Instructions to use greyDonald/HumanAwareDesignGeneration with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use greyDonald/HumanAwareDesignGeneration with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline from diffusers.utils import load_image # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("greyDonald/HumanAwareDesignGeneration", dtype=torch.bfloat16, device_map="cuda") prompt = "Turn this cat into a dog" input_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png") image = pipe(image=input_image, prompt=prompt).images[0] - Notebooks
- Google Colab
- Kaggle
Download THIRD_PARTY_NOTICES.md from greyDonald/HumanAwareDesignGeneration: direct link, hf CLI and curl.
- Browser
- Download file 4.14 kB
-
https://huggingface.co/greyDonald/HumanAwareDesignGeneration/resolve/main/THIRD_PARTY_NOTICES.md
- Command line
-
hf download hf://greyDonald/HumanAwareDesignGeneration/THIRD_PARTY_NOTICES.md
-
curl -L -o THIRD_PARTY_NOTICES.md https://huggingface.co/greyDonald/HumanAwareDesignGeneration/resolve/main/THIRD_PARTY_NOTICES.md
Third-party notices
This release combines separately licensed source code, user-provided research checkpoints, model resources and demonstration materials. No single license is asserted over all of them. Preserve upstream licenses and follow the terms applicable to each component.
| Component | Source and terms |
|---|---|
| FlexDM encoder source | CyberAgentAILab/flex-dm, Apache License 2.0. The license is retained at vendor/flexdm/LICENSE. The supplied encoder checkpoint was extracted from the project's locally trained FlexDM model; provenance is recorded in vendor/flexdm/provenance.json. |
| Pose token decoder | User-provided original project's VQ-VAE source and 52joints_tokenization.pth. Source provenance is in vendor/render/PROVENANCE.json; no new blanket license is assigned to that checkpoint. |
| Rotation utilities | Meta PyTorch3D, BSD-style license; retain the copyright header and accompanying vendor/render/LICENSE.pytorch3d. |
| LayerDiffuse source | rootonchair/diffusers-LayerDiffuse, based on lllyasviel/LayerDiffuse. The upstream source license is retained in vendor/pose_renderer/. Model weights have separate terms. |
| Qwen Image 2.1 | Qwen/Qwen-Image-2.1. Qwen RESEARCH LICENSE AGREEMENT, September 20, 2026; non-commercial use subject to its terms. Full LICENSE and NOTICE are retained with the model resources. Built with Qwen. |
| SMPL-X body model | Official SMPL-X model license. Obtain an authorized copy directly from the provider. The public resource repository does not redistribute it. |
| SDXL base (optional) | stabilityai/stable-diffusion-xl-base-1.0, CreativeML Open RAIL++-M. |
| SDXL OpenPose ControlNet (optional) | thibaud/controlnet-openpose-sdxl-1.0; the model card refers to the OpenPose license. That license restricts redistribution. The checkpoint is excluded from this public resource mirror; obtain it from its provider under the applicable terms. |
| IP-Adapter (optional) | h94/IP-Adapter, Apache License 2.0. Its CLIP ViT-H image encoder matches laion/CLIP-ViT-H-14-laion2B-s32B-b79K, MIT. |
| SDXL fp16 VAE fix (optional) | madebyollin/sdxl-vae-fp16-fix, model-card license MIT; preserve upstream attribution and applicable base-model terms. |
| LayerDiffuse weights (optional) | Original LayerDiffusion weights declare CreativeML Open RAIL-M. The rootonchair conversion declares MIT; conversion metadata does not replace original model terms. |
| DejaVu font | Bitstream Vera permission notice, with DejaVu changes in the public domain. The font and notice are stored under weight/assets/fonts/. |
The public weight bundle retains component-specific license texts and a source/license inventory. SMPL-X and the OpenPose ControlNet remain available only in the owner's local resource directory unless separate redistribution rights are established. Other downloaded packages retain their own licenses; see their distributions.
Original example design elements and paper figures are included for demonstration and research. Their inclusion does not grant a new blanket license to the underlying artwork, imagery, fonts or depicted material. The project README and page display only original paper figures, not development inference results.
Required Qwen notice:
Qwen is licensed under the Qwen RESEARCH LICENSE AGREEMENT, Copyright (c) 2026 Hangzhou Tongyi Laboratory Technology Co., Ltd. All Rights Reserved.