WAN2.2 I2V GGUF NSFW (8GB VRAM / 32GB RAM) WORKFLOW
Details
Download Files
About this version
Model description
GOONING WORKFLOW FOR THE VRAM POOR!
If you are VRAM poor just like me, this workflow is for you! You can generate NSFW videos with just 8GB VRAM and 32 GB RAM. Maybe even with lower specs if you use lower GGUF models. Everything is written as notes in the ComfyUI workflow but I will write it them again here.
STEP 1 - MODELS
WAN2.2 I2V A14B GGUF:
Put WAN GGUF or SMOOTH MIX WAN GGUF models under unet folder.
I recommend Q5_K_M for 8GB VRAM, you can download smaller versions if you have less VRAM or bigger versions if you have more.
Text encoder GGUF:
I recommend Q5_K_M for 8GB VRAM, you can download smaller versions if you have less VRAM or bigger versions if you have more.
VAE:
CLIP VISION:
STEP 2 - LORAs
LoRAs:
!! DO NOT USE IT WITH SMOOTHMIX WAN! SMOOTHMIX ALREADY HAS LIGHTX2V BAKED IN !!
This LORA reduces generation time by A LOT. I do NOT recommend removing this LORA.
All-in-one general NSFW LORA. Don't forget to download both HIGH and LOW versions. You can use this LORA together with other LORAs; however, not every other LORA works well together. You need to test it.
If you want to add more LORAs, just add them~!
STEP 3 - IMAGE AND PROMPT
START IMAGE:
The image proportions should be the same as video generation proportions. For example, if you are putting a 16:9 image, your generation proportion should be 16:9. Otherwise, weird things might happen.
END IMAGE:
You can disable the end image by selecting it and pressing CTRL+B. I do NOT recommend putting the START IMAGE as END IMAGE, it reduces the movement and motion a lot. However, feel free to experiment.
PROMPTS:
Check CIVITAI generations for more prompts and keywords for the LORAs you are using. As for the negative prompt, I have no idea what's best. Chinese? English? Less keywords? More keywords? No idea.
STEP 4 - WAN PROCESS
STEPS ( ! IMPORTANT ! ):
If you are using WAN2.2 14B I2V, total steps is 4 with LIGHTX2V LORA. If you are using SMOOTH MIX WAN, total steps is 6 WITHOUT LIGHTX2V LORA. Make sure you change BOTH KSampler steps.
VIDEO SIZE & LENGTH:
Dimensions must be divisible by 32! For quick generations (testing), use these dimensions:
- 512 x 512 (SQUARE)
- 480 x 864 (9:16)
- 864 x 480 (16:9)
For final render, this is what my machine was capable of (8GB VRAM / 32GB RAM):
- 864 x 864 (SQUARE)
- 576 x 1024 (9:16)
- 1024 x 576 (16:9)
LENGTH:
81 frames for 5 seconds. I do not recommend trying longer or shorter duration using this workflow. But if you must, the frame length must be divisible by 16 + 1.
KSampler:
Change noise_seed generation from "randomize" to "fixed" if you are happy with the testing result but want a higher resolution. Otherwise, leave everything else as it is if you do not know what you are doing.
STEP 5.1 & 6 - Upscale and Frame Interpolation
DISABLE THESE WHILE TESTING OUTPUT (CTRL+B)
UPSCALE:
This model works great with anime images/videos. Feel free try other models.
FRAME INTERPOLATION
Free FPS increase! If you do not like the results, you can disable the frame interpolation and just save the upscaled video.