Donate some BUZZ to make this available for Generation <3
Prompt Guide -> Click Here
⚡️ High-Definition Art Network Architecture ⚡️
by CHROMEKIDD
If you enjoy using HANA, drop a little BUZZ ⚡️ to support the vibes.
HANA 4 CHROME is more than just a model — it’s a creative playground. Built with precision and a passion for 3D aesthetics, this model is fine-tuned to give you freedom-to-create, no matter your vision. From stylized portraits to surreal concept art, HANA 4 CHROME brings your imagination to life. + NSFW friendly 🙂
🎨 My LoRA: CHROMEKIDD's Style LoRA ← Click me
📦 Format: .safetensors (VAE integrated)
⚠️ Note: Portable versions may yield different or subpar results.
🔧 Recommended Settings
Preset A
Steps: 30
CFG scale: 5.5
Sampler: Euler a
Preset B
Steps: 30–60
CFG scale: 9–14
Sampler: dpmpp_2s_ancestral (tested on AUTOMATIC1111)
Image Dimensions
Optimal: 832×1216 or 1216×832
Hires Fix
Upscale: 1.5 – 2.0
Hires steps: 15 – 17
Upscaler: R-ESRGAN 4x+ Anime6B
Denoising strength: 0.35 – 0.43
✨ Prompt Suggestions
Positive PromptHDR, raytracing, pathtracing, realistic shadows, masterpiece, best quality, newest, absurdres, highres,
Negative Promptworst quality, bad quality, low quality, lowres, anatomical nonsense, artistic error, bad anatomy,
💬 Final Notes
HANA 4 CHROME was crafted with balance, clarity, and style in mind. The VAE is already baked in — but if you face color issues, try switching to the original SDXL VAE.
Let HANA transform your canvas into a sculpted 3D dream.
Create with freedom. Create with HANA.
Description
FAQ
Comments (2)
For ComfyUI users!
Since there are no ComfyUI-specific instructions in the description, I spent some time experimenting with this unique model and figured out a workaround.
Just as the author mentioned, you absolutely need the 'FreeU' and 'DynamicThresholdingFull' nodes. However, here is the catch: if you set the mimic_scale in DynamicThresholdingFull to the recommended 8, your images will look very washed out/gray.
Here is what worked for me:
For the v9 model: Set mimic_scale to 10, and keep the CFG in your KSampler about 10-12.
For the 'huge' model: The settings above won't work. You need to lower both the mimic_scale and CFG.
If your generations are coming out gray and flat, adjusting these two settings should be your first troubleshooting step
How should I use it in ComfyUI?