<h2 id="inference">Inference</h2>For inference I used ComfyUI.The strength of the LoRA can differ from prompt to prompt. Usually 0.9 works but in some cases it can be too strong (lower it to 0.8-0.85) or too weak (increase it to 1.0).<ul><li>Trigger words: Nfj1nx, blue hair</li><li>Strength: 0.8-0.9</li></ul>You can find example prompts in the video examples!<h2 id="trainig-details">Trainig details</h2><ul><li><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">dataset size: 127 x 960x544 (with various frame lenghts)</li><li><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">steps: 29000</li><li>optimizer_type: adamw</li><li>scheduler_type: linear</li><li>learning_rate: 0.0001</li><li>first_frame_conditioning_p: 0.5</li><li>alpha: 32</li><li>dropout: 0.0</li><li>rank: 32</li><li>target_modules:<ul><li>to_k</li><li>to_q</li><li>to_v</li><li>to_out.0</li></ul></li></ul>Important Notes: <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">This LoRA is created as part of a fan project for research purposes only and is not intended for commercial use. It is based on the movies, which are protected by copyright. Users utilize the model at their own risk. Users are obligated to comply with copyright laws and applicable regulations. The model has been developed for non-commercial purposes, and it is not my intention to infringe on any copyright. I assume no responsibility for any damages or legal consequences arising from the use of the model.

LTX-2 19b Arcane Jinx LoRA

874132-ltx2-Nfj1nx_29000.safetensors

Arcane_Jinx

v1.0

LTXV Arcane Jinx character LoRA v1 for the 13B modelInferencePlease note that the distilled model might not work properly with this lora. Based on my tests it might have a weaker effect or you might notice some visual artifacts.This LoRA is quite overtrained and in the captioning I intentionally avoided using text that describes clothing, so it would generate the character's original outfit/look.If you still want Jinx to wear different outfit from the usual, then I recommend using a 0.6 LoRA strength (or even lower) for the first passand then increase the LoRA strength to as high as 0.9 for the upscale pass.For inference I used ComfyUI.Trigger words: csetiarcane Nfj1nx blue hairStrength: 0.6-0.9Trainig details:Trained only on videos.- dataset: 40 videos 768x480x25- steps: 12000- LR: 2e-4- optimizer: adamw- scheduler: contant- rank: 128- batch size: 1- gradient accumulation steps: 1For training I used the official <a href="https://github.com/Lightricks/LTX-Video-Trainer" rel="ugc noopener noreferrer nofollow" target="_blank">LTX-Video Trainer</a> repo.Important Notes:This LoRA is created as part of a fan project for research purposes only and is not intended for commercial use. It is based on the TV series called Arcane which are protected by copyright. Users utilize the model at their own risk. Users are obligated to comply with copyright laws and applicable regulations. The model has been developed for non-commercial purposes, and it is not my intention to infringe on any copyright. I assume no responsibility for any damages or legal consequences arising from the use of the model.## AcknowledgementSpecial thanks to <a href="https://github.com/Lightricks/LTX-Video" rel="ugc noopener noreferrer nofollow" target="_blank">Lightricks Team</a> for making this available for the community for free!

Files with SHA256: 5a178aab0047622b69b4a9b8eddc509afeb050a81c1418c528e126872262eadd