datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Hunyuan3D-FLUX-Gen
Orient Anything V2 Dataset
Project Page | Paper | GitHub
Orient Anything V2 is an enhanced foundation model for unified understanding of object 3D orientation and rotation from single or paired images. This dataset repository supports the model by providing assets for orientation estimation, 6DoF pose estimation, and object symmetry recognition.
Data Preparation
You can download the absolute orientation, relative rotation, and symm-orientation test datasets using the… See the full description on the dataset page: https://huggingface.co/datasets/Viglong/Hunyuan3D-FLUX-Gen.fomo_in_flux_dataflux-dev-portrait
FLUX-dev portrait
This dataset is generated with FLUX.1-dev using default settings (guidance scale = 3.5, 50 steps) at 1024x1024 resolution. The prompts are taken from k-mktr/improved-flux-prompts-photoreal-portrait. The images are converted to uint8 and encoded with lossless WEBP, and packed in WebDataset format.
This can be used for distillation to another model or self-distillation when training new capabilities for FLUX.
Currently the generated faces are not diverse. Future… See the full description on the dataset page: https://huggingface.co/datasets/gaunernst/flux-dev-portrait.logos_fluxThe_full_pure_image_of_FLUX-Reason-6Msoundnet-flux-captionflux-ultra-stockimagesflux_gen_imagesflux_rl_embeddingsffhq-flux1-latents-256flux_image_intrinsicflux_gen_imagesflux_image_edit_masksflux_image_edit