I have been trying to achieve this effect for a long time with existing models, now I finally managed to train my first video Lora.It requires a fairly specific image in order for teh effect to be realistic, an image of a womans vagina with good detail from a lower angle, and the slit needs to be somewhat dilated.I trained it on LTX 2.3 Base model, and it does not understand the vaginal anatomy well, so I recommend using another lora for that purpouse (Dr3aml4y or Sexgod nudity).Version 2:Trained a second run with a few more clips and on 10Eros which helps the anatomy a lot.Same triggerword but can now use a bit more temporal control in the prompt like:<pre><code>pre orgasm phase, orgasm buildup phase, orgasm peak phase</code></pre>Some control over the degree of moaning:<pre><code>subtle moaning, slight moaning, intensified moaning, intense moaning, hyperventiaating</code></pre>I also tried to achieve different types of motion effects with:<pre><code>vagina pulsations, vagina contractions</code></pre>It seems to have some effect, but vagina pulsations being the dominant motionI also experimented with:<pre><code>wet vagina, glistening vagina, creamy vagina, cream bubbles</code></pre>..with varying results.T2V works, but anatomy is a bit off, need some more variety to the dataset.

vaginal orgasm - convultions

Trained a second run with a few more clips and on 10Eros which helps the anatomy a lot.Same triggerword but can now use a bit more temporal control in the prompt like:<pre><code>pre orgasm phase, orgasm buildup phase, orgasm peak phase</code></pre>Some control over the degree of moaning:<pre><code>subtle moaning, slight moaning, intensified moaning, intense moaning, hyperventiaating</code></pre>I also tried to achieve different types of motion effects with:<pre><code>vagina pulsations, vagina contractions</code></pre>It seems to have some effect, but vagina pulsations being the dominant motionI also experimented with:<pre><code>wet vagina, glistening vagina, creamy vagina, cream bubbles</code></pre>..with varying results.T2V works, but anatomy is a bit off, need some more variety to the dataset.

Files with SHA256: aafb405896cc6d04df10436aebff058bf3d96b9e4d03ccc5868dd270e97b2046