<div data-youtube-video><iframe src="https://www.youtube.com/embed/YYsbOE1PoB4" width="640" height="360" allowfullscreen="true"></iframe></div>
https://youtu.be/YYsbOE1PoB4 
 
Creator: AI-KSK · Collection: Video 328 / six-workflow pose x outfit matrix episode 
 
Workflow file: <code>01_Qwen21_Pose_UI.json</code> · Cover file: <code>AI姿势库批量换姿势-2.png</code> 
 
<h3>What this workflow does</h3> 
 
Takes one character reference image and re-renders that same character in a batch of poses selected from the built-in pose gallery. The character identity, hairstyle, body proportions, original outfit and rendering style come from the reference image; the pose gallery only supplies body geometry. 
 
<h3>Exact values from this JSON</h3> 
 
- <code>PoseGalleryBatch</code>: manual selection mode, built-in gallery, two sample poses pre-selected (<code>full_sexy_P016_front</code>, <code>full_sexy_P022_front</code>), fit mode <code>等比留边</code>, gallery render 768x1024. 
- <code>RebatchImages</code>: batch size 1. 
- <code>AIKSK_Qwen21ModeLoader</code>: base <code>qwen_image_2.1_int8_convrot.safetensors</code> with turbo variant available. 
- Text encoder <code>qwen3vl_8b_int8_convrot.safetensors</code>, VAE <code>qwen_image_2.1_vae_bf16.safetensors</code>. 
- Output size 768x1152. 
 
<h3>How to use it</h3> 
 
Load your own authorized character image in <code>LoadImage</code>, select the poses you want in the gallery node, then queue. The workflow is the Qwen-Image 2.1 counterpart of workflow 04; the two stacks produce different aesthetics from the same inputs. 
 
<h3>Limits</h3> 
 
The gallery is a third-party community asset. Pose transfer is a generation, not a rig or a guarantee of anatomical correctness; hands, occlusion and extreme poses still need review. 
 
<h3>Dependencies, tutorial and deployment</h3> 
 
ComfyUI with the AI-KSK custom nodes (https://github.com/AI-KSK/KSK-Nodes). The JSON references model files that must exist in your installation; no model weights are bundled with this release. The pose and outfit galleries are third-party community assets with their own licenses. 
 
<a rel="ugc" href="https://www.compshare.cn/images/fFwdv0x1vlhz?referral_code=A7fTQUYY9lCDFJwA6MMvoq">Deploy AI-KSK's pose x outfit matrix image on UCloud CompShare</a>. Referral disclosure: the link includes an AI-KSK referral code and may support the creator; check the provider's billing and environment before deploying. 
 
<a rel="ugc" href="https://youtu.be/YYsbOE1PoB4">Watch the complete six-workflow YouTube tutorial</a> 
 
&lt;div class="data-youtube-video"&gt;&lt;iframe src="https://www.youtube.com/embed/YYsbOE1PoB4" width="560" height="315" frameborder="0" allowfullscreen&gt;&lt;/iframe&gt;&lt;/div&gt;
Chinese version: <a href="https://youtu.be/KtYimd16ApM" rel="ugc">https://youtu.be/KtYimd16ApM</a>

Qwen-Image 2.1 Batch Pose Change

<div data-youtube-video></div>
https://youtu.be/KtYimd16ApM 
 
Creator: AI-KSK · Collection: Video 328 / six-workflow pose x outfit matrix episode 
 
Workflow file: <code>01_Qwen21_Pose_UI.json</code> · Cover file: <code>AI姿势库批量换姿势-2.png</code> 
 
What this workflow does 
 
Takes one character reference image and re-renders that same character in a batch of poses selected from the built-in pose gallery. The character identity, hairstyle, body proportions, original outfit and rendering style come from the reference image; the pose gallery only supplies body geometry. 
 
Exact values from this JSON 
 
- <code>PoseGalleryBatch</code>: manual selection mode, built-in gallery, two sample poses pre-selected (<code>full_sexy_P016_front</code>, <code>full_sexy_P022_front</code>), fit mode <code>等比留边</code>, gallery render 768x1024. 
- <code>RebatchImages</code>: batch size 1. 
- <code>AIKSK_Qwen21ModeLoader</code>: base <code>qwen_image_2.1_int8_convrot.safetensors</code> with turbo variant available. 
- Text encoder <code>qwen3vl_8b_int8_convrot.safetensors</code>, VAE <code>qwen_image_2.1_vae_bf16.safetensors</code>. 
- Output size 768x1152. 
 
How to use it 
 
Load your own authorized character image in <code>LoadImage</code>, select the poses you want in the gallery node, then queue. The workflow is the Qwen-Image 2.1 counterpart of workflow 04; the two stacks produce different aesthetics from the same inputs. 
 
Limits 
 
The gallery is a third-party community asset. Pose transfer is a generation, not a rig or a guarantee of anatomical correctness; hands, occlusion and extreme poses still need review. 
 
Dependencies, tutorial and deployment 
 
ComfyUI with the AI-KSK custom nodes (https://github.com/AI-KSK/KSK-Nodes). The JSON references model files that must exist in your installation; no model weights are bundled with this release. The pose and outfit galleries are third-party community assets with their own licenses. 
 
<a href="https://www.compshare.cn/images/fFwdv0x1vlhz?referral_code=A7fTQUYY9lCDFJwA6MMvoq" rel="ugc">Deploy AI-KSK's pose x outfit matrix image on UCloud CompShare</a>. Referral disclosure: the link includes an AI-KSK referral code and may support the creator; check the provider's billing and environment before deploying. 
 
<a href="https://youtu.be/KtYimd16ApM" rel="ugc">Watch the complete six-workflow YouTube tutorial</a> 
 
&lt;div class="data-youtube-video"&gt;&lt;iframe src="https://www.youtube.com/embed/KtYimd16ApM" width="560" height="315" frameborder="0" allowfullscreen&gt;&lt;/iframe&gt;&lt;/div&gt;

Files with SHA256: d1180154017cba76fac18a040a29d18b7b6801a6515e9f2741de4c9cc7a74cb0