{"id":29854896,"url":"https://github.com/ryanontheinside/comfyui_david","last_synced_at":"2025-07-29T22:19:16.992Z","repository":{"id":306783380,"uuid":"1027206541","full_name":"ryanontheinside/ComfyUI_DAViD","owner":"ryanontheinside","description":"Wrapper for Microsoft's DAViD, providing depth and normal estimation, as well as foreground segmentation. ","archived":false,"fork":false,"pushed_at":"2025-07-27T14:52:01.000Z","size":18,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-07-27T16:44:29.172Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ryanontheinside.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-07-27T14:40:45.000Z","updated_at":"2025-07-27T15:13:23.000Z","dependencies_parsed_at":"2025-07-27T16:44:31.124Z","dependency_job_id":"64bc345e-8efb-418f-be59-713376b214fa","html_url":"https://github.com/ryanontheinside/ComfyUI_DAViD","commit_stats":null,"previous_names":["ryanontheinside/comfyui_david"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/ryanontheinside/ComfyUI_DAViD","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ryanontheinside%2FComfyUI_DAViD","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ryanontheinside%2FComfyUI_DAViD/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ryanontheinside%2FComfyUI_DAViD/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ryanontheinside%2FComfyUI_DAViD/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ryanontheinside","download_url":"https://codeload.github.com/ryanontheinside/ComfyUI_DAViD/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ryanontheinside%2FComfyUI_DAViD/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":267771843,"owners_count":24142081,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-07-29T02:00:12.549Z","response_time":2574,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2025-07-29T22:19:13.686Z","updated_at":"2025-07-29T22:19:16.964Z","avatar_url":"https://github.com/ryanontheinside.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# ComfyUI DAViD Custom Node\n\nComfyUI wrapper for [DAViD (Data-efficient and Accurate Vision Models from Synthetic Data)](https://microsoft.github.io/DAViD) - human-centric computer vision models for depth estimation, foreground segmentation, and surface normal estimation.\n\n## Features\n\n- **Automatic Model Downloads**: Models are automatically downloaded to `models/DAViD/` when first used\n- **Multiple Model Options**: Support for both Base (ViT-B/16) and Large (ViT-L/16) variants\n- **Efficient Multi-Task Processing**: Single model can perform all three tasks simultaneously\n- **ComfyUI Native Outputs**: Proper IMAGE and MASK tensor formats for seamless workflow integration\n\n## Available Nodes\n\n### DAViD Model Loader\nLoads and manages DAViD ONNX models with automatic downloading.\n\n**Inputs:**\n- `model_type`: Dropdown selection from available models\n  - `multitask_large` (recommended) - Performs all tasks simultaneously\n  - `depth_base` / `depth_large` - Depth estimation only\n  - `foreground_base` / `foreground_large` - Foreground segmentation only  \n  - `normal_base` / `normal_large` - Surface normal estimation only\n\n**Outputs:**\n- `model`: DAVID_MODEL type for use with inference nodes\n\n### DAViD Multi-Task Estimator\nPerforms all three tasks simultaneously using the multi-task model (most efficient).\n\n**Inputs:**\n- `image`: IMAGE - Input image to process\n- `model`: DAVID_MODEL - Multi-task model from loader\n\n**Outputs:**\n- `depth_map`: IMAGE - Relative depth map visualization\n- `foreground_mask`: MASK - Human silhouette segmentation\n- `normal_map`: IMAGE - Surface normal map visualization\n\n### Individual Task Nodes\n\n#### DAViD Depth Estimator\n**Outputs:** `depth_map` (IMAGE)\n\n#### DAViD Foreground Segmenter  \n**Outputs:** `foreground_mask` (MASK)\n\n#### DAViD Normal Estimator\n**Outputs:** `normal_map` (IMAGE)\n\n## Usage Example\n\n1. Add **DAViD Model Loader** node, select `multitask_large`\n2. Add **DAViD Multi-Task Estimator** node\n3. Connect your input IMAGE to the estimator\n4. Connect the model output from loader to estimator\n5. Use the three outputs (depth, mask, normals) in your workflow\n\n## Technical Details\n\n- **Input Resolution**: Models expect 384x384 input (automatically handled)\n- **Output Format**: \n  - IMAGE tensors: (1, H, W, 3) in range [0, 1]\n  - MASK tensors: (1, H, W) in range [0, 1]\n- **Model Storage**: Downloaded to `{ComfyUI}/models/DAViD/`\n- **Performance**: Multi-task model is most efficient for getting all outputs\n\n## Dependencies\n\nRequired packages (automatically installed with DAViD):\n- numpy\n- onnx  \n- onnxruntime-gpu\n- opencv-python\n- torch (from ComfyUI)\n\n## Model Information\n\nAll models are trained on synthetic human data and optimized for human-centric scenes. Models are provided under MIT License by Microsoft Research.\n\nFor more details, see the [DAViD paper](https://arxiv.org/abs/2507.15365) and [official repository](https://github.com/microsoft/DAViD). ","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fryanontheinside%2Fcomfyui_david","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fryanontheinside%2Fcomfyui_david","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fryanontheinside%2Fcomfyui_david/lists"}