Watch the full video first if you want to understand how this Boogu-Image Base text-to-image workflow works in practice. The video shows how a standard prompt can be converted into a finished image through a clean Boogu-Image generation route, with stronger quality control than the Turbo version.
This ComfyUI workflow is designed for Boogu-Image Base text-to-image generation. Its main purpose is to provide a stable and reusable prompt-to-image pipeline for creators who want better detail, stronger composition, and more controlled image quality than an ultra-fast draft workflow. Compared with the Turbo workflow, this Base version uses more sampling steps and a full negative prompt route, making it more suitable for polished image generation, visual concept development, character images, commercial-style drafts, posters, thumbnails, and creative testing.
The workflow is built around boogu_image_base_bf16.safetensors as the main image generation model. This model is loaded through UNETLoader and sent directly into the KSampler generation stage. The text encoder uses qwen3vl_8b_fp8_scaled.safetensors with the Boogu type setting. This encoder turns the user prompt into positive conditioning for the image generation process. The VAE route uses ae.safetensors to decode the final latent result into a visible image.
The prompt section is simple but practical. A CLIPTextEncode node handles the positive prompt. In the sample workflow, the prompt describes an image of a red panda holding a sign that says “BOOGU!”. A second CLIPTextEncode node provides the negative prompt. This negative prompt is much more complete than the Turbo version and is designed to suppress common image problems such as low quality, blur, bad anatomy, bad hands, distorted text, watermark artifacts, compression artifacts, excessive particles, cluttered backgrounds, chaotic details, and low commercial quality.
The latent canvas is created through EmptyLatentImage at 1024×1024 with batch size 1. This makes the workflow immediately useful for square-format image generation, AI app covers, social media images, character concepts, product-style tests, icons, and high-quality visual drafts. Users can later adjust width, height, and batch size depending on the target format.
The sampling stage uses KSampler with 35 steps, CFG 3.5, Euler sampler, SGM Uniform scheduler, random seed mode, and denoise set to 1. This configuration is slower than the Turbo workflow, but it gives the model more room to build structure, improve detail, and respond to prompt and negative prompt constraints.
After sampling, VAEDecode converts the latent output into the final image. PreviewImage gives an immediate visual check inside the workflow, while SaveImage exports the final result. The graph is compact, readable, and easy to reuse as a base text-to-image application on RunningHub.
This workflow is suitable when you want a more stable “type prompt → generate high-quality image” experience. It is best used for prompt testing, visual design, image concept generation, character creation, social media covers, product-style images, and polished creative drafts.
Main features:
Boogu-Image Base text-to-image workflow
Stable prompt-to-image generation
boogu_image_base_bf16.safetensors main model
qwen3vl_8b_fp8_scaled.safetensors Boogu encoder
ae.safetensors VAE decoding
Positive CLIPTextEncode prompt route
Full negative prompt control
1024×1024 default latent canvas
Batch size 1 default output
KSampler generation route
35-step sampling
CFG 3.5 setting
Euler sampler
SGM Uniform scheduler
Randomized seed mode
Denoise 1 full generation
PreviewImage for quick checking
SaveImage for final export
Clean and reusable Base text-to-image structure
Suggested workflow:
Write a clear prompt that describes the subject, composition, style, lighting, camera angle, color palette, and key visual details. Use this Base workflow when you need more quality and control than the Turbo version. Start with the default 1024×1024 canvas and 35-step setting. If the image lacks precision, make the subject and composition more specific. If the result becomes cluttered, simplify the prompt and rely on the negative prompt to suppress low-quality details. Use Turbo for fast idea exploration, then use this Base workflow for cleaner and more polished final image generation.
⚙️ RunningHub Workflow
Try the workflow online right now — no installation required.
👉 Workflow: https://www.runninghub.ai/post/2067856528262127617?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/BV17ujH6eEZk/
☕ 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/2067856528262127617?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/BV17ujH6eEZk/
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。