<h3 id="ga1n3rb0t-ntrq9vud9">ga1n3rb0t</h3>ga1n3rb0t is a LORA created for the Wan2.1 14B T2V + Hunyuan T2V models. Designed to produce realistic BBW content both SFW and NSFW.Use the trigger word: ga1n3rb0t at the beginning of your prompts.Notable keywords/phrases included in dataset: jiggling, rubbing, caressing, massaging, lifting.<h3 id="version-info:-rebwwnpz0">Version Info:</h3><ul><li>Wan2.1 14B T2V v1.0 (Recommended): Currently the most stable release for Wan2.1 14B T2V (will work fine with GGUF versions). IMO this is the best version the Lora, strongly recommended over Hunyuan versions.</li><li>Hunyuan T2V v2.0 (Recommended): Currently the most stable release for Hunyuan T2V (will work fine with GGUF versions). Better movement prompt adherence, more reliable results.</li><li>Hunyuan T2V v2.1 (Experimental/Not Recommended): Tendency to be unstable, but can be interesting. Main highlight is better clothing prompting (will work fine with GGUF versions). Read version notes for more info. </li></ul>

ga1n3rb0t

Additional videos and images were added to the existing dataset from version 2.0, doubling the size of the dataset. The result is version 2.1 (Experimental) which is moderately successful. Version 2.0 is more stable than 2.1, with better prompt adherence and more reliable outcomes. I would say version 2.0 is better overall. Key differences with v2.1: easier and more diverse clothing prompting, significantly larger women generated without necessarily prompting for it, better overall physics. The shortcomings of v2.1 are: weaker prompt adherence when it comes to specific movements, less predictable movement, less flexibility overall. Nonetheless, some people might find it fun/interesting to play around with - it is certainly different and has its quirks, it isn't perfect. I would recommend a lora strength of 0.80.

Files with SHA256: e3114b945d7074aba609359b075910fe1ce0d78a791d6223affc2053173c8bf8