Anima Base 1.0 + LLite ControlNet-Style Generation Workflow

Watch the full video first if you want to understand the workflow logic quickly. The video shows how the Anima Base + LLite control pipeline works, why it is useful, and how to launch the workflow online without dealing with a complicated local setup.

This ComfyUI workflow is designed for Anima Base 1.0 controlled image generation using LLite as a lightweight ControlNet-style guidance module. The goal is to give Anima Base stronger composition control while keeping the workflow simple, modular, and easy to reuse in larger production graphs.

The workflow is built around anima_baseV10.safetensors, qwen_3_06b_base.safetensors, and qwen_image_vae.safetensors. It also includes optional enhancement modules such as NAGuidance and anima-highres-aesthetic-boost.safetensors. These optional modules can help improve prompt response, composition stability, and final visual polish, but they can also be bypassed when a cleaner baseline is needed.

The core control section uses a reference image as the control input. The image is resized to match the generated canvas, then processed with DepthAnythingPreprocessor. The resulting depth-style control image is passed into AnimaLLLiteApply, which injects LLite guidance into the Anima model. This gives the generation stronger spatial structure, pose direction, scene layout, and depth consistency compared with pure text-to-image generation.

This workflow is useful when prompt-only generation is not stable enough. With LLite control, creators can keep the freedom of Anima Base while guiding the output through a visual structure. It is suitable for anime key visuals, fantasy scenes, character posters, pose control, depth-guided illustration, scene layout transfer, and reference-composition-based image generation.

The workflow uses FluxResolutionNode to calculate canvas size, then creates an empty latent based on the selected aspect ratio. The prompt section includes separate positive and negative prompt inputs, making it easy to change the visual direction while keeping the same control structure. The positive prompt defines the subject, style, lighting, atmosphere, and final image concept. The negative prompt suppresses low-quality output, blur, artifacts, and unwanted visual problems.

The generation stage uses a two-pass structure. The first sampling stage focuses on building the controlled base image under LLite guidance. After that, LatentUpscaleBy enlarges the latent by 1.5x. The second sampling stage inherits the first-stage composition and refines the image at a higher level of detail. This structure gives better final quality than a single-pass generation route while keeping the workflow easier to understand than a full repair-heavy graph.

The workflow intentionally removes the later face and hand repair chain. This makes it cleaner, faster, and better suited as a reusable Anima + LLite control module. The final result is decoded through the VAE and exported with native PreviewImage and SaveImage nodes, improving platform compatibility.

Main features:

• Anima Base 1.0 controlled generation workflow

• LLite used as a ControlNet-style guidance module

• DepthAnything control preprocessing

• AnimaLLLiteApply control injection

• Optional NAGuidance model patch

• Optional Anima high-resolution aesthetic boost LoRA

• Qwen Image CLIP text encoder

• Qwen Image VAE output

• FluxResolutionNode canvas size control

• Two-stage sampling pipeline

• 1.5x latent upscale refinement

• Native PreviewImage and SaveImage output

• Clean module structure, easy to copy into larger workflows

Suggested workflow:

Prepare a clear control image first. The control image should have readable pose, depth, or scene structure. Load it into the workflow, let DepthAnything generate the control map, and use AnimaLLLiteApply to inject the control signal into the model. Then write a prompt that describes the final subject, style, lighting, and atmosphere. Use the first sampling stage to establish the controlled composition, then use the second stage for higher-resolution detail refinement. This workflow is best used when you want the visual freedom of Anima Base but need more layout control than pure text prompts can provide.

⚙️ RunningHub Workflow

Try the workflow online right now — no installation required.
👉 Workflow: https://www.runninghub.ai/post/2057391382855241730?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 & Asia-Pacific)

If you’re 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/BV1jALa6xEqH/

☕ 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 — just like a spark that can ignite a blazing flame.
👉 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/2057391382855241730?inviteCode=rh-v1111
如果觉得效果理想，你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，畅玩 4090 体验 48 G 超级性能！

📺 Bilibili 更新（中国大陆及南亚太地区）

如果你在中国大陆或南亚太地区，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/BV1jALa6xEqH/

我会在 夸克网盘 持续更新模型资源：
👉 https://pan.quark.cn/s/20c6f6f8d87b
这些资源主要面向本地用户，方便进行创作与学习。