Skip to content

Qwen / Krea / Z Image

What is included. What to know before you deploy.

All templates
Qwen / Krea / Z-Image. Generate, edit and build image datasets. Up to 15 workflows across 5 model sets. HearmemanAI.

Qwen / Krea / Z Image

Use Qwen Image 2512 for detailed portraits and text layouts, or Edit 2511 for identity-preserving edits. Add Krea 2 for visual concepts and Z-Image Turbo for photoreal images with English or Chinese text.

  • ComfyUI and JupyterLab with selectable image model sets
  • Qwen generation, editing, camera-angle and dataset graphs
  • Krea 2 RAW with Turbo LoRA; Z-Image Turbo with upscale variants

15 workflows across 5 optional model sets

See the workflow breakdown
Qwen Image 2512 4 workflows
Generation variants, including a PiD upscaling graph.
Qwen Image Edit 2511 3 workflows
Image editing, multiple angles and dataset creation.
Z-Image Turbo 3 workflows
Generation, tiled upscaling and PiD upscaling.
Krea 2 2 workflows
RAW plus Turbo LoRA generation, with a PiD variant.
Boogu 3 workflows
Separate Base, Turbo and Edit graphs.
Before you deploy

Before deploying, click Edit Template and open Environment Variables. Select at least one model family by changing its download variable from false to true. Leave unused families false; enable several only if you need them.

Model selection

Qwen Image 2512DOWNLOAD_QWEN_IMAGEDefault: false
Enable if wanted false → trueImage generation.
Qwen Image Edit 2511DOWNLOAD_QWEN_IMAGE_EDITDefault: false
Enable if wanted false → trueImage editing and camera angles.
Z-Image TurboDOWNLOAD_Z_IMAGEDefault: false
Enable if wanted false → trueImage generation and upscaling.
Boogudownload_booguDefault: false
Enable if wanted false → trueBase, Edit and Turbo models.
Krea 2download_krea2Default: false
Enable if wanted false → trueRAW, Turbo and identity-edit LoRA downloads with two generation workflows.

OptionalModel precision

QWEN_IMAGE_PRECISIONDefault: bf16
Use bf16 or fp8 for Qwen Image generation. This does not change Qwen Image Edit. Copied workflows use the selected files.
BOOGU_PRECISIONDefault: bf16
Use bf16 or fp8 for the Boogu models, independently of Qwen. Copied workflows use the selected files.

OptionalCivitAI downloads

civitai_tokenDefault: empty
Your CivitAI API token. Add it if you want to download your own LoRAs or checkpoints.
CIVITAI_LORASDefault: empty
Comma-separated LoRA model-version IDs to download into models/loras. Replace any replace_with_ids placeholder with your IDs.
CIVITAI_CHECKPOINTSDefault: empty
Comma-separated checkpoint model-version IDs to download into models/checkpoints. Leave empty if you do not need any.
How to download CivitAI LoRAs

OptionalOptional access

HF_TOKENDefault: empty
Optional Hugging Face read token for authenticated model downloads.
GITHUB_PATDefault: empty
Access token for the extra node packs used by purchased workflows. Not needed for the free workflows included in this template.
Deploy on Runpod

HearmemanAI may earn a commission when you use this Runpod link. GPU and storage charges apply.