DaSiWa WAN 2.2 I2V 14B Lightspeed

darksidewalkerdarksidewalker
738
15K
base model
LureNoir HighLureNoir LowRadiantCrush HighRadiantCrush LowTeasingKiss HighTeasingKiss LowCheekyDream HighCheekyDream LowHotSpring HighHotSpring LowSweetSpot HighSweetSpot Low

Details

Download Files

Model description

🗡️💀 DaSiWa-WAN 2.2 I2V 14B Lightspeed 💀🗡️

🔮 Key Features:

  • 🔥 LoRA-Free Generations
    Generate high-quality videos without stacking Wan 2.2 LoRAs (unless you want adding spacial styles/concepts).

  • ☄️Fast: 4 step generation

  • Extreme versatile (more build in concepts)

  • Quality motions (less slowdowns)

  • 🔞 NSFW:

    • Enhanced anatomy + poses + framing

    • NSFW (BIAS) and SFW capable

    • Better understanding of sexual concepts

    • You know, if you know! 💦

  • 🪄 Better Prompt Responsiveness

  • 🥺👉👈Better understanding of anime/manga style composition

  • 🪡 FP8 precision

  • ⚠️ Read version details for the version you are using for more information!

  • It is not perfect and has some minor issues, but it can shine at some points I could not create with some LoRAs!

  • ⚠️Do not use any extra speed-up LoRAs, this is baked in already

🍒Make sure to checkout my easy to use Workflow!

🛠️ Recommended Settings

  • Steps: 4

  • CFG: 1

  • Sampler/Scheduler: Euler/Normal or UniPC/Simple

  • Resolution up to 720p (native quality).

My go to settings:

  • Aspect ratio 3:4 (784x1136) or 9:16 (720x1264)

  • CFG 1

  • Euler/Normal

  • 4 steps

  • 16 fps

  • Sigma Shift: 5

  • Add other LoRAs with 0.6-1

  • 16 fps, 81 frames ~ 5s

VAE: Wan 2.1 VAE

CLIP: umt5_xxl_fp8_e4m3fn_scaled

💫 Speed + Examples

On 16GB VRAM, 64GB RAM, 4 steps, cfg 1, 81 frames

  • 480 pixel side length (384x576): 70-90 sec

  • 720 pixel side length (592x848): 120-140 sec

  • 768 pixel side length (576x1008): 3 min

  • 896 pixel side length (672x1184): 6 min

  • 960 pixel side length (784x1136): 9-11 min

  • All examples are without any additional LoRAs

  • Initial example images made by me are made with my model 🗡️💀 DaSiWa-Illustrious-XL 💀🗡️

  • Some initial examples are borrowed - 🖤 the creators for their awesome art! 😊

🩻 Known issues

  • Tell me 🫵🫢

🩺 Fixes & Feedback

  • If you use LoRAs, try to respect the LoRA training triggers and try some versatile descriptions, most LoRAs will work with 0.6-1.2 (start with 1.0)

  • Do not mass add LoRAs, just add 1 or 2 (x2 High+Low)

  • Negative prompting do not work with cfg 1, thats a limitation of speed-ups with cfg 1

  • Low resolution (e.g.384x576) are only for fast samples and will blur fine details, do a higher resolution if you want clear details

Before posting any questions I suggest reading my guide.

🖤 Why I Made This

I was tired of using all these massive list of LoRAs, just to get a remotely good result after 10 generations, consuming hours of time.
So I can just make my videos with 1 or 2 concept LoRAs without pushing 6 till 10 LoRAs (Low/High) into a generation.

This checkpoint is also my personal playground.

Closing words

🤩Inspired by

  • Lightx2v

  • ComfyQuants

  • Lightning LoRAs

  • ... and more

I want to thank all the fantastic other creators who made super nice LoRAs and concepts to play with!

🐞Recommended offline Frontend/UI: SwarmUI

YOU are responsible for outputs as always! If you make ToS violating content and I get aware I WILL report this.

Images made by this model

No Images Found.