Back
MILF SWINGING
1 day ago
333
0:59
Generation Instructions:
1. Buy a Macbook M3 Max 48GB or equivalent hardware
2. Install ComfyUI for MacOS.
3. Inside ComfyUI, there's a "template" button, which can be used to start generating immediately. But do not download any models from ComfyUI itself, because the suggested models are in the FP8 format which is incompatible with Apple Silicon. The MacOS compatible replacements are GGUF files, which are provided for free on the profile "CITY96".
4. Download both the low noise and high noise models from huggingface from CITY96. Download both the high noise and low noise GGUF models with Q5_K_S. You only need 2, and this GGUF format is the perfect balance of speed and quality.
5. Go to CIVITAI.red, and search for LORAS or LYCORIS specific to the Wan 2.2 i2v 14B i2v. By itself, Wan won't produce adult content, but people figured out how to bypass that, and this generation is an example.
6. You only need to know 1 trick for video generation; Always use the image-to-video workflow, never use the text to video. Use either screenshots from porn, or download an image generator that has NSFW capabilities.
7. In ComfyUI, do not try to generate anything from scratch; there's a large number of parameters that you need to set, and the templates are useful because you can just change the "POSITIVE PROMPT" with your goal. Always copy the CIVITAI generation parameters; there are special keywords for the NSFW LORAS.
8. For video generation, you'll need to add CIVITAI LORAS to your ComfyUI workflow. Place them directly after the "lightx2v" nodes, there should be a low and high noise corresponding to the model.
9. Do not type any text manually into the prompt nodes in your workflow. It is not a chat-box and lots of things can break it, so you'll waste hours trying to get one prompt working. There is a faster solution: Download an abliterated or uncensored local LLM, then start a conversation by telling the model you need help with prompt generation, specifically for ComfyUI and your chosen model. The smaller chatbots are good enough to give you good prompts, and if you tell it the task, it will pretty much always give a valid prompt.
10. Video generation runs are slow on MacOS, and expect a 24FPS 5 second video generation to take close to an hour. if you want complex movement in the video, simply set it to 10FPS and the same number of frames will generate a longer clip.
The last few steps are difficult, but I'd be happy to give anonymous help to anyone wanting to have a private uncensored video generator. I'm also hoping that I can earn a living doing commissions /donations, so if like my generations, then please leave a comment.
1. Buy a Macbook M3 Max 48GB or equivalent hardware
2. Install ComfyUI for MacOS.
3. Inside ComfyUI, there's a "template" button, which can be used to start generating immediately. But do not download any models from ComfyUI itself, because the suggested models are in the FP8 format which is incompatible with Apple Silicon. The MacOS compatible replacements are GGUF files, which are provided for free on the profile "CITY96".
4. Download both the low noise and high noise models from huggingface from CITY96. Download both the high noise and low noise GGUF models with Q5_K_S. You only need 2, and this GGUF format is the perfect balance of speed and quality.
5. Go to CIVITAI.red, and search for LORAS or LYCORIS specific to the Wan 2.2 i2v 14B i2v. By itself, Wan won't produce adult content, but people figured out how to bypass that, and this generation is an example.
6. You only need to know 1 trick for video generation; Always use the image-to-video workflow, never use the text to video. Use either screenshots from porn, or download an image generator that has NSFW capabilities.
7. In ComfyUI, do not try to generate anything from scratch; there's a large number of parameters that you need to set, and the templates are useful because you can just change the "POSITIVE PROMPT" with your goal. Always copy the CIVITAI generation parameters; there are special keywords for the NSFW LORAS.
8. For video generation, you'll need to add CIVITAI LORAS to your ComfyUI workflow. Place them directly after the "lightx2v" nodes, there should be a low and high noise corresponding to the model.
9. Do not type any text manually into the prompt nodes in your workflow. It is not a chat-box and lots of things can break it, so you'll waste hours trying to get one prompt working. There is a faster solution: Download an abliterated or uncensored local LLM, then start a conversation by telling the model you need help with prompt generation, specifically for ComfyUI and your chosen model. The smaller chatbots are good enough to give you good prompts, and if you tell it the task, it will pretty much always give a valid prompt.
10. Video generation runs are slow on MacOS, and expect a 24FPS 5 second video generation to take close to an hour. if you want complex movement in the video, simply set it to 10FPS and the same number of frames will generate a longer clip.
The last few steps are difficult, but I'd be happy to give anonymous help to anyone wanting to have a private uncensored video generator. I'm also hoping that I can earn a living doing commissions /donations, so if like my generations, then please leave a comment.
Uploaded by
patrician
