<h3 id="slow-progressive-depth-and-stable-deep-holds."><strong>slow progressive depth</strong> and <strong>stable deep holds</strong>.</h3><p>training large video LoRAs like H3 on cloud GPUs is an absolute money burner 💸. High-tier cloud compute costs add up so quickly with every experiment and epoch iteration.</p><p>Every single bit of support helps cover those heavy cloud GPU rental fees, and it directly serves as the motivation and fuel to keep this <strong>Bunny</strong> training and bringing more high-quality models to you all! 🐰⚡</p><p><span style="background-color:rgba(34, 139, 230, 0.15);color:rgb(230, 73, 128);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">☕ If my works bring you some fun and you would like to support my future creations:</span></p><p><span style="background-color:rgba(34, 139, 230, 0.15);color:rgb(230, 73, 128);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"> Ko-fi:</span><span style="background-color:rgba(34, 139, 230, 0.15);color:rgb(254, 254, 254);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"> </span><a target="_blank" rel="ugc" href="https://ko-fi.com/fourbunny"><u>https://ko-fi.com/fourbunny</u></a></p><hr /><p>Instead of emphasizing speed, this V1 is designed for longer controlled movements, readable depth changes, and stopping at the deepest position without immediately falling back into repetitive motion.</p><p>It also strengthens the small secondary responses around the <strong>cheeks, jaw, under-jaw area, front of the neck, and throat</strong>, together with <strong>synchronized wet contact sounds, gagging, breathing changes, and recovery audio</strong>.</p><h3 id="main-strengths">Main strengths</h3><ul><li><p>Slow progressive deep movement</p></li><li><p>Long continuous insertion instead of short repetitive strokes</p></li><li><p>Stable deep-position holds</p></li><li><p>Small cheek, jaw, neck, and throat responses during the hold</p></li><li><p>Controlled withdrawal</p></li><li><p><strong>Synchronized breathing, gagging, wet contact, and recovery sounds</strong></p></li><li><p>Can combine <code>oralmove1</code> and <code>oraldetail1</code> depending on the desired behavior</p></li></ul><h3 id="suggested-lora-weight">Suggested LoRA weight</h3><p>Start around <strong>0.8</strong> and adjust according to the character, reference image, and desired intensity.</p><hr /><h2 id="a-generic-template-for-simple-copy-and-paste"><strong>A generic template for simple copy-and-paste</strong></h2><h2 id="example-1-deep-hold">Example 1 — Deep Hold</h2><pre><code>integrated_multimodal_description:
oraldetail1. [Shot 1] She begins from established oral contact and advances slowly through one long continuous movement into a deeper position. The insertion depth increases steadily. Her jaw angle changes progressively, while her cheek, the area beneath the jaw, and the front of the neck show small synchronized changes as the depth increases. At the deepest position, the forward-and-back movement stops and she remains there for approximately two seconds. Her head position changes very little while small movements continue through the cheek, beneath the jaw, along the front of the neck, and around the throat with her breathing. She then performs one slow controlled partial withdrawal, with the insertion depth decreasing gradually as the cheek, jaw, and neck progressively relax.

overall_soundscape:
Heavy muffled gagging and strained choking sounds continue while she is held completely still at maximum depth. Her breathing is cut off, only short desperate nasal attempts escape. Wet saliva dripping sounds. No dialogue.

non_diegetic_music:
None.</code></pre><h2 id="example-2-slow-deep-motion">Example 2 — Slow Deep Motion</h2><pre><code>integrated_multimodal_description:
oralmove1, oraldetail1. [Shot 1] She begins from established oral contact and moves gradually into a deeper position through one long continuous movement. The insertion depth increases steadily and remains clearly readable throughout the movement. Her jaw angle changes progressively with the increasing depth, while her cheek, the area beneath the jaw, and the front of the neck respond with small synchronized changes. She then performs one slow controlled withdrawal, with the insertion depth decreasing continuously as the jaw, cheek, and neck gradually return toward their previous state.

overall_soundscape:
Loud wet gagging and deep choking sounds rise on every full insertion, then lighten into wet suction and her short strained whimpers on each withdrawal. Continuous messy saliva sounds. No dialogue.

non_diegetic_music:
None.</code></pre><hr /><h2 id="example-3-entry-deep-hold-struggle-full-withdrawal">Example 3 — Entry → Deep Hold → Struggle → Full Withdrawal</h2><pre><code>
integrated_multimodal_description:
oralmove1, oraldetail1,[Shot 1] The man grips her head and steadily pushes the cock deeper until her lips are pressed tightly against his balls. He then holds her completely still at this maximum depth with no thrusting. While held there she begins to struggle from lack of air — her body tenses, her hands weakly move, and her expression tightens as she tries to breathe. After several seconds he suddenly pulls the cock all the way out. Long thick strings of saliva stretch between her lips and the tip before breaking. The moment the cock leaves she gasps in a huge desperate breath, mouth wide open, chest heaving hard.

overall_soundscape:
Wet sounds transition into heavy muffled gagging and strained choking while she is held balls-deep. Her struggle produces short desperate nasal sounds and suppressed whimpers. On the sudden extraction there is a thick wet pop followed by the stretch of saliva strings and her sharp, full-lung gasp for air. No dialogue.

non_diegetic_music:
None.</code></pre><h3 id="notes">Notes</h3><p>This V1 is mainly tuned for <strong>Slow Deep + Hold</strong> behavior rather than high-speed repetitive motion.</p><p>For the cleanest results, describe the movement as a continuous physical sequence rather than stacking abstract intensity words. Depth changes, stopping, holding, withdrawal, breathing, and local facial/neck responses should be written as visible or audible events.</p><h3 id="compatibility-note">Compatibility Note</h3><p>This version mainly focuses on <strong>Slow Deep + Hold</strong> behavior, depth progression, and local motion response. </p><p>For cleaner results, I currently recommend pairing it with a dedicated<strong> penis LoRA</strong>, or using reference images where the male <strong>penis</strong> is already fully visible and structurally clear.</p><p>If you rely entirely on the H3 base model to generate it from scratch, the result can still look strange in this version, especially in terms of shape consistency, structural continuity, and geometry during insertion or withdrawal.</p><p>This is a known limitation of V1, and I plan to address it in a future update by adding a small set of geometry-focused training samples while keeping the current <strong>Slow Deep + Hold</strong> strengths intact.</p><p></p><p><span style="background-color:rgb(26, 27, 30);color:rgb(250, 176, 5);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"><strong>This LoRA is provided as a motion-generation tool. Users are responsible for how they use their reference images and generated outputs.</strong></span></p>

Files with SHA256: e85ab40e9b7267f92cab8f9e066e7b448791290146518c0e8414a87168218a4f