WAN 2.2 / SVI Pro 2 / I2V for 12GB VRAM
V2.1 - small fix, I accidentally removed the required SVI Low LoRa in Stage 2.
Obviously an easy fix but for beginners it might be easy to miss.
check my Aerith video (3rd video in the list) at the top. (Download Video then drag n' drop it into comfyui) it's a perfect example video of NAG negative prompts working correctly. Note the settings changing per stages. Sometimes nag_scale 8 has to be used with weighted prompts for it to work correctly. This is using all 7 stages and a good example of the many things you can do at CFG 1.
V2.0 - Adds about 30 seconds to generation time, due to NAG and Inject Latent Noise nodes.
SamplerCustomWithNAG replaced the old Custom Samplers. If you do not need Negative prompts and want faster gen times, stick with v1.31.
There is a lot of new stuff to this, so expect some trial and error.
Added CFG High and Low to the settings.
Shift is still 9 for better prompting but if you want to retain more face detail use Shift 5. It's a good trade off though.
Upscale Model:
please report any BUGS!
Pro Tip!
Mix and Match High/Low Checkpoints:
For example: Try using Smooth Mix High and Dasiwa's Low. I got some interesting results!
Inject Latent Noise
Reduces prompt bleeding/carry over to next stage.
SamplerCustomWithNAG
Open CMD: navigate to ComfyUI/custom_nodes folder.
Run: git clone https://github.com/BigStationW/ComfyUI-NAG
If it fails to install correctly (node not showing up):
File 1: ComfyUI/custom_nodes/ComfyUI-NAG/chroma/layers.py
Line 5:
Change:
from comfy.ldm.chroma.layers import DoubleStreamBlock, SingleStreamBlock
To:
from comfy.ldm.flux.layers import DoubleStreamBlock, SingleStreamBlock
File 2: ComfyUI/custom_nodes/ComfyUI-NAG/chroma/model.py
Lines 9–14:
Change to:
from comfy.ldm.flux.layers import DoubleStreamBlock, SingleStreamBlock
from comfy.ldm.flux.layers import timestep_embedding
from comfy.ldm.flux.layers import (
DoubleStreamBlock,
SingleStreamBlock,
)
from comfy.ldm.chroma.model import Chroma
after, close comfyui (don't restart).
v1.31 - QoL Update and change to Shift Values.
576x832, This is the resolution I've had the most success with.
overlap frames added from subgraph. Recommended to not touch unless you understand what you are changing.
Added Preview Animation and Control Random Seed to each sampler instead of a subgraph in the beginning, this way you can preview each clip as they finish, and let the generation continue endlessly if you're doing batches. Just remember to change it from Fixed -> Randomize. Also, don't forget to change it back to Fixed when happy with generation, otherwise it will just start at Sample 1 again.
Also, use Shift 9.0, it seems much better than 5.0. Don't forget when using ANY video LoRa, especially High LoRas, start them at 0.5-0.6, the low can usually stay safe at 1.0. This will avoid a lot jittery, jelly animations.
If you ever think you lost a seed... just drag the video into Comfyui workspace, and it will always have the seed number it used.
If you dont want endless 'preview animation' -> right-click Preview Window and select 'Reload Node', this can be done during generating as well. Or, you can CTRL+B before generation to disable it. This is a QoL update.
v1.2 - Please update to this version.
Seed per Sample - fixed.
Seed arrangement were out of order. (I'm not sure how this slipped through.)
Added labels on seed per sample.
Tested each seed on each clip and changed seed to make sure its working correctly.
Added a preview image after Resize Image so you can check and compare your image to the original.
For 832x1216 Images use 576x832 resolution if you're using the crop option on resize.
Modified version of [SVI Pro 2.0 for Low VRAM (8GB)]
And [Wan2.2 SVI Pro Example KJ]
7 Stage Sample Setup, with each Stage having their own Loras, combined with Sage Attention Cuda for faster speeds.
Can save each stage clip if needed.
Final Output w/ Upscaler + RIFE for smooth 60FPS.
Fast Group Bypasser - for quick access.
### Required Models & LoRAs
GGUF Main Models:
* [DaSiWa-Wan 2.2 I2V] or
* [Smooth Mix Version] or
* [Enhanced NSFW Camera Prompt Adherence]
> Note: Use a suitable quantization (e.g., Q4 or Q5) based on your available VRAM. I highly recommend DaSiWa-Wan high/low Models, as the Lightning Loras are BAKED in, leaving you only with SVI Loras being required.
SVI PRO LoRAs (Wan2.2-I2V-A14B):
* Both Required
Text Encoders:
[WAN UMT5] or
VAE:
The following is for Speed Boosts for nVidia Cards - If its already working then skip this!
Patch Sage Attention Node (sageattn_qk_int8_pv_fp16_cuda) + Model Patch Torch Settings Node (Faster Speed Times):
Prompt executed in 136.56 seconds <- Sage Attention Disable/FP16 Accumulation = Disable/Allow Compile = False
Prompt executed in 104.38 seconds <- Sage Attention Enabled/FP16 Accumulation = Enabled/Allow Compile = False
Prompt executed in 96.26 seconds <-- Sage Attention Enabled/FP16 Accumulation = True/Allow Compile = True
With this setup you can save a massive 40+ seconds just for one Stage!
If Sage Attention is not working/crashing comfyui then do the following or use (CTRL+B to bypass the nodes but I highly recommend getting it working for massive speed boost):
The following is for Comfyui_windows_portable, do not do it this way if you are using a different setup!
Step 1 — Check your PyTorch + CUDA version
Open CMD in your ComfyUI Portable folder (SAME directory as run_nvidia_gpu.bat) and run the following command:
.\python_embeded\python.exe -c "import torch; print(torch.__version__, torch.version.cuda)"
output = 2.9.1+cu130 13.0
check Python embeded version:
.\python_embeded\python.exe -V
output = Python 3.13.9
Which Means:
Python: 3.13 (embeded)
PyTorch: 2.9.1
CUDA: 13.0
Warning! If you are unsure how to proceed with the following steps, then paste your error code into Grok/ChatGPT
for a more detailed analysis.
Pick the wheel that matches your Python + PyTorch + CUDA output from Step 1.
That means the correct SageAttention wheel for your setup would be something like this:
sageattention-2.2.0.post3+cu130torch2.9.0-cp313-cp313-win_amd64.whl
download the correct wheel for your setup from:
It matches Python 3.13 (cp313-cp313), PyTorch 2.9.x, and CUDA 13.0.
The slight difference in patch version (2.9.1 vs 2.9.0) is fine — this wheel works with PyTorch 2.9.x.
Step 2 — Install Wheel (make sure the file is in \ComfyUI_windows_portable, same directory as run_nvidia_gpu.bat)
Open CMD in your ComfyUI Portable folder and run with the correct wheel file (example below):
.\python_embeded\python.exe -m pip install "sageattention-2.2.0.post3+cu130torch2.9.0-cp313-cp313-win_amd64.whl"
Step 3 — How to check if it works:
Open CMD in your ComfyUI Portable folder and run:
.\python_embeded\python.exe -c "import sageattention; print('SageAttention import successful!'); print(dir(sageattention))"
You should see:
SageAttention import successful!
['__builtins__', '__cached__', '__doc__', '__file__', '__loader__', '__name__', '__package__', '__path__', '__spec__', '_fused', '_qattn_sm80', '_qattn_sm89', '_qattn_sm90', 'core', 'quant', 'sageattn', 'sageattn_qk_int8_pv_fp16_cuda', 'sageattn_qk_int8_pv_fp16_triton', 'sageattn_qk_int8_pv_fp8_cuda', 'sageattn_qk_int8_pv_fp8_cuda_sm90', 'sageattn_varlen', 'triton']
Step 4 — confirm if triton attention mode is available:
Open CMD in your ComfyUI Portable folder and run:
.\python_embeded\python.exe -c "import sageattention; print('SageAttention import successful!'); print('Triton mode available:' , hasattr(sageattention, 'sageattn_qk_int8_pv_fp16_triton'))"
You should see:
SageAttention import successful!
Triton mode available: True
if any triton errors run this command:
.\python_embeded\python.exe -m pip install triton
Step 5 - now you should be able to use "sageattn_qk_int8_pv_fp16_cuda" with Patch Sage Attention + Model patch Torch Settings Nodes properly.
Description
updated default settings. If you're a beginner please read the tips at the bottom of notes.
FAQ
Comments (29)
I get reconnecting after "patching torch setting" on sampler start
paste the full error if you can.
@oatis81259 no error in console, just red box appears with "reconnecting" in right corner
I've never seen this before but it definitely sounds like something is silently crashing.
Change the sage mode to auto. The node is under the prompt box.
@Proxy00 just be aware switching to auto you will not be saving the max amount of time.
It works for me, however, every single video comes out extremely grainy and low quality. Not sure what the problem is.
what resolution you using?
@oatis81259 Sorry, I actually got it working. I believe I was using the incorrect SVI loras. Thanks for your work!
This workflow is solid! Love the option to enable the samplers as you go to help stage things. Really helps keep the flow nice.
I get a GGUFLoaderKJ
'dict' object has no attribute 'startswith' Error when I run this workflow. no idea why since everything is in its right place.
did you rewire anything?
@oatis81259
nope, i downloaded it again and ran it with the example image without touching anything since i have all the models in the same place. I get the same error so yeah I'm not entirely sure why. I also tried it on both comfy desktop and portable.
scratch that, looks like i needed an update to the gguf nodes. it works now.
@fjphoenix hwo to update kj ggufnodes? i got this GGUFLoaderKJ
invalid character '·' (U+00B7) (nodes.py, line 122)
@ponystalk69990 Just used comfy manager
Hope to have a non GGUF version of the workflow
Thank you for the workflow. How do you upscale longer videos, it always crashes my ram (i have 64gb).
if its crashing on Rife then remove the node connected to clear_cache_after_n_frames, and set it to something low like 2-4. If it still crashes reduce Multiplier from 4x to 2x.
i forgot to mention try to use only a 2x upscale model, 4x will definitely crash on this with 12GB of VRAM. Other than that i suggest the first thing I mentioned above. I'm also using 64gb of ram, with the default upscale template it doesn't crash for me, so make sure you have nothing else open on your PC, you need every bit of VRAM you can get, when you boot up windows you should have 0.4-0.5 used up of vram. If more is used open task manager and see whats eating it up.
The only WF that worked for me, thx brother.
Hi there, thanks for this workflow I've mostly got it working but for some reason the Fast Groups Bypasser says "Enable Models / Loras / Image / Settings" in every field and doesn't turn the sample stages on/off only the Models / Loras / Image / Settings stage. I've used ctrl+drag select to bypass stage 2 but instead of getting a 10 second clip I'm still getting a 5 second clip but with the combined loras of stage 1 and 2. Any suggestions? I assume the idea was that each stage would apply a new lora over the next 5 seconds of the video?
Sounds like you got some outdated plugins or an outdated Comfyui? Are you using the web version or application? It's been tested multiple times on the web version to be working with the latest comfyui. I'm not entirely sure what problem is, I'd start with those things first. Let me know how it works out. Could also do a fresh install into a separate directory without affecting your current comfyui.
@oatis81259 thanks for the help, the fresh install route ended up working out, looks like I didn't realise installing tons of custom nodes for different workflows was a terrible idea!
@mistergoodbytes464 I'm not sure if this will help everyone, but disabling Nodes 2.0 fixed the issue for me.
Edit this file: ComfyUI\user\default\comfy.settings.json
Change this line to: "Comfy.VueNodes.Enabled": false,
Word to the wise, if you set dequant_dtype and patch_dtype to target on the GGUFLoaders, that speeds up switching between the models a good deal. patch_on_device may also play a role in this, I forgot, but that's how I had it set on another workflow before :P
Absolutely fantastic workflow that really changed how I work with Wan. I had to change it to pull the seed gen outisde the subgraph because that's known to cause problems with randomisation. I also added the same input for frame length so now each section can be a different length.
Plus I added the new NVidia RTX node for 2x upscaling along with a slight change to the Rife interpolation letting me gen at 12fps and interpolate to 24fps, that's all I needed.
This WF was a real game changer for me though, absolutely love it. Thank you for sharing.
Thanks I appreciate the kind words!
I'm not sure what you mean about the seed gen, it is outside of the subgraph and in the main section with 1 seed per section.
Anyways, different lengths per section is actually interesting though, I'll play around with it since sometimes some sections dont flow the way you want.
@oatis81259 Maybe it was a ComfyUI problem on my setup then because if I changed the seed it wouldn't regenerate unless I refreshed the page or changed the prompt. I had seen a problem previously with the Randomise function getting stuck inside a subgraph so I thought it was the same thing.
Either that or I was working with two different workflows and got confused about which one had the problem... I am an old man so that happens a lot, I solve all kinds of problems that don't really exist... sorry about that. ☹
@youtougle2422 no worries, I'll retest again to make sure the seed gen outside is working correctly. Comfyui is definitely NOT comfy lol. I'd be using Forge if they could produce longer videos like comfy can. Forge can make some really good clips.