SocksLatentPatcher GGUF Subgraph — v1.0
This is a custom workflow from: . What it do it circumnavigates the loss of detail, saturation/desaturation for infinite length video generation, the node bypasses the vae decode and directly patches the latent...
Runtime profile
- Access
- External workflow
- Status
- Listed
- Version
- SocksLatentPatcherGGUF.json
- Source
- Civitai
- Updated
- Oct 19, 2026
Source description
This is a custom workflow from: https://github.com/synystersocks/ComfyUI-SocksLatentPatcher . What it do it circumnavigates the loss of detail, saturation/desaturation for infinite length video generation, the node bypasses the vae decode and directly patches the latent tensor. experimental for i2v. I implemented Painteri2v, ⚡ Rife Tensorrt, Upscaler Tensorrt ⚡ a choice between rife vf1 and rife tensorrt. All in with subgraph.
NODE:
PAINTERI2V: https://github.com/princepainter/ComfyUI-PainterI2V
Sockslatentpatcher: https://github.com/synystersocks/ComfyUI-SocksLatentPatcher
⚡ Rife Tensorrt: https://github.com/yuvraj108c/ComfyUI-Rife-Tensorrt , If your interested with this i made a ALL-IN-ONE click (Comfyui_portable and desktop version is Compatible) that will install everything, otherwise its tricky to install the node. Don't worrie it will not mess with your comfyui i made sure of it and tested it. Here is the link : https://civitai.com/models/2122077/rifetensorrt-interpolation-using-tensortcore-and-upscaler
Upscaler TensorRT ⚡: https://github.com/yuvraj108c/ComfyUI-Upscaler-Tensorrt . Once you have installed the ALL-IN-ONE click you can install the upscaler from comfyui manager.
Comfyui_Essentials: https://github.com/cubiq/ComfyUI_essentials?tab=readme-ov-file I had trouble installing it via manager idk why but if it happen. Then just Git clone into custom_node folder.
All the link to from model to clip vae are link in the workflow.
EDIT: Thing as been fixed it was on my part. You need to update the node mainly COMFYUI-GGUF i would recommand to update manually https://github.com/city96/ComfyUI-GGUF delete in your custom_node and git clone it. i had to the same for https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite . I really recommand to do it manually and install the requirements.txt but if your already up-to-date enjoy.
What you need
- Model files
- 0.6 GiB known subtotal
Remaining hardware or download-size estimates are not available yet.
Requirements
Models
7lightx2v_I2V_14B_480p_cfg_step_distill_rank256_bf16.safetensors
PossibleLoRA · Unknown
- Found
Checkpoint · 20.4 MB · PTH · Unknown
umt5_xxl_fp8_e4m3fn_scaled.safetensors
PossibleText encoder · Unknown
LoRA · 601 MB · SAFETENSOR · Unknown
- File unverified
VAE · Hugging Face · Comfy-Org/Wan_2.2_ComfyUI_Repackaged
wan2.2_i2v_high_noise_14B_Q4_K_S.gguf
File unverifiedCheckpoint · Unknown
wan2.2_i2v_low_noise_14B_Q4_K_S.gguf
File unverifiedCheckpoint · Unknown
Node packs
10- Found
Node pack · Registry
Node pack · Registry
- Found
Node pack · Registry
Node pack · Registry
ComfyUI-GGUF
PossibleNode pack · Registry
- Found
Node pack · Registry
Node pack · Registry
Node pack · Registry
Node pack · Registry
- Found
Node pack · Registry
Sources
1 source
Sources
1 sourceSource excerpts
2 excerpts
Source excerpts
2 excerptsSource context: 245 downloads · Type Workflows · Base model Wan Video 2.2 I2V-A14B
Source context: 223 downloads · Type Workflows · Base model Wan Video 2.2 I2V-A14B
