⚡ Recommendation
You can use my custom checkpoint V2 NSFW, Lightning included:
➡️ WAN 2.2 Enhanced – Lightning Edition
for wf lightningedition long v2.1
https://civarchive.com/models/2053259?modelVersionId=2477539
Or for wf svi , no Lightning included:
➡️ SVI NSFW
https://civarchive.com/models/2053259?modelVersionId=2609141
Update: 2 or 3 KSamplers
The triple-KSampler workflow added dynamism to avoid slowdowns, but it could cause issues in certain sections, such as high brightness or other bugs. To fix this, simply enable the Double KSampler in the problematic section and change the seed.
🟣 uptade SVI LoRAs – Strengths & Weaknesses
⚡ Overview
✅ Strengths
Best solution for making long videos
Excellent transitions between video segments
Reduced degradation compared to other solutions
Strong character coherence: the model retains information from the previous video, helping maintain consistency
⚠️ Weaknesses
Weaker prompt understanding
Weaker camera understanding
Videos are less dynamic
Sometimes slow-motion effect
(can be improved with proper Lightning LoRAs and dynamic prompts)
🟣 SVI LoRAs – Download Links
⚡ Download
Note: Both LoRAs must be loaded manually in your workflow.
🟣 Suggested Lightning LoRA Combos (Optional)
⚡ Overview
You don’t have to use these Lightning LoRA combos. They are optional and allow you to fine-tune motion and degradation.
You can also use other Lightning LoRAs or assign different combos per video for more control.
🔥 Combo 1 – More Motion (Rapid Video Degradation)
High LoRA →
https://huggingface.co/Kijai/WanVideo_comfy/blob/709844db75d2e15582cf204e9a0b5e12b23a35dd/Lightx2v/lightx2v_I2V_14B_480p_cfg_step_distill_rank128_bf16.safetensors
Weight: 4
or
💜 Combo 2 – Less Image Degradation
or
or
🟢 Combo 3 – Balanced Motion / Moderate Degradation
High LoRA →
https://huggingface.co/Kijai/WanVideo_comfy/blob/709844db75d2e15582cf204e9a0b5e12b23a35dd/Lightx2v/lightx2v_I2V_14B_480p_cfg_step_distill_rank128_bf16.safetensors
Weight: 3
Low LoRA →
https://huggingface.co/Kijai/WanVideo_comfy/blob/709844db75d2e15582cf204e9a0b5e12b23a35dd/Lightx2v/lightx2v_I2V_14B_480p_cfg_step_distill_rank128_bf16.safetensors
Weight: 1.5
If you want to create several-minute-long videos while maintaining high quality, it is possible, but it will take a very long time. You just need to avoid using Lightning LoRAs and use the full model instead.
🧠 Advanced Usage Tip
You can disable global Lightning LoRAs and assign different combos per video:
Combo 1 for Video 1, Combo 3 for Video 2, Combo 2 for Video 3.
Each combo produces different motion and degradation behavior.
If you don’t use Lightning LoRAs, degradation will be even lower, but generation time will be very long.
🧠 Dynamic Prompts – Better Control & More Motion
You can use dynamic prompts to have better control and this helps to make the video more dynamic.
You just need to give this example prompt to an LLM like ChatGPT. It will be enough for it to describe the image you have and what you want as a video while keeping the same structure of the prompt in the following example.
⚠️ This will be a NSFW prompt; ChatGPT will not accept it.
You can use GROK https://grok.com, which can make NSFW prompt modifications.
For more examples of prompts with different poses, check:
Enhanced FP8 Model.
Give these prompts to GROK with the dynamic prompt structure if you want.
Example 1
(At 0 seconds: Wide shot showing a slightly overweight man casually walking down a city street, camera fixed in front, urban environment with buildings and cars.)
(At 1 second: Suddenly, a massive shark bursts from the pavement ahead, looking terrifying at first, pavement cracking, dust and debris flying, camera from side angle.)
(At 2 seconds: Medium shot from the side, the man stumbles backward in shock, while the shark dramatically slows down and strikes a comically exaggerated sexy pose, revealing large, exaggerated shark breasts, covered by a colorful bikini.)
(At 3 seconds: Close-up on the man’s face, eyes wide in disbelief, as he turns to look at the shark, small cartoon-style hearts floating above his head to emphasize his amazement, camera slightly low-angle.)
(At 4 seconds: Dynamic travelling shot showing the man frozen in the street, the shark maintaining its sexy pose, water splashes and debris still moving realistically, urban chaos around.)
(At 5 seconds: Wide cinematic shot pulling back, showing the man standing in the street, staring at the bikini-wearing shark with hearts above his head, epic perspective highlighting absurdity and humor.)
Example 2 – Anime NSFW
(At 0 seconds: The couple in a cozy bedroom, anime style, soft lighting highlighting their intimate embrace, her back arched slightly as he positions himself.)
(At 1 second: The man’s hips moving rhythmically, the head of his penis sliding effortlessly into her vagina, her body responding with a gentle, fluid motion, anime-style motion lines emphasizing the smooth penetration.)
(At 2 seconds: Her back arching deeply against him to intensify the pleasure, hips swaying with each thrust, breasts bouncing subtly, small hearts floating around them to capture the erotic energy.)
(At 3 seconds: Her face, eyes closed in bliss, a soft moan escaping, hands resting behind her head, anime-style blush on her cheeks, the air filled with a seductive aura.)
(At 4 seconds: The man penetrating her deeply, her body moving in sync with his, the bed sheets slightly rumpled, the room’s warm lighting enhancing the intimate, lustful atmosphere.)
(At 5 seconds: The couple locked in a passionate embrace, the scene exuding vibrant, seductive energy, anime style with smooth lines and soft shadows.)
findperfectresolution
https://github.com/ashtar1984/comfyui-find-perfect-resolution
rgtree
https://github.com/rgthree/rgthree-comfy
kj-node
https://github.com/kijai/ComfyUI-KJNodes
iamccs
https://github.com/IAMCCS/IAMCCS-nodes
🔧 What’s New – V2.1
The workflow is now much simpler than before.
You can create videos up to 20 seconds.
New sections can be added very easily by copying and pasting an existing section.
🎨 New Node: PainterI2VAdvanced
🔗 https://github.com/princepainter/ComfyUI-PainterI2Vadvanced
Main Parameters
color_protect
Enables or disables color drift correction.
Recommended: True.
Prevents green tint, darkening, and color distortion.
correct_strength
Controls the strength of color correction.
Higher value = stronger color correction
findperfectresolution
https://github.com/ashtar1984/comfyui-find-perfect-resolution
_______________________________________________________________________________________________
New FunVace workflow: it's extremely easy to use. If you have separate videos and want to join them with a seamless transition, use this! Thanks to Bob once again, excellent work: https://civarchive.com/models/2277993?modelVersionId=2563927
Workflow Update – Lightning Edition V2
📌 New Version Available
The Lightning Edition V2 is the latest update of my workflow.
This version is cleaner, better organized, and scene transitions are improved for smoother results.
🔧 Fixes & Improvements – Lightning Edition V2
Workflow fully reorganized and clarified for better readability.
Scene transitions improved for smoother results.
Optimized nodes and settings for easier use.
Confirmed compatibility with the Lightning Edition V2 model.
🔹 V1
Version without Super Power LoRA
A version without Super Power LoRA has been added for those who are more cautious.
If you want to try the Super Power LoRA without using the folder included in the ZIP, check the GitHub page:
➡️ GitHub – rgthree workflow
Tutorial
The workflow is divided into 6 sections:
Section 1: generates a 5-second video.
Section 5: handles transitions and interpolation.
Important notes:
Running only sections 1 + 2 will not merge the video.
To create a 15-second video, run sections 1 + 2 + 5.
Interpolation can also be enabled in section 1 if you prefer.
You can use different prompts and LoRAs for each section.
Resolution is automatically adjusted to avoid cropping.
💥 Super Power LoRA – V1
Requires the custom rgthree node:
➡️ rgthree ComfyUIInstall via git clone or the Manager, then replace the folder with the one included in the ZIP.
Allows you to save and load templates, very useful when working with multiple LoRAs, especially for WAN 2.2.
A version without Super Power LoRA is also available for those who prefer.
For the version with Super Power LoRA, see: GitHub – rgthree workflow
✅ Tips – V1
Keep the default settings for fast results.
Increase the steps if you want better quality.
Description
FAQ
Comments (38)
Hi sir and thanks for your effort for the workflow and even for the models you released. But sadly i have an annoying problem. I don't know why but many times i've eyes glitching in my videos. At the first frame it's fine but after eyes start to glitch. The only time i had success without glitch was with 720x1280 res but it took 27mn on my 3090 for a 15 sec video. Have you another idea ? I use your FP8 models with 3 steps on High pass and 2 steps on Low pass.
Hello and thank you for your feedback. I also have this problem when I use certain LoRAs: I have to lower the strength and it works well. If you are not using a LoRA, a higher resolution is the best way to get a good video, but I am surprised that it takes so long with your card. I have a colleague who also has a 3090 and it is much faster. How much RAM do you have? Have you tried the GGUF versions?
@taek75799 hmmm i'm using mql_I2V_doggystyle lora and male and futanaris animation lora from smoothmix collection. Yep for a 15 sec video it took this time. Actually i've 64 GB of DDR4, i enable Sage attention node but i think i'll probably try your Q8 Checkpoint since FP16 acceleration works on it. FP8 torch compile isn't fully support on Ampere architecture sadly. I'll try your advice and thanks for your response sir :D. Atm i'm using 720x720 or 480x832 res
Ого какой рабочий процесс! Огромная благодарность. работает как автомат калашников.. у кого артефакты используйте последние модели fast и шагов больше ставьте!! Всех с наступающим Новым Годом! Всех благ!!)))
Привет! Спасибо за ответ. Извини, я пользуюсь переводчиком, так как не говорю по-русски (ха-ха). Спасибо за советы! С наступающим тебя и всю Россию!
Hi and thanks for the workflow. But for some reason my output is just blurry.
A Little bit of info on how I used your WF:
I use your model "NSFW FAST MOVE V2 Q8 H and L" in the Unet Loader. In the Diffusion Model High and Low Node I use "wan2.2_i2v_high_noise_14B_fp16.safetensors" same for the low node just with the low model of course.
Load Clip --> "umt5_xxl_fp16.safetensors
vae --> wan_2.1_vae
Can you somehow help me. Should I use the fp8 models? I´m running a 5090
Share your workflow
Hello, I didn't quite understand are you using the Q8 (high and low) and also the FP16 safetensor at the same time?
If you want, send the workflow to qek, he will be able to help you.
@taek75799 I´ve sent the WF to qek.
To clarify it. At the top is the GGUF Loader "Unet Loader GGUF High" and the "Unet Loader GGUF Low"
I downloaded both of your Models and put it in there (the low and the high).
In the Load Diffusion Model High and Low I put in those models: wan2.2_i2v_high_noise_14B_fp16.safetensors and wan2.2_i2v_low_noise_14B_fp16.safetensors
I shared my wf with qek
@keralo The two cannot work at the same time you must connect either my models (the Q8 ones) or the FP16 ones. However, I don't think that's the problem. Gek should be able to help you don't hesitate to come back if you encounter any other issues. Either I or others will be able to find a solution. Thank you for your feedback.
I feel so stupid right now. I fixed it. Both of the GGUF nodes werent connected. I disconnected the Diffusion Model nodes and connected the Unet Nodes/GGUF Nodes to the wf and now it works -_- sorry
and thanks for your help
EDIT: wow such a nice gguf model big thanks!
@keralo Haha, don't worry about it, we all make mistakes and I've done much worse than you! I'm glad it's fixed and thank you for your feedback.
this is so nice to work with! is it possible to extend the sessions to have longer vids than 15 secs
Hello, yes you can extend it further, but be careful the more you extend, the more the quality will degrade. This is why I didn't go beyond 15 seconds. However, there is a solution beyond 15 seconds: take the last frame of your video and upscale it with a good upscaler.
I use either SeedVR2 (which unfortunately doesn't always give good results, but when it does, it's incredible), Topaz Photo, or a ComfyUI upscaler with a small denoise. Then, load this upscaled image into the workflow. If you want to extend it even more now (for example, 4 sections), copy the 2nd section: open the subgraph and copy the merge, upscale, and KSampler nodes. Just paste them to create a 3rd section.
I know this might seem a bit difficult at first. In fact, if I have time, I'll probably add 5 sections as I did in my previous versions, so that people have as many choices as possible. And thanks for your feedback!
@taek75799 youre so kind haha i am not used to this anymore, thank you ill give it a shot and fingers crossed for a 5 sections one too xoxo
@kikiazusa856 Alright, thanks. If you can't manage to do it, I'll do it, but I can't tell you when I'm a bit busy at the moment.
@taek75799 I tried but failed, when you find time it would be amazing if you can make one like that, keep the good work, take care and thanks again
@taek75799 little update, I was able to modify it to infinite extensions, I made a copy of the Stage 2, called Stage 2a, I wired the Stage 2 IMAGE to Stage 2a previous video, after this I had to delete Stage 3 because it was looping twice Stage 1, but it works as long as I make a copy of Stage 2a into Stage 2b etc, its a messy but its working for now.
@kikiazusa856 Well done then, and thanks for your feedback! To be honest, this workflow is already obsolete, lol. I'm currently working on a new one. The AI world moves very fast, but that's for the best: we're reaching near-perfection now. Honestly, I can barely see where the transition is. I'll try to post it in a few hours.
@taek75799 cant wait!
@kikiazusa856 And I thought of you: I did a 20-second one and you can extend it even further, very easily, you'll see. On the other hand, I'm not a fan of anything longer than 15 seconds.
Cool workflow, thanks.
But i'm somehow struggling to test it on a standard diffusion model. It gives me blurry results, I've tried several models (14B_fp8, or 14B_fp16, or fp8_e4m3fn_scaled) with appropriate text encoders, and they give me blurry results. I've also tried standard q8 GGUF - still blurry. It's only working good with your checkpoint (FASTMOVEV2FP8H). Is there another model I need to get in order to get it to work?
Hello, my model contains the Lightning lorass that’s why it works well with the settings, meaning 2+2 steps on the samplers. If you want to use other models that don't contain Lightning loras, you either need to use Lightning loras to generate with few steps, or add steps to the various KSamplers (for example 10+10). To access the KSampler, go to the subgraph nodes and click at the top left.
@taek75799 thanks, all is clear now. BTW your model is superior, I just like to test other stuff also
hi, im a bit confused as to which models and checkpoints to download. in UNET Loader GGUF High/Low i put the NSFW Fast Move V2 Q8 H/L files? What do i put in the diffusion box then? whats the best combination for Q8? or is it just the flux2_dev _fp8 one? Shouldnt it be Q8 too?
Hello, the best choice is the Q8 models: they are the most powerful. Load them into a "GGUF Loader" node and connect the links instead of the "Load Diffusion Model" nodes. Having both "GGUF Loader" and "Load Diffusion Model" nodes is intended to let people choose between GGUF or FP8. In your case, if you choose the Q8, the "Load Diffusion Model" node will therefore be disconnected.
@taek75799 Thank you! Which Model do you recommend for "Load Clip"? And how do i properly generate the videos so they are chained together? i couldnt find an instruction sadly
@ssense996175 Re-download the latest workflow; I updated a note to explain a bit. Regarding the CLIP, fp16 will be better, even though fp8 works fine.
@taek75799 okay perfect, is the linked bf16 clip the same as fp16? or should i use a fp16 one
@ssense996175 The bf16 version seems to be "NSFW", while the normal one is fp16. Check the description; there is an example comparing the two. Let's just say that if you're doing a sexual pose, it helps, and it works for horror too, but it can also introduce some issues.
Hello !
Very good workflow ! the first video is always awesome but for some reason I have a problem with the consistancy of the person used in the first image in the second video.
It always change the face/hairs ...
Any idea on how to fix that ? (I'm using V2 Fastmove Q8H)
Unfortunately, with Wan, we have a limit. Keep one thing in mind to start: try to have the cleanest image possible. If you can upscale it, that will help. Also, the resolution is important; by default, it's at 720x720. If you can, don't hesitate to increase it. But the most important thing is that Wan generates 5-second videos. So your first video will be 5 seconds long; at the last frame of this video, the workflow simply takes that last image, upscales it, and creates a continuation. If the face looks good on that last frame, the rest will be good. On this V2, I only included 3 sections, whereas there were 5 on the others. I decided to only include 3 because with each generation, the quality will degrade. Unfortunately, we have no choice.
@taek75799 I wanted to say thank you for taking your time to explain all this. Your updated workflow is super good and worked right out of the box.
hey @taek75799! great job with the workflows and model! one thing i cant get is better quality. the video looks like its all blurry with enormous pixel. When i increase the step (to whatever above 4 mostly) i just get noise on the video, no images at all? How to make it a high resolution video?
in the description, you do say to increase the steps for better quality, but for me, its just produce noise when i do increase it while keeping every other params as you do provide them.
@DoumLapointe Hello, could you send me the workflow that is giving you poor quality?
hey @taek75799, i was missing the guff and for some reasons, comfyui was still going through the process and giving me really poor quality video. Ive fixed it and managed to generate great stuff! thanks for all your work!
@DoumLapointe Thanks for your feedback, good news then! I should be releasing a new workflow, but this time the transitions will be perfect. Don't hesitate to give it a try.