Workflow Overview: Lightspeed Wan Series
These workflows are optimized for the Lightspeed Wan 2.1 and 2.2 Image-to-Video (I2V) models by darksidewalker. They are categorized into the streamlined SVI-V2-Pro series and the feature-rich Standard (v12 and below) series.
1. SVI-V2-Pro-I2V (Barebone Series)
The SVI series is designed for users who want a clean workspace without unnecessary overhead. It focuses on the core I2V logic with two critical utility modules.
A. Resolution Management (Resolution Master)
To ensure compatibility with Wan 2.2’s specific latent requirements, the source image must be precisely resized.
Step 1: Load the source image into the Load Image node.
Step 2: Use the dropdown (next to the TV icon) to select a target resolution (e.g., 480p, 720p).
Step 3: Click the TV Icon to activate the selection.
Step 4: Under 'Actions', set the slider to 16 and click the Snap button.
Note: This "snapping" ensures the dimensions are multiples of 16, preventing tiling artifacts or model errors.
B. Temporal Smoothing (Interpolation)
Located in a dedicated group, this module handles frame rate conversion.
Primary Use: Converting low-FPS raw generations (e.g., 16 FPS) into smooth, high-fidelity output (e.g., 60 FPS).
2. Standard Wan Workflows (v12 and Below)
The legacy/standard workflows are "all-in-one" solutions, containing a broader suite of post-processing and utility tools:
Image to Video
The core generative engine for Wan 2.1/2.2.
FLF Loop
Facilitates seamless looping by utilizing First-Last Frame logic.
Downscale Image
Safely shrinks oversized source images to prevent VRAM OOM (Out of Memory) errors.
Extract Last Frame
Pulls the final frame of a generation to be used as a starting point for the next clip (Storyboarding).
Combine Videos
Merges multiple generated clips into a single sequence.
Interpolation
Increases frame rate for smoother motion.
Upscale Video
Increases the spatial resolution of the final video output.
Description
Removed the Load Upscale Model and Upscale Image (Using Model) nodes from the Image to Video group.
If your input image has jagged edges in it, they will probably remain in the video. Having those (now removed) nodes helped reduce that when using the right upscale models, but at the same time apparently confused at least one person as to what its function was (even with the note there).
On the plus side, this reduced the width of the Image to Video group.
Added a Basic Upscale group to the workflow.
It literally only upscales the video. No denoise. No frills. SUPER basic.
This is not something I plan on using and have no desire to, unless I suddenly come into having better hardware.
Other than that, the workflow is the same as v5.
FAQ
Comments (9)
Thanks for the workflow. It renders super fast, and I appreciate the notes throughout. It may be disconfigured, however. In your Wan2_2-480p-i2v_v6.json workflow, you have High Model > CLIP running to CR Clip Input Switch > 1: Checkpoint Clip, and Load CLIP > CLIP running to CR Clip Input Switch > 2: Load CLIP. I tried running the workflow out of the box after adding the checkpoints (Darksidewalker's Wan 2.2 LureNoire High and LureNoir Low), but kept getting an “CLIPTextEncode - ERROR: clip input is invalid” error.
It wasn’t until I connected the CLIP of the Load CLIP node to the CR Clip Input Switch > 1: Checkpoint Clip, and another connector from Load Clip > CLIP to CR Clip Input Switch > 2: Load CLIP that the workflow executed without any errors.
There is also something in the workflow that is causing the outputs to hallucinate. In every generation, the character remains, but the background scene changes to something completely different. It even had a penis appear out of nowhere, and the chick just hopped on it and started riding! 🤣 I did not add any LoRAs.
I checked it against Darksidewalker's DaSiWa WAN 2.2 i2v FastFidelity Comfy FLF (1.0).json workflow with the same LureNoir checkpoints and no LoRAs, and the I2V held up and did not produce any scene changes.
Any idea what I need to change? I already tried changing “noise_seed” to 0 and “control after generate” to “fixed” in the Wan MoE KSampler node. Should I instead be tweaking the steps in the Wan MoE Schedule node (both High and Low are set to 2)?
that kinda sounds like what was happening to me when the Early Access model from DarkSideWalker was messed up. So not really sure what's causing that for you tbh regarding the weird hallucinations.
As far as the CLIP error: the checkpoint CLIP should be connected to where it says Checkpoint CLIP on the switch and the Load CLIP should be connected to the other. That error appears when the wrong one is toggled, but I can double check the connections on the saved workflow. I still kinda prefer RadiantCrush that has the CLIP baked-in, otherwise the switch isn't really needed for convenience.
If you're using LureNoire, then the CLIP switch should have been set to "2" to use the Load CLIP node, since DSW didn't bake the CLIP in on that model AFAIK.
Note: I just ran the LureNoire model with no issues. I did move one of the model noodles, but that's just to connect the lora loader to the KSampler. Otherwise, the connections should be the same. I'll reupload the workflow file JIC.
@Gladas I will try that, thanks! On another note, Darksidewalker's Wan 2.2 LureNoire High and LureNoir Low didn't seem to play well with this model for me. Only when I changed the checkpoints to Radiantcrush High/Low did the background switching stop.
@PMAX365 yeah that's 100% what was happening for me before DSW reuploaded LuneNoire. I reuploaded the workflow, but if you're using the previous upload of LuneNoire, it will probably hallucinate still.
@Gladas good to know! I will re-download both the LureNoir High/Low checkpoints to see if that makes a difference
Re-uploaded the v6 workflow on 27 October 855PM UTC.
If you're not using RadiantCrush or a model that has the CLIP baked-in, remember to toggle the CLIP switch to "2".
Also, make sure to toggle your model/clip/vae in the loaders to make sure they are actually pointing to your model/clip/vae.
Appreciate the update!
Update: v6 worked right out of the box! And no more background switching after updating the LureNoir High/Low models from Darksidewalker. Thanks for the update to the workflow and for its continued support! Just blessed you with 800 buzz ⚡️
@PMAX365 glad to hear it and thanks!