This model accurately captures Lae'zel from Baldur's Gate 3, trained on 35 carefully curated, long-captioned images including screenshots and renders for versatility.<h3 id="launching-v2.0-lora-26bkp092c">Launching v2.0 - LoRa</h3>Now available as a Rank 8 LoRa trained with an improved dataset and hyperparameters for better likeness and prompt adherence over the v1.0 LoKr version but with similar or slightly worse compatibility with other LoRas. The v2.0 LoKr version is in development and should improve even further.⚠️ Important: If using the v1.0 LoKr model, please read the "ComfyUI Setup" section below before use.<h3 id="usage-mpk8zbei2">Usage</h3>Trigger Word: <code>l43z3l elf</code>Important: <ol><li>Do not include "woman" for likeness, like <code>l43z3l elf woman</code></li><li>Include "she" or "her" in your prompt to ensure female generation, otherwise the model may generate a male Lae'zel spinoff.</li></ol><h3 id="usage-tips-j3ttun8y8">Usage Tips</h3><ul><li>Start with weight 1.0 and adjust as needed</li><li>The more mundane or realistic the prompt is, the more likely you will need to increase the strength to 1.1 or even 1.2.</li><li>For baldur's gate like prompts (e.g. mentioning armor) you might need to decrease the strength to 0.9 or even 0.8 in order to avoid a too strong 3d render like output</li><li>When combining with other models, first consider increasing this model's strength up to 1.4 if the likeness isn't quite there yet. If still not quite there, try then lowering the other loras strength.</li><li>This is because the model has a weaker interaction vs most other loras, depending on how realistic and/or overfitted they are.</li></ul><h3 id="comfyui-setup-for-lokr-model-(v1.0-only)-4u1b747b2">ComfyUI Setup for LoKr Model (v1.0 Only)</h3>Important: The following setup instructions are only required for the v1.0 LoKr model. The v2.0 LoRa model works out of the box with ComfyUI.For some reason, LoKr models for Wan 2.1 are not yet supported in ComfyUI and you need to do a small code change. This change should not impact any other LoRas or LoKr models you use for Wan or other base models.Installation Steps:<ol><li>Navigate to your ComfyUI folder: <code>comfy/lora.py</code></li><li>Find the <code>model_lora_keys_unet</code> function</li><li>Replace this section:</li></ol><pre><code>for k in sdk:
 if k.startswith("diffusion_model."):
 if k.endswith(".weight"):
 key_lora = k[len("diffusion_model."):-len(".weight")].replace(".", "_")
 key_map["lora_unet_{}".format(key_lora)] = k
 key_map["{}".format(k[:-len(".weight")])] = k #generic lora format without any weird key names
 else:
 key_map["{}".format(k)] = k #generic lora format for not .weight without any weird key names
</code></pre>With this:<pre><code>for k in sdk:
 if k.startswith("diffusion_model."):
 if k.endswith(".weight"):
 key_lora = k[len("diffusion_model."):-len(".weight")].replace("attn.o", "attn.to.out.0").replace("ffn.0", "ffn_net_0_proj").replace("ffn.2", "ffn_net_2").replace(".", "_")
 key_map["lora_unet_{}".format(key_lora)] = k
 key_map["{}".format(k[:-len(".weight")])] = k #generic lora format without any weird key names
 key_map["lycoris_{}".format(key_lora)] = k #SimpleTuner lycoris format
 else:
 key_map["{}".format(k)] = k #generic lora format for not .weight without any weird key names
</code></pre>Note: This modification should not impact using other LoRa or LoKr models for any other base models.Other UIs: Setup instructions currently only available for ComfyUI. Please share your experiences with other interfaces in the comments.<h3 id="feedback-zif0yvhet">Feedback</h3>Share your results and report any issues in the comments.❤️ Last but not least, if you enjoy this model, please give it a like and share your generations in the comments! Your support helps other creators discover it.

Lae'zel (Baldur's Gate 3) - WAN 2.1 14B T2V

v2.0 - LoRaChanges:<ul><li>Adding a Rank 8 LoRa model option for compatibility as LoKr is not widely supported out of the box in the UIs</li><li>Removed lower quality images and added new and better images to the dataset</li><li>Added EMA to model training</li><li>Used a lower starting lr and longer warmup</li></ul>Improvements over v1.0 - LoKr:<ul><li>Better likeness</li><li>Better prompt adherence</li><li>Should still be compatible with other LoRas</li></ul>

l43z3l_v2_lora.safetensors

v2.0 - LoRa

v1.0 - LoKr

<a target="_blank" rel="noopener noreferrer nofollow" href="https://civitai.com/models/1713597/laezel-baldurs-gate-3-lokr-model?modelVersionId=1939159">https://civitai.com/models/1713597/laezel-baldurs-gate-3-lokr-model?modelVersionId=1939159</a>

Lae'zel  Baldur's Gate 3

l43z3l elf sits comfortably at a coffee shop table. She wears a cozy oversized sweater. The camera captures her from a natural eye-level angle. Warm golden hour light streams through the window beside her, creating a soft glow on her skin and highlighting the steam rising from her cup. Her phone lies forgotten on the table next to a half-eaten pastry, and her relaxed posture shows she's completely absorbed in the moment as she drinks the coffee in her cup. The background bustles with other café patrons, but the shallow depth of field keeps focus on her calm expression. The camera slowly rotates around her.

Files with SHA256: 429d87b9b51f79023c1408fca378c2aeeb3705c92e3e357e9d2e5ff04134d7d3