<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);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">Changes the pose and framing of a character displayed in Image 1 to instead match the one provided in Image 2.<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);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">It will attempt to use what is already provided in the original image, but if the original lacks a required element, it will insert what is required. For example, if the target pose depicts someone drinking a beverage, they will be holding a beverage that was on the table of the original image, but if the original image didn't have one, then it will create a new beverage for them to hold.<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);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">For best results, try to have comparable framing, camera angle, and number of subjects in both images.<pre><code>change the actions and poses in Image 1 to match those in Image 2</code></pre>There is sometimes some subject bleed in Qwen and Klein 9B, so I will probably revisit those to make a V2. Klein 4B is significantly more stable.Model Comparisons:<ul><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2380153/copy-pose-qwen-and-klein">Copy Pose</a> will preserve the subject and background in Control 1, while bringing the framing/pose of Control 2</li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2508392">Replace Subject</a> will preserve the framing/pose and background in Control 1, while bringing the subject of Control 2</li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2517219">Replace Scene</a> will preserve the subject and pose in Control 1, while bringing the background scene of Control 2</li></ul>California AB 2013 Training Data Disclosure<ul><li>This LoRA was fine-tuned using visual data consisting entirely of synthetic still images. The training data may include copyrighted material owned by third parties. No training data was licensed or purchased. This LoRA is provided for non-commercial use only under the terms of its distribution.</li><li>The dataset consists of 538 image sets (1614 images total). Each set was created by generating an image, using ControlNet to create a matching duplicate (for use as Control 2 and Target), then using Qwen Edit 2509 to change both images to be different (for use as Control 1), resulting in four images that can be used as two sets (two images are used in both sets). Low quality sets were then culled prior to training. Dataset was created in 2026.</li><li>Image data was processed through standard resizing, cropping, normalization, and labeling steps. Synthetic images were included as part of the training dataset.</li><li>This model is intended for non-commercial, experimental, and educational use. Generated outputs may reflect copyrighted visual styles or themes associated with the underlying training data. Users are responsible for ensuring compliance with applicable copyright law, other intellectual property laws, and all other applicable laws.</li></ul>

Files with SHA256: b2148be5556e620d25f41f744430d200d1b37e7163f04f90e75e0e6ae857d0c1