For the REF2VA model.Don't expect great details without a reference image, training tool authors are still working on it.Somehow also help the base model who struggles making the camera look directly up.<pre><code>facesitting on the viewer</code></pre><pre><code>looking upward</code></pre><pre><code>above the viewer</code></pre><pre><code>She is thrusting her hips back and forth</code></pre><pre><code>The camera is placed between her legs</code></pre><pre><code>The camera capture a very wide angle with a fisheye lens.</code></pre><pre><code>Worm eye view</code></pre>

Files with SHA256: cb040b6068d2f063eb101ae3f31eba6a1dd826515c80385b333e180887274c04