https://youtu.be/aCEuir7zy-w 
 
This workflow is the online batch restyle version of the Krea 2 RAW + Turbo image refinement setup. It is built for creators who prefer to run a batch image cleanup pipeline through RunningHub instead of maintaining every dependency locally. The workflow focuses on taking multiple source images and improving clarity, edge quality, material texture, and overall polish while keeping the original subject count, pose, composition, depth, lighting, and color logic. 
 
The inspected graph uses a two-model Krea 2 route. `krea2_raw_bf16_0630.safetensors` handles the first high-noise shaping stage, while `krea2_turbo_fp8.safetensors` handles the lower-noise convergence and refinement stages. The text encoder is `qwen3vl_4b_fp8_scaled.safetensors`, and the VAE is `qwen_image_HDR_vae_fp32_comfy.safetensors`. The active structural control route uses `depth-control-lora.safetensors` in three stages with strengths 1.00, 0.85, and 0.65, which helps the online batch process stay close to the source layout. 
 
This version keeps the same RAW/Turbo logic as the local refinement family but is packaged for online batch use. The active chain includes a 36-step RAW/Turbo scheduler, RAW high-noise shaping, Turbo convergence at source resolution, latent upscale stages targeting 1024 x 1024, a 10-step Turbo upscale refinement stage, several preview outputs, and an active SaveImage path. A legacy final KSampler and a 2048 x 2048 scale node are bypassed or disconnected, so this listing does not claim them as active output stages. 
 
Main features: 
 
- Online batch Krea 2 restyle workflow 
- RunningHub-friendly batch refinement setup 
- Krea 2 Raw BF16 high-noise shaping route 
- Krea 2 Turbo FP8 low-noise refinement route 
- Qwen3VL 4B FP8 Krea2 text encoder 
- Qwen Image HDR VAE route 
- Three-stage Krea2 depth control 
- Depth ControlNet LoRA strengths 1.00, 0.85, and 0.65 
- 1024 x 1024 latent refinement target 
- Active 4x-UltraSharp upscale route 
- Multiple preview outputs before final save 
 
Suggested workflow: 
 
Use this version when you want to test or process a batch online with the same preservation-first restyle logic. Upload source images with related quality goals, keep the prompt consistent, and check the preview outputs before scaling the batch. If the batch becomes too stylized, reduce descriptive style terms and emphasize preservation. If the batch still looks under-refined, make the restoration target more explicit and let the RAW/Turbo stages add detail gradually. 
 
Related resources: 
 
Krea 2 Model Collection (Quark): https://pan.quark.cn/s/07bdc81784ce 
Krea 2 Raw: https://huggingface.co/krea/Krea-2-Raw 
Krea 2 Turbo: https://huggingface.co/krea/Krea-2-Turbo 
Krea 2 Depth ControlNet LoRA: https://huggingface.co/Patil/Krea-2-depth-controlnet 
 
RunningHub Workflow 
 
Try the workflow online right now - no installation required. 
Workflow: https://www.runninghub.ai/post/2080681462248071170?inviteCode=rh-v1111 
 
If the results meet your expectations, you can later deploy it locally for customization. 
 
Fan Benefits: Register to get 1000 points + daily login 100 points - enjoy 4090 performance and 48 GB super power. 
 
Bilibili Updates (Mainland China &amp; Asia-Pacific) 
 
If you are in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 
Bilibili Video: https://www.bilibili.com/video/BV1GBgi6JEwL/ 
 
Support Me on Ko-fi 
 
If you find my content helpful and want to support future creations, you can buy me a coffee. 
Every bit of support helps me keep creating. 
Ko-fi: https://ko-fi.com/aiksk 
 
Business Contact 
 
For collaboration or inquiries, please contact aiksk95 on WeChat. 
 
打开下方链接即可在线体验，无需安装。 
工作流：https://www.runninghub.ai/post/2080681462248071170?inviteCode=rh-v1111 
 
如果觉得效果理想，也可以在本地进行自定义部署。 
 
粉丝福利：注册即送 1000 积分，每日登录 100 积分，可体验 4090 / 48G 级别算力。 
 
Bilibili 更新（中国大陆及亚太地区） 
 
如果你在中国大陆或亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。 
B站视频：https://www.bilibili.com/video/BV1GBgi6JEwL/ 
 
本期 Krea 2 相关模型资源： 
Quark: https://pan.quark.cn/s/07bdc81784ce 
Krea 2 Raw: https://huggingface.co/krea/Krea-2-Raw 
Krea 2 Turbo: https://huggingface.co/krea/Krea-2-Turbo 
Krea 2 Depth ControlNet LoRA: https://huggingface.co/Patil/Krea-2-depth-controlnet 
 
这些资源主要面向本地用户，方便进行创作与学习。

Krea2 RAW-Turbo Online Batch Restyle Workflow

krea2RAWTurboOnline_v10.json

v1.0

<div data-youtube-video="https://youtu.be/aCEuir7zy-w"><iframe src="https://www.youtube.com/embed/aCEuir7zy-w" width="853" height="480" allowfullscreen></iframe></div> 
RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2080681462248071170?inviteCode=rh-v1111
<h2>Krea 2 Cleanup 2.6 - RunningHub Online Batch (Three-Stage Depth Control + Presets)</h2>
Purpose: the RunningHub-hosted batch variant of the Krea 2 cleanup graph. Same three depth-conditioned stages and the same preset dial as the local version, but the batch I/O runs through RunningHub nodes: upload a ZIP, extract images, batch them, and bring LLM captions back through the RunningHub LLM node.
<h2>Graph (as saved)</h2>
104 nodes / 169 links / 57 groups.
<ul>
<li>Depth chain: three <code>Krea2ControlImageEncode</code> stages (high / mid / final size) each feeding <code>Krea2ControlApply</code> and a <code>Krea2ControlLoRALoader</code> on <code>depth-control-lora.safetensors</code> at 1.00 / 0.85 / 0.65 (preset-driven).</li>
<li>The three <code>LoraLoaderModelOnly</code> slots on the same LoRA file are saved at strength 0: empty/disabled slots, not active LoRAs.</li>
<li>Models: <code>krea2_raw_bf16_0630.safetensors</code> for the high-noise stage, <code>krea2_turbo_fp8.safetensors</code> for mid/low. CLIP <code>qwen3vl_4b_fp8_scaled.safetensors</code> (type <code>krea2</code>), VAE <code>qwen_image_HDR_vae_fp32_comfy.safetensors</code>.</li>
<li>RunningHub-specific nodes: <code>RHUploadZip</code>, <code>RHBatchImages:</code>, <code>RH_ExtractImage</code>, <code>RH_LLMAPI_Pro_Node</code> - the online batch path and the automatic caption/JSON path.</li>
<li>Samplers: three <code>SamplerCustomAdvanced</code> with one shared <code>SamplerEulerAncestral</code> (1, 1); shared RAW/Turbo schedule 36 steps denoise 0.92; Turbo upscale refine 10 steps denoise 0.35; active intermediate KSampler 10 steps, CFG 1, euler_ancestral, simple, denoise 0.40; legacy KSamplers saved bypassed.</li>
<li>Depth maps via DA3 nodes; upscale via 4x-UltraSharp + <code>ImageUpscaleWithModel</code>; SaveImage prefix <code>Krea2_线上批量洗图_2.4_RAWTurbo</code>.</li>
<li>Preset dial: 0 = tightly controlled, 1 = balanced (default), 2 = diverging high-noise repaint; switches denoise, Split checkpoints, RAW CFG and depth-LoRA strengths together.</li>
</ul>
<h2>How to use</h2>
Upload a ZIP of images, select the preset, and queue. The LLM node expects a RunningHub API/LLM endpoint; if you run this outside RunningHub, that branch and the RH I/O nodes will not resolve, so either run it on the platform or replace them with local loaders.
<h2>Limits and not-verified items</h2>
No inference was run for this listing: no speed, VRAM, batch throughput, fidelity or upscale-quality claim is made. RunningHub node behaviour and availability are platform-side; this file is a saved snapshot, not an installation guarantee. Weights, LoRA, depth model and input media are not bundled. The cover image is a promotional still, not a frame produced by this graph.
Tutorial: <a href="https://www.youtube.com/watch?v=aCEuir7zy-w" rel="ugc">YouTube</a> · <a href="https://www.bilibili.com/video/BV1GBgi6JEwL" rel="ugc">Bilibili</a>. Workflows supplied through AI-KSK; verify workflow, node, model, LoRA, input-image and cover rights separately. No blanket commercial or redistribution license is asserted.

Krea 2 Cleanup 2.6 - RunningHub Online Batch (Three-Stage Depth Control + Presets)

<div data-youtube-video="https://youtu.be/aCEuir7zy-w"></div> 
RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2080681462248071170?inviteCode=rh-v1111
Krea 2 Cleanup 2.6 - RunningHub Online Batch (Three-Stage Depth Control + Presets)
Purpose: the RunningHub-hosted batch variant of the Krea 2 cleanup graph. Same three depth-conditioned stages and the same preset dial as the local version, but the batch I/O runs through RunningHub nodes: upload a ZIP, extract images, batch them, and bring LLM captions back through the RunningHub LLM node.
Graph (as saved)
104 nodes / 169 links / 57 groups.
<ul>
<li>Depth chain: three <code>Krea2ControlImageEncode</code> stages (high / mid / final size) each feeding <code>Krea2ControlApply</code> and a <code>Krea2ControlLoRALoader</code> on <code>depth-control-lora.safetensors</code> at 1.00 / 0.85 / 0.65 (preset-driven).</li>
<li>The three <code>LoraLoaderModelOnly</code> slots on the same LoRA file are saved at strength 0: empty/disabled slots, not active LoRAs.</li>
<li>Models: <code>krea2_raw_bf16_0630.safetensors</code> for the high-noise stage, <code>krea2_turbo_fp8.safetensors</code> for mid/low. CLIP <code>qwen3vl_4b_fp8_scaled.safetensors</code> (type <code>krea2</code>), VAE <code>qwen_image_HDR_vae_fp32_comfy.safetensors</code>.</li>
<li>RunningHub-specific nodes: <code>RHUploadZip</code>, <code>RHBatchImages:</code>, <code>RH_ExtractImage</code>, <code>RH_LLMAPI_Pro_Node</code> - the online batch path and the automatic caption/JSON path.</li>
<li>Samplers: three <code>SamplerCustomAdvanced</code> with one shared <code>SamplerEulerAncestral</code> (1, 1); shared RAW/Turbo schedule 36 steps denoise 0.92; Turbo upscale refine 10 steps denoise 0.35; active intermediate KSampler 10 steps, CFG 1, euler_ancestral, simple, denoise 0.40; legacy KSamplers saved bypassed.</li>
<li>Depth maps via DA3 nodes; upscale via 4x-UltraSharp + <code>ImageUpscaleWithModel</code>; SaveImage prefix <code>Krea2_线上批量洗图_2.4_RAWTurbo</code>.</li>
<li>Preset dial: 0 = tightly controlled, 1 = balanced (default), 2 = diverging high-noise repaint; switches denoise, Split checkpoints, RAW CFG and depth-LoRA strengths together.</li>
</ul>
How to use
Upload a ZIP of images, select the preset, and queue. The LLM node expects a RunningHub API/LLM endpoint; if you run this outside RunningHub, that branch and the RH I/O nodes will not resolve, so either run it on the platform or replace them with local loaders.
Limits and not-verified items
No inference was run for this listing: no speed, VRAM, batch throughput, fidelity or upscale-quality claim is made. RunningHub node behaviour and availability are platform-side; this file is a saved snapshot, not an installation guarantee. Weights, LoRA, depth model and input media are not bundled. The cover image is a promotional still, not a frame produced by this graph.
Tutorial: <a href="https://www.youtube.com/watch?v=aCEuir7zy-w" rel="ugc">YouTube</a> · <a href="https://www.bilibili.com/video/BV1GBgi6JEwL" rel="ugc">Bilibili</a>. Workflows supplied through AI-KSK; verify workflow, node, model, LoRA, input-image and cover rights separately. No blanket commercial or redistribution license is asserted.

Files with SHA256: 9982b16ce9c8b2e30621bb3c244d05e011da8691063544f1a67cb252b103157d