ComfyUI Workflows (CLIP Text Encode) You can use OVERTLI as a **drop-in prompt enhancer** without rebuilding your graph. 1. Add any text-capable OVERTLI node (for example `GZ_TextEnhancer`, `GZ_LLMTextEnhancer`, or `GZ_AdvancedTextEnhancer` with `active_engine = text`). 2. Connect the node's `STRING` output to your existing **CLIP Text Encode** node `text` input. 3. Keep the rest of your workflow unchanged (sampler, model, VAE, etc.). This lets you enhance or rewrite any prompt while still using your current ComfyUI pipeline. ### Example Workflow For users who want a sample workflow with Flux2Klein 9B GGUF, use workflows/OvertliStudioSuite_x_Flux2Klein9B-GGUF.json from this repo. ----- ## โจ Core Capabilities * **Unified Advanced Routing:** Manage multi-engine workflows through a single, powerful provider node. * **Native Media Contracts:** Full support for standard ComfyUI output types. * `AUDIO` for text-to-speech and text-to-music generation. * `VIDEO` for video generation capabilities. * The Advanced Router dynamically returns `STRING + IMAGE + VIDEO + AUDIO`. * **Native Previews:** Fully integrated with ComfyUIโs native image and text preview rendering. ----- ## ๐งฉ Registered Nodes | Node | Purpose | Supports | | :--- | :--- | :--- | | โก **`GZ_AdvancedTextEnhancer`** | **(Recommended)** Unified provider/engine router. | Provider-aware engine routing: `text`, `image`, `video`, `text_to_speech`, `speech_to_text`, `text_to_music` with validation. | | ๐ **`GZ_OpenAICompatibleTextEnhancer`** | Dedicated OpenAI-compatible all-engines node. | `text`, `image_gen`, `video_gen`, `text_to_speech_gen`, `speech_to_text_gen`, `text_to_music_gen` through OpenAI-compatible APIs. | | ๐ **`GZ_TextEnhancer`** | Pollinations text and optional vision enhancement. | Pollinations text generation, optional vision image context, grouped mode presets. | | ๐จ **`GZ_ImageGen`** | Pollinations image generation. | Pollinations image generation models. | | ๐ฌ **`GZ_VideoGen`** | Pollinations video generation. | Pollinations video generation models (`VIDEO` output). | | ๐ฃ๏ธ **`GZ_TextToSpeech`** | Pollinations speech generation. | Pollinations text-to-speech models (`AUDIO` output). | | ๐ง **`GZ_SpeechToText`** | Pollinations speech transcription. | Pollinations speech-to-text models (`STRING` transcript output). | | ๐ต **`GZ_TextToAudio`** | Pollinations text-to-music generation. | Pollinations text-to-music generation (`AUDIO` output). | | ๐ฅ๏ธ **`GZ_LLMTextEnhancer`**| **(Recommended for local)** Local/OpenAI-compatible text+vision route (LM Studio, Ollama, similar endpoints). | Text generation with optional image context over local/OpenAI-compatible chat endpoints. | | ๐ค **`GZ_CopilotAgent`** | GitHub Copilot CLI route. | Copilot CLI text generation with optional image attachment context. | | โ๏ธ **`GZ_ProviderSettings`** | Persisted provider settings helper. | Save/load provider model, endpoint, and API key settings for this suite. | | ๐ **`GZ_PromptLibraryNode`** | Prompt library CRUD/refresh utility. | Prompt management and reusable preset selection. | | ๐ฅ **`GZ_StyleStackNode`** | Reusable style stack utility. | Composable style bundles for prompt layering across nodes. | ----- ## ๐ Provider Coverage ### โจ Pollinations.ai Full support for text, image, video, text-to-speech, speech-to-text, and text-to-music. ### ๐ฅ๏ธ LM Studio Local text generation with optional image context for vision-capable local models. ### ๐ค GitHub Copilot CLI Local CLI-based text enhancement with optional image context. ### ๐ OpenAI-Compatible APIs Available through both: * `GZ_OpenAICompatibleTextEnhancer` (dedicated node) * `GZ_AdvancedTextEnhancer` with `provider = openai_compatible` Supported engines: * `text` * `image` * `video` * `text_to_speech` * `speech_to_text` * `text_to_music` > **Note:** exact modality availability still depends on the target OpenAI-compatible provider and selected model. The suite now exposes these engines with explicit runtime validation and terminal-visible errors when a provider/model endpoint cannot satisfy a request. ----- ## ๐ Provider Setup & Authentication The Overtli Studio Suite is designed to be as "zero-config" as possible, supporting cloud, local, and standard API protocols. ### โจ Pollinations.ai (Cloud) API Key: Visit enter.pollinations.ai to generate your key. Setup: Enter your key into the GZ_ProviderSettings node or set the GZ_POLLINATIONS_API_KEY environment variable, as well as support in Pollinations nodes to enter api key directly. Pricing: `Free` tagged models allow free generations, but pollen usage is determined based on your free plan tier within enter.pollinations.ai so aim to reach higher tiers for more free generations or top up using gems while waiting for your pollen to reset. `Paid` models are supported as well if gems have been purchased. ### ๐ค GitHub Copilot CLI (Cloud/Local Hybrid) Auto-Detection: If you are already signed in via the GitHub CLI (gh auth login), Overtli will automatically detect your session and "just work." Prerequisites: Requires an active Copilot subscription and the GitHub CLI installed (It will open a copilot-cli terminal silently by default for communication). Vision/Image context: OVERTLI writes Comfy `IMAGE` inputs to temporary local files and references them with Copilot CLI `@path` attachment syntax so vision-capable models receive the actual image bytes. Background behavior: Copilot auth recovery and vision-cache retry handling are managed automatically in the background so the node surface stays simple. With `vision_enabled` on, prior runtime vision-cache entries do not silently block new image attempts by default. ### ๐ฅ๏ธ LM Studio (Local) Host: Defaults to http://localhost:1234. Setup: Ensure LM Studio is running and the "Local Server" is started. No API key is required by default. Vision: Support for vision-capable local models is integrated directly into the GZ_LLMTextEnhancer. ### ๐ OpenAI-Compatible APIs (Custom) Flexibility: Use any provider that supports the OpenAI standard (e.g., Groq, Together AI, LocalAI). Setup: Requires a Base URL and API Key configured in the GZ_ProviderSettings node (or just use the GZ_AdvancedTextEnhancer or GZ_LLMTextEnhancer node). ----- ## ๐ง Prompting & Output Contracts ### The Composition Model All primary routes follow the same strict sequence: 1. **Custom Instructions** 2. **Selected Mode Preset** 3. **Raw Prompt** (User Prompt) 4. **Style Layers** (`style_preset_1..3` + `additional_styles`) > ๐ก **Tip:** The `additional_styles` input is socket-first (`forceInput`) and is designed to chain seamlessly from the `GZ_StyleStackNode`. ### Quick Instruction Modes Overview Instruction presets are grouped by task family and can be toggled on/off per route. At-a-glance preset counts (current build): - **Total instruction presets:** **63** (excluding `Off` options) - **Text:** 14 - **Image:** 24 (4 vision + 10 generation + 6 editing + 4 narrative) - **Video:** 11 (8 generation + 3 analysis) - **TTS:** 8 - **Text-to-Audio:** 3 - **Speech-to-Text:** 3 - **Text modes:** examples include `Enhance` (expand a short idea into a richer prompt), `Refine` (tighten wording while preserving intent), and `Translation Prompt` (convert to generation-ready English). - **Image modes:** examples include `Tags` (comma-separated visual tags), `Simple Description` (single concise visual description), `Concept Art` (design-forward scene direction), and `Upscale Image Prompt` (detail-preserving enhancement guidance). - **Video modes:** examples include `Cinematic Video Prompt` (film-style motion direction), `Loop Video Prompt` (seamless repeating motion guidance), and `Video Summary` (chronological visual recap of footage). - **TTS modes:** examples include `Voiceover Script` (clean narration pacing for speech synthesis), `Character Voice` (expressive dialogue delivery), and `SSML Enhancement` (markup-driven timing and emphasis control). - **Text-to-Audio modes:** examples include `Audio Prompt Enhance` (clarify generation intent), `Music Direction` (genre/instrument/energy guidance), and `Ambience and Foley` (layered environmental sound design). - **Speech-to-Text modes:** examples include `Clean Transcript` (readability cleanup while preserving meaning), `Punctuation and Casing` (restore sentence structure), and `Structured Notes` (concise key points and actions). If a mode is set to `Off`, the route relies on custom instructions + raw prompt + style layers. *Only enable ONE instruction group* ### Quick Style Layering Overview - **Built-in style library:** **656** presets (+ `Off` option) - **Built-in style slots:** up to **3** per primary node (`style_preset_1..3`). - **Extended style stack:** connect `additional_styles` from **`GZ_StyleStackNode`** for up to **7** extra styles. - **Total style layers available:** up to **10** (3 built-in + 7 stacked). Style examples (brief): - **Photograph (Real Life):** grounded, natural realism with real lens/lighting behavior. - **Cinematic Still:** film-like framing, layered depth, and motivated key/fill/rim lighting. - **Anime Illustration:** expressive line work, controlled cel-shading, and clean silhouette readability. - **Line Art (Clean):** minimal contour-focused visuals with crisp outlines and low noise. - **Concept Art:** production-style world/design visualization with clear value hierarchy. ### Output Contracts | Route Type | Native ComfyUI Output | | :--- | :--- | | **Text** | `STRING` | | **Image** | `IMAGE` | | **Video** | `VIDEO` | | **Audio-producing** | `AUDIO` | | **Advanced Router** | `STRING`, `IMAGE`, `VIDEO`, `AUDIO` | ----- ## โ๏ธ Settings and Precedence Persistent settings are stored locally in: ```text ComfyUI/user/overtli_studio_settings.json ``` **Resolution Precedence:** 1. Runtime node input *(Highest priority)* 2. Environment variable 3. Persisted settings 4. Suite default *(Lowest priority)* <details> <summary><b>๐ง Click to view useful Environment Variables</b></summary> * `GZ_POLLINATIONS_API_KEY` * `GZ_POLLINATIONS_CHAT_TIMEOUT` * `GZ_POLLINATIONS_IMAGE_TIMEOUT` * `GZ_POLLINATIONS_VIDEO_TIMEOUT` * `GZ_POLLINATIONS_TTS_TIMEOUT` * `GZ_POLLINATIONS_STT_TIMEOUT` * `GZ_POLLINATIONS_AUDIO_TIMEOUT` * `GZ_LMSTUDIO_HOST` * `GZ_LMSTUDIO_PORT` * `GZ_LMSTUDIO_API_KEY` * `GZ_COPILOT_EXECUTABLE` * `GZ_COPILOT_MODEL` * `GZ_COPILOT_TIMEOUT` * `GZ_OPENAI_COMPAT_BASE_URL` * `GZ_OPENAI_COMPAT_API_KEY` * `GZ_OPENAI_COMPAT_MODEL` * `GZ_LOG_LEVEL` * `GZ_TEMP_DIR` </details> ----- ## ๐งฏ Troubleshooting **OpenAI-compatible route does not appear as its own node** That is expected. Use `GZ_AdvancedTextEnhancer` with `provider = openai_compatible`. **STT fails with missing input** `GZ_SpeechToText` requires an `AUDIO` input connection to function. **Video generation blocks the UI** That is expected for synchronous provider-side video generation. The UI will resume once rendering is complete. **LM Studio/Ollama does not return models** Make sure LM Studio/Ollama is running and exposing its API endpoint correctly. **Copilot route fails unexpectedly** Check that: * GitHub Copilot CLI is installed and reachable. * The configured executable path is correct if adjusted manually. * Your authentication session is valid. * Environment token overrides are not conflicting. ----- ## ๐ค Support & Funding If this suite accelerates your workflow, consider supporting the continued development of Overtli tools and other projects using the GitHub Sponsor link on the repository sidebar. ## ๐ License MIT. See `LICENSE`.
WAN2.2 14B GGUF (I2V,T2V) Low-Vram Workflow
Coming SoonInfinite talkMMaudioLastframefirstframeUltimate VersionSupported WorkflowsWan2.2 T2V (TextโtoโVideo) 14Bย Wan2.2 I2V (ImageโtoโVideo)ย 14BWan2.2 TI2Vโ5B (Hybrid Text/ImageโtoโVideo) (old version)Nodes Used (Install via ComfyUI Manager or links below)๐ WanVideoWrapper๐ KJNodesModel Downloads + VAE*these are conversions from the original models to run on less VRAM.๐ WAN2.2 14Bย T2V๐ WAN2.2 14B I2VAll these GGUF conversions are done at the moment by:https://huggingface.co/city96https://huggingface.co/bullerwinshttps://huggingface.co/QuantStack*If you cant find the model you are looking for check out there profiles!Additional Required Files๐ CLIP Vision, Text Encoder๐ Loras (i2v,T2v)(low and high required)https://huggingface.co/lightx2v
Self Forcing | Autoregressive Keyframe-to-Video Generation
SUPER FAST! 5-second video in 45 seconds! **Who it's for:** creators who want this pipeline in ComfyUI without assembling nodes from scratch. **Not for:** one-click results with zero tuning โ you still choose inputs, prompts, and settings. ### Open preloaded workflow on RunComfy [Open preloaded workflow on RunComfy (browser)](https://www.runcomfy.com/comfyui-workflows/self-forcing-autoregressive-keyframe-to-video-generation?utm_source=civitai&utm_medium=referral) **Why RunComfy first** - **Fewer missing-node surprises** โ run the graph in a managed environment before you mirror it locally. - **Quick GPU tryout** โ useful if your local VRAM or install time is the bottleneck. - **Matches the published JSON** โ the zip follows the same runnable workflow you can open on RunComfy. **When downloading for local ComfyUI makes sense** โ you want full control over models on disk, batch scripting, or offline runs. **How to use (local ComfyUI)** 1. Load inputs (images/video/audio) in the marked loader nodes. 2. Set prompts, resolution, and seeds; start with a short test run. 3. Export from the Save / Write nodes shown in the graph. **Expectations** โ First run may pull **large weights**; cloud runs may require a **free RunComfy account**. --- ### Overview Self Forcing trains autoregressive video diffusion models by simulating the inference process during training, performing autoregressive rollout with KV caching. It resolves the train-test distribution mismatch and enables real-time, streaming video generation on a single RTX 4090 while matching the quality of state-of-the-art diffusion models. In simple terms, it allows you to generate smooth videos by providing start and end keyframe reference images along with a guiding text prompt, producing identity-consistent and motion-smooth video synthesis. ### Notes **Self Forcing | Autoregressive Keyframe-to-Video Generation** โ see [RunComfy page](https://www.runcomfy.com/comfyui-workflows/self-forcing-autoregressive-keyframe-to-video-generation?utm_source=civitai&utm_medium=referral) for the latest node requirements.
Wan FusionX Workflow ComfyUI | Ultimate Video Generation
Most powerful video generation solution yet! Cinema-grade detail, your personal film studio.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewThis comprehensive Wan FusionX workflow represents the new king in video generation technology. Built on the fusion of CausVid, AccVideo, MoviiGen1.1, and MPS Reward LoRA, this all-in-one solution delivers cinema-grade video quality in just 6 steps. The workflow integrates complete Text-to-Video and Image-to-Video capabilities, plus the full VACE 14B 5-function solution (reference-based subject replacement, dual-image video generation, first/last frame control, object modification, and video expansion). This represents the pinnacle of current video generation technology, offering unprecedented speed, quality, and versatility in a single unified workflow.Important nodes:AccVideoMoviiGen1.1Positive Prompt:Negative Prompt:CFG:Steps:Shift:Sampler:Frames:TeaCache:Lower Shift (1-2):Higher Shift (3-9):Frame Rate:Model Selection:Enhanced Quality:Enhanced Reference Image + ControlNet:Depth Control:OpenPose Control:NotesWan FusionX Workflow ComfyUI | Ultimate Video Generation โ see RunComfy page for the latest node requirements.
MultiTalk Workflow in ComfyUI | Photo to Talking Video
Millisecond lip sync + Wan2.1 = 15s ultra-detailed talking videos!Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewThis ComfyUI workflow showcases MultiTalk's powerful audio-driven technology for creating 15-second talking videos from static photos. With millisecond-precise lip synchronization that outperforms Sonic, MultiTalk delivers stunning results for singing and speaking scenarios. Enhanced by Wan video generation backbone, optional Uni3C camera stabilization, and acceleration LoRA, this complete solution supports 480p-720p output with natural facial expressions and body movements.NotesMultiTalk Workflow in ComfyUI | Photo to Talking Video โ see RunComfy page for the latest node requirements.
ComfyUI VACE 14B | All-in-One Video Creation & Editing Workflow
Create, edit and transform videos with the powerful VACE Wan2.1 14B.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewUpdated 6/16/2025: ComfyUI version updated to v0.3.39 for improved stability and compatibility, with SageAttn installed for 30% faster generation. Updated 5/24/2025: Speed Mode (FP8E5M2 + quantization) for faster generation or Quality Mode (FP8E4M3 + disable quantization) for better results now available. Harness the power of Wan2.1-VACE 14B, the all-in-one video creation and editing model. This comprehensive workflow enables five key capabilities: replacing subjects in videos using reference images, generating videos from two reference images, controlling video generation with first and last frames, modifying specific objects within videos, and expanding video dimensions. Experience masterful video manipulation with this unified solution.Important nodes:Tip:NotesComfyUI VACE 14B | All-in-One Video Creation & Editing Workflow โ see RunComfy page for the latest node requirements.
Uni3C ComfyUI Workflow | Video-Referenced Camera & Motion Transfer
Extract camera movements and human motions from reference videos for professional video generationWho it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewThis advanced Uni3C workflow extracts camera movements and human motions from reference videos and applies them to your input images for professional video generation. Unlike traditional manual animation tools, Uni3C automatically understands motion patterns from reference footage and seamlessly transfers both camera work and character animations to new scenes. The workflow supports motion transfer between different visual styles, video-referenced camera movements, and natural human animation synthesis from any reference footage. With 4x speed optimization, you can generate high-quality videos with extracted motion patterns in minutes.Important nodes:video reference extractionGood Examples:Key Point:Preserve Original Motion:Modify Actions:NotesUni3C ComfyUI Workflow | Video-Referenced Camera & Motion Transfer โ see RunComfy page for the latest node requirements.
ComfyUI Phantom | Subject to Video
Reference-driven video generation using Wan2.1 14B **Who it's for:** creators who want this pipeline in ComfyUI without assembling nodes from scratch. **Not for:** one-click results with zero tuning โ you still choose inputs, prompts, and settings. ### Open preloaded workflow on RunComfy [Open preloaded workflow on RunComfy (browser)](https://www.runcomfy.com/comfyui-workflows/comfyui-phantom-subject-to-video?utm_source=civitai&utm_medium=referral) **Why RunComfy first** - **Fewer missing-node surprises** โ run the graph in a managed environment before you mirror it locally. - **Quick GPU tryout** โ useful if your local VRAM or install time is the bottleneck. - **Matches the published JSON** โ the zip follows the same runnable workflow you can open on RunComfy. **When downloading for local ComfyUI makes sense** โ you want full control over models on disk, batch scripting, or offline runs. **How to use (local ComfyUI)** 1. Load inputs (images/video/audio) in the marked loader nodes. 2. Set prompts, resolution, and seeds; start with a short test run. 3. Export from the Save / Write nodes shown in the graph. **Expectations** โ First run may pull **large weights**; cloud runs may require a **free RunComfy account**. --- ### Overview ComfyUI Phantom is a unified video generation framework for single and multi-subject references, built on existing text-to-video and image-to-video architectures. It achieves cross-modal alignment using text-image-video triplet data by redesigning the joint text-image injection model. Additionally, it emphasizes subject consistency in human generation while enhancing ID-preserving video generation. In simpler terms, ComfyUI Phantom allows you to generate videos based on reference images and a text prompt, making it perfect for identity-consistent human video synthesis. ### Notes **ComfyUI Phantom | Subject to Video** โ see [RunComfy page](https://www.runcomfy.com/comfyui-workflows/comfyui-phantom-subject-to-video?utm_source=civitai&utm_medium=referral) for the latest node requirements.
SeedVR2: one-step 4X video/image upscaling (and beyond) with BlockSwap and great temporal consistency
Restore and upscale any video to 4X and beyond in a single step with ByteDance's revolutionary SeedVR2. Tutorial update for SeedVR2.5๐ Support our workIf you found this tutorial helpful and want to support more open-source content like this, any contribution helps us continue creating in-depth guides for the community: https://www.ainvfx.com/donate/Every donation enables us to dedicate more time to research, testing, and sharing knowledge. Thank you for being part of this journey!๐ Follow AInVFX- Website: https://www.ainvfx.com- LinkedIn: https://www.linkedin.com/company/ainvfx- Instagram: https://www.instagram.com/ainvfxcom- Facebook: https://www.facebook.com/ainvfxcom- TikTok: https://www.tiktok.com/@ainvfxcom- GitHub: https://www.github.com/AInVFX
Video upscaler and Interpolator [RIFE VFI + Nvidia Video Super Resolution]
Turn your low-resolution, low-framerate videos into high-quality, high-FPS videos.**Whatโs Interpolation?** A method that adds extra frames between the original ones to make the video look smoother. It basically โfills in the gapsโ between frames, so motion appears more fluid and less choppy.This workflow uses NVIDIA VSR, so it does not add extra detail to the base video. Instead, it preserves the original quality while upscaling without introducing pixelation.It uses RIFE VFI for interpolation.Tested with 1-minute-long videos targeting 1440p at 64 FPS on an NVIDIA RTX 4090.To install ComfyUI_NVIDIA_RTX_Nodes:1. Update ComfyUI to the latest version.2. Using ComfyUI Manager, search for the ComfyUI_NVIDIA_RTX_Nodes custom node and install it.COMFYUI SECURITY ERROR FIXIf you hit a security level error while trying to install the ComfyUI_NVIDIA_RTX_Nodes custom node from ComfyUI manager:You need to temporarily change the security level in your config.ini file.If you are using the portable version, the file is located at:ComfyUI/user/__manager/config.iniChange the value to:security_level = weakIMPORTANT:Do NOT forget to restore it back to its original value. It is safe to temporarily lower the security level for this node since it comes directly from the ComfyUI team, but you should restore it afterward for safety reasons.Perform a hard restart of ComfyUI (completely stop it, then start it again).Install the custom node using ComfyUI manager and double-check that it is the official one. The link should point to:https://github.com/Comfy-Org/Nvidia_RTX_Nodes_ComfyUIRestart ComfyUI and check if the node works properly.If you encounter a dependency error, keep reading...Once everything is working properly, restore your original security level in the config.ini file and restart ComfyUI one final time.COMFYUI PORTABLE DEPENDENCY ERROR FIXIf you hit a dependency error trying to run the Nvidia VSR node, try this:Create a backup of your embedded Python directory, in case you break something you can restore your embedded directory later using this backup. Then run the following command: python -m pip install -U --no-build-isolation nvidia-vfx --index-url https://pypi.nvidia.com Run the command in your embedded Python directory.For example: D:\ComfyUI\python_embeded.This command is designed for CMD. If you're using PowerShell, replace the initial python with .\python.exe.Detailed use instructions are contained within the workflow it self:- Yellow nodes are input and settings you can configure to suit your needs.- Red nodes are instructions and useful notes.
Very low VRAM LTXAV 2.3 I2V workflow. - 4GB
This is the only LTX 2.3 video workflow so far I that I can run on my NVIDIA GeForce RTX 3050 4GB vram GPU and 8GB ram laptop.It's not fast on 4GB and probably much faster on 6GB and 8GB machines.I recommend having these settings in your run_nvidia_gpu.bat file:--windows-standalone-build --lowvram --force-fp16 --use-split-cross-attentionSo change the first line to something like this:.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build --lowvram --force-fp16 --use-split-cross-attentionHere are the filenames and where to Place them. The files can be found on huggingface.C:\AI\ComfyUI_windows_portable\ComfyUI\models\unetltx-2-3-22b-dev-Q4_K_M.ggufC:\AI\ComfyUI_windows_portable\ComfyUI\models\clipgemma_3_12B_it_fp4_mixed.safetensorsltx-2.3_text_projection_bf16.safetensorsC:\AI\ComfyUI_windows_portable\ComfyUI\models\vaeLTX23_video_vae_bf16.safetensorsLTX23_audio_vae_bf16.safetensorsC:\AI\ComfyUI_windows_portable\ComfyUI\models\latent_upscale_modelsltx-2.3-spatial-upscaler-x2-1.0.safetensorsC:\AI\ComfyUI_windows_portable\ComfyUI\models\lorasltx-2.3-22b-distilled-lora-dynamic_fro09_avg_rank_105_bf16.safetensorsIf you change the frame rate, do it on both video and audio so it stays synced.
Wan 2.1 Workflow in ComfyUI | Premium Text & Image to Video Creation
Create incredible videosย from text or images with breakthrough AIย running on everydayย CPUs.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewUpdated 6/16/2025: ComfyUI version updated to v0.3.39 for improved stability and compatibility. Wan 2.1 surpasses all competitors in video creation benchmarks. Its 1.3B model needs only 8.19GB VRAM, supporting both Text-to-Video and Image-to-Video workflows to produce 480P videos in 4 minutes on standard hardware. The Wan 2.1 14B model offers enhanced 720P quality via RunComfy's cloud. As the first model generating both Chinese and English text in videos, Wan 2.1 expands creative options while its Wan-VAE backend efficiently handles 1080P video processing with temporal consistency.Important nodes:High-quality output:Hardware accessibility:Versatile generation:Multilingual support:VAE efficiency:Fast processing:Inputs:Best for:Characteristics:KSampler:WanImageToVideo:ModelSamplingSD3:EmptyHunyuanLatentVideo:Memory Optimization:Quality Optimization:NotesWan 2.1 Workflow in ComfyUI | Premium Text & Image to Video Creation โ see RunComfy page for the latest node requirements.
Push-In Camera - A Motion LoRA for Wan 2.1
One image in, blockbuster push-in shots out. Zero complexity.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewTransform your AI videos from static to cinematic with this Push-In Camera - A Motion LoRA for Wan 2.1 workflow. Trained on 100 real film clips through 40+ iterations, this LoRA adds professional push-in drone camera movements that breathe life into your generations. Just load your image and hit run - the workflow handles everything automatically, delivering Hollywood-quality push-in shots that sweep from establishing views to dramatic close-ups. No complex prompts needed, just pure cinematic motion.Important nodes:Push-in cameraNotesPush-In Camera - A Motion LoRA for Wan 2.1 โ see RunComfy page for the latest node requirements.
Wan 2.2 + Lightx2v V2 ComfyUI Workflow | Fast Image & Text to Video
Dual Light LoRA setup, 4X faster.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewThis ComfyUI Wan 2.2 + Lightx2v V2 workflow combines the power of Wan 2.2 14B model with Lightx2v V2 LoRA for lightning-fast image-to-video and text-to-video generation. Transform static images into dynamic videos or create videos directly from text prompts using advanced distillation techniques. Generate high-quality results in 8-14 steps with reduced sampling parameters, dramatically cutting generation time while maintaining excellent output quality. Choose from optimized configurations: 8 steps (maximum speed), 12 steps (balanced), or 14 steps (optimal quality), perfect for rapid prototyping and efficient video creation workflows.Important nodes:EmptyHunyuanLatentVideoCLIP Text Encode (Positive Prompt)LoadImageNotesWan 2.2 + Lightx2v V2 ComfyUI Workflow | Fast Image & Text to Video โ see RunComfy page for the latest node requirements.
Wan 2.2 Low Vram Kijai Wrapper | ComfyUI Video Generation
Low VRAM. No longer waiting. Kijai wrapper included.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewWan 2.2 Low VRAM is a memory-efficient ComfyUI workflow based on Kijai's wrapper implementation of the Wan 2.2 video generation model. It features an OpenAI-powered prompt refinement node and a two-stage sampling system (4 steps each for high and low noise). This approach balances quality with performance, enabling smooth video interpolation between frames even on VRAM-constrained setups. Perfect for creators who need efficient video generation without sacrificing quality.NotesWan 2.2 Low Vram Kijai Wrapper | ComfyUI Video Generation โ see RunComfy page for the latest node requirements.
Wan 2.2 Fun Control | ComfyUI Video Generation
Structured video generation with Wan 2.2 Fun Control using Canny, Depth, Pose, and inputs for precise creative guidance.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewThe Wan 2.2 Fun Control workflow in ComfyUI enables multimodal control by leveraging depth, canny edges, pose, and trajectory inputs to drive structured, expressive video generation using the new Wan 2.2 Fun Control model. It offers precise motion and styling control beyond prompt-only methods, perfect for animation, pose-driven content, or stylized transformations, all powered by Wan 2.2โs next-gen MoE-based cinematic generation.NotesWan 2.2 Fun Control | ComfyUI Video Generation โ see RunComfy page for the latest node requirements.
Wan 2.2 Video Restyle | Try First Frame Style Transfer in ComfyUI Workflow
Change the first frame, folks, your style makes the whole video look amazing. Pure magic.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewWan 2.2 Video Restyle locks in your videoโs look from the first frame with Flux Kontext or custom styles, then keeps it consistent across every scene. Perfect for cinematic effects, anime vibes, or bold artistic transformations.NotesWan 2.2 Video Restyle | Try First Frame Style Transfer in ComfyUI Workflow โ see RunComfy page for the latest node requirements.
Wan 2.2 Lightning T2V I2V ComfyUI | Ultra Fast Video Gen
Wan 2.2 now 20x faster! T2V + I2V in 4 steps.Who it's for: creators who want this pipeline in ComfyUI without assembling nodes from scratch. Not for: one-click results with zero tuning โ you still choose inputs, prompts, and settings.Open preloaded workflow on RunComfyOpen preloaded workflow on RunComfy (browser)Why RunComfy first- Fewer missing-node surprises โ run the graph in a managed environment before you mirror it locally.- Quick GPU tryout โ useful if your local VRAM or install time is the bottleneck.- Matches the published JSON โ the zip follows the same runnable workflow you can open on RunComfy.When downloading for local ComfyUI makes sense โ you want full control over models on disk, batch scripting, or offline runs.How to use (local ComfyUI)1. Load inputs (images/video/audio) in the marked loader nodes.2. Set prompts, resolution, and seeds; start with a short test run.3. Export from the Save / Write nodes shown in the graph.Expectations โ First run may pull large weights; cloud runs may require a free RunComfy account.OverviewThis ComfyUI workflow features the upgraded Wan 2.2 Lightning distillation model, replacing the previous lightx2v lora with native Wan 2.2 compatibility. Generate professional-quality videos in just 4 steps with 20x speed improvement while maintaining the same visual fidelity. Create complex motion sequences, animate static images, and produce cinematic content at lightning speed. Perfect for rapid prototyping, content creation, and real-time video generation workflows.Important nodes:WanVideoTextEncodeWanVideoEmptyEmbedsNotesWan 2.2 Lightning T2V I2V ComfyUI | Ultra Fast Video Gen โ see RunComfy page for the latest node requirements.