NIH 3D is an open, interactive platform for discovering, sharing, and generating 3D models related to bioscience and medicine. It supports uses across 3D printing, augmented and virtual reality, animation, illustration, and web-based visualization. Models include molecular structures, cellular components, anatomical features, and open source labware.

NIH 3D is the next-generation version of the NIH 3D Print Exchange, with expanded capabilities beyond 3D printing. It offers a modernized interface, new tools, support for additional formats and technologies, and workflows for generating 3D models from scientific data.

• Researchers: Visualize complex scientific and clinical structures
• Educators & Students: Use 3D models as interactive or printable teaching aids
• Presenters: Improve scientific communication with 3D visualizations
• 3D Enthusiasts: Explore and remix bioscientific models

• Purpose-built for biomedical and scientific visualization
• Free to use
• Integrated tools to convert scientific data into printable or interactive formats
• Model submission includes support for metadata, licensing, and citation

Anyone with a free NIH 3D account can submit models relevant to bioscience, medicine, or biological research.

From the submission page page, you can either:

• Use QuickSubmit to generate models from external databases (PDB, AlphaFold, PubChem, EMDB).
• Upload your own 3D files (e.g., .glb, .stl, .x3d, .wrl) or scientific data (e.g., .pdb, .mol, .sdf).

NIH 3D supports upload and automatic conversion of:

• 3D mesh formats: .glb, .x3d, .stl, .wrl
• Scientific formats: cif, .pdb, .sdf, .mol, mol2, .map, .omap, .mae, .pdb1, .ent, .tiff, and .tif
• Medical imaging formats: zip archive of dcm files

Yes, but all Protected Health Information (PHI)/Personally Identifiable Information (PII) must be removed. We do not allow sharing of raw data that includes the face. NIH 3D automatically strips personal data and flags raw DICOM uploads for moderator review, but ultimately the user is responsible for ensuring that no sensitive data is included. Many dicom viewers allow you to review and edit DICOM tags (e.g., MicroDicom DICOM Viewer). NIH 3D will rename your input file in case identifiers have been included as part of the filename.

Yes, but with some limitations based on data type:

• For mesh files, you can upload multiple mesh files at a time (e.g., .stl, .glb, .x3d, .wrl).
• For molecular data files (e.g., .pdb, .mol, .sdf) or zip archives of .dcms, only one file can be uploaded per submission.

This setting controls whether raw input files (e.g., .cif or .pdb files) are downloadable. For medical imaging we do not allow data that includes unobscured facial information to be made public.

Mesh repair is a process that attempts to fix common 3D printability issues by generating a watertight, manifold mesh. It is enabled by default on workflows that use ChimeraX for representations intended for printing and that are not in color (since color prints can’t be repaired by our system). If mesh repair fails, we recommend running with mesh repair disabled, then processing the output with the free version of Autodesk NetFabb.

Yes. After submission, you can replace the input files. After you publish, this would result in a major version change (e.g., 1.0 to 2.0) and the old file will still be available on the site in that past version.

You can generate a 3D model manually using Omega or Avogadro then upload the resulting file.

Your model will be processed by our workflows to create multiple downloadable formats. Optionally, you can run mesh repair.

• Versions track changes to files or metadata
• Remixes are derivative works based on existing entries

Our workflows use Blender and ChimeraX that are also free for public use. These are both good environments to build and test models you might like to upload to NIH 3D.

All content must be relevant to biomedical science. We reserve the right to remove links to commercial websites that aren't directly relevant to the uploaded content. Inappropriate or irrelevant content is moderated and may result in account removal. Should you have content that you think is appropriate for NIH 3D but it doesn’t not have an applicable Category for upload, please contact us.

Yes. Most models are downloadable in .glb, .stl, .x3d, and .wrl. Licensing varies, so please review the terms of the assigned license.

• .stl: Best for monochrome 3D printing
• .wrl: For full-color powder bed printing
• .glb/.x3d: Best for interactive or web use

You can 3D print files at a variety of locations, including local libraries, makerspaces, and commercial print services. Many public libraries now offer 3D printing as part of their tech labs, while community makerspaces provide access to printers and technical support. For more advanced or large-scale printing, online services allow you to upload files, choose materials, and have your print shipped to you.

NIH Staff Only: If you are with NIAID, please use contact us and we can connect you with the NIAID 3D printing service, which is offered at no cost to NIAID staff (contact niaid3dservices@nih.gov). The NIH Library offers a free printing services for NIH staff, and NIBIB’s Beta Center Makerspace provides the NIH research community with access to various fabrication tools and equipment. You can also join the NIH 3D Printing SIG for updates and events.

Yes, .glb files can be inserted into Word and PowerPoint documents. This is a convenient and powerful way to let your audience interact with a 3D model directly within the slide. It transforms static visuals into dynamic, engaging content to convey spatial or structural relationships clearly. Note that only the desktop versions of Word and PowerPoint support this capability.

You should cite a 3D model entry in the NIH 3D database when submitting a manuscript to a journal for several important reasons — both scientific and ethical:

• Ensure reproducibility and transparency
  Citing the specific NIH 3D model entry provides a persistent, traceable link to the exact version of the model used in your research or figures. This allows reviewers and future readers to:
  • Verify the data source and metadata (e.g., structure origin, annotations, or processing method)
  • Reproduce your visualizations, simulations, or 3D prints accurately
• Give proper credit to model creators and contributors
  Many NIH 3D entries are derived from original scientific work—such as PDB structures, imaging datasets, or computational reconstructions. A formal citation ensures that the authors, labs, and repositories who developed and curated those models receive appropriate scholarly credit.
• Strengthen data provenance and compliance
  Citing the NIH 3D entry supports FAIR data principles (Findable, Accessible, Interoperable, Reusable) and helps journals and funders verify that:
  • The 3D data are from a trusted, permanent, and open-access repository
  • You are complying with data sharing and open science requirements often expected by NIH and publishers
• Provide a stable reference for future use
  Unlike supplemental figures or lab-specific links, NIH 3D assigns each model a persistent identifier (PID or accession number) and a stable landing page, ensuring long-term accessibility even if your institution’s systems change.
• Align with best practices in scholarly publishing
  Just as you would cite a PDB ID for a molecular structure or a GenBank accession number for a gene sequence, citing an NIH 3D entry follows community standards for referencing digital scientific assets.

When citing an entry in NIH 3D, please refer to the ‘Attribution Instructions,’ as the author may have included specific guidance on how their work should be cited. This may include referencing a manuscript for which the entry was developed.

In addition to any instructions provided by the author, to cite an entry in NIH 3D, you should include the author/username, version number, year, URL, and DOI (if applicable). Below are examples of a few commonly used citation formats. Citation structures may vary among journals. You should check the publisher's instructions before submitting a manuscript.

• MLA

  BioVisUser123. Human Heart Model. Version 2.1, NIH 3D, 2024. https://3d.nih.gov/entries/3DPX-012345. https://doi.org/10.xxxxx/3dpx/xxxx.x

• APA

  BioVisUser123. (2024). Human Heart Model (Version 2.1) [3D model]. NIH 3D. https://3d.nih.gov/entries/3DPX-012345. https://doi.org/10.xxxxx/3dpx/xxxx.x

• NLM

  BioVisUser123. Human Heart Model [3D model]. Version 2.1. NIH 3D. 2024. https://3d.nih.gov/entries/3DPX-012345. https://doi.org/10.xxxxx/3dpx/xxxx.x

• AMA

  BioVisUser123. Human Heart Model [3D model]. Version 2.1. NIH 3D. Published 2024. Accessed April 2, 2025. https://3d.nih.gov/entries/3DPX-012345. https://doi.org/10.xxxxx/3dpx/xxxx.x

A DOI is a persistent identifier for citing digital content. NIH 3D automatically assigns DOIs to QuickSubmit entries. Other submissions may request a DOI, which is reviewed by NIH 3D admins.

Since the Quick Submit process relies solely on data from PDB, EMD, AlphaFold or PubChem and does not involve user-provided data, NIH 3D assumes creator rights because it oversees all the creative and technical aspects of entry creation.

For Source Data uploads the user can specify creator(s) and NIH 3D is automatically included as our workflows provide the creative and technical aspects of the model creation.

For Mesh File uploads the user can specify the creator(s).

Your DOI corresponds to the major version of your entry at the time the DOI request is made. If you need to implement modifications that affect the input/output files, you must request an update for your existing DOI. Minor alterations to your entry, such as text or metadata changes, do not warrant a new DOI.

NIH 3D uses records from the Research Organization Registry (ROR) to list affiliations in the DOI request form. If you do not see your organization listed, you can suggest a new record using the ROR Request Form on their website.

The DOI is specifically assigned to encompass only the input and output files, if they are present, and does not encompass supplementary files or associated metadata.

No. A DOI is a permanent identifier. Once it is created and published it is permanent. Please contact us if you run into issues.

WebXR is an application programming interface that enables communication between a web browser and an extended reality (XR) device. “XR” is a collective term that includes virtual, augmented, and mixed reality.

The WebXR Viewer is an immersive mode that allows users to explore NIH 3D models in virtual reality (VR) using a supported web browser and VR headset.

You will need a VR headset with a WebXR-compatible browser (the NIH 3D WebXR Viewer has been tested in the Meta Quest 2 and Quest 3 using the Meta Quest Browser).

We recommend going to the NIH 3D entry you would like to view using the Meta Quest Browser in your Quest 2 or 3 headset. Once at the entry, click the “WebXR” button to launch the viewer application. From there, you can click the “Enter XR” button that will display in the browser to enter VR mode.

We have not tested other browsers/headsets for compatibility, but welcome feedback through our contact us form.

Only models that are <50MB will be loaded as larger files often cause the browser to crash. There will be a warning if the files are approaching this value as they may still experience lag depending on your device and/or network.

The viewer will attempt to load the model if it is <50MB, but it may experience significant lag or crash the browser depending on your hardware, the browser used, and the size of the model.

The WebXR experience may not be suitable for all users, especially those with visual impairments or motion sensitivity. We are committed to accessibility and provide a standard viewer as an alternative.

Users may experience lag or reduced performance when using the NIH 3D WebXR Viewer due to several factors. These include hardware limitations, such as insufficient processing power or memory, which can affect how smoothly complex 3D models are rendered. A slow or unstable internet connection may also lead to delays, especially when loading large models. Performance can vary depending on browser version and system load. Running multiple applications or browser tabs in the background can further impact responsiveness. In some cases, performance issues may also stem from the 3D content itself if it has not been optimized for real-time rendering in immersive environments.

Some models are intentionally anchored at specific real-world positions rather than centered in your view. For example, several models in the Human Reference Atlas Collection are designed to assemble into a full-scale human body. This means their default coordinates place them where they would exist anatomically — so if you’re seated, you may need to look up to see the brain. Take a good look around, and if you still can't find the model, please contact us with the entry ID and our team will look into the issue.

The Heart Library is a curated collection of 3D-rendered congenital heart models created through expert collaboration between radiologists, congenital cardiologists, and biomedical engineers. All models are peer-reviewed and reflect a commitment to accuracy, reproducibility, and educational value.

While NIH 3D does not generate these models, the rendering process follows established quality guidelines developed by the 3D Heart Collaborative. Each model originates from high-resolution DICOM datasets (ideally 1–1.5 mm isotropic) that are free of artifacts interfering with critical anatomy. All identifiable patient information is removed prior to submission.

Models typically follow one of three segmentation approaches:

• Solid Blood Pool Segmentation

  Contrast-enhanced blood structures are segmented to produce a “negative cast” of the heart, often used for extracardiac vascular assessment.

• Blood Pool with Myocardial Border

  A thin, uniform layer is applied to the blood pool segmentation to simulate the myocardial surface. This is the most used method and supports both intra- and extracardiac visualization. Vessel ends are digitally opened to prevent occlusion artifacts.

• Myocardial and Vessel Wall Segmentation

  This detailed method highlights heart walls and intracardiac structures. It uses standardized thresholding techniques to avoid manual errors and supports high-fidelity printing, though it may require specialized print methods for fine structures.

Each model undergoes a final quality check comparing the 3D segmentation to the original imaging data.

The library aims to:

• Educate: Provide accurate, peer-reviewed models for learners at all levels.
• Collaborate: Support the sharing of expert-created resources and workflows.
• Inspire: Encourage new use cases and innovation in 3D modeling for congenital heart disease.
• Facilitate: Raise awareness of the clinical and educational utility of 3D modeling.
• Standardize: Establish best practices for segmentation and quality control.
• Advance: Promote continued refinement of modeling techniques and call for creative solutions to shared challenges.

This collection was launched to address urgent PPE shortages by rapidly evaluating and disseminating 3D-printable medical devices. Its goals were:

• To test and validate 3D-printed devices for potential clinical use.
• To guide makers and manufacturers on design and material safety.

• NIH 3D hosted the collection and made it accessible to the public.
• FDA provided regulatory guidance via Emergency Use Authorizations (EUAs).
• VA Innovation Ecosystem reviewed and tested devices in clinical settings.
• America Makes facilitated connections between health care providers and manufacturers.

No. Unless explicitly stated, designs are not FDA approved, or NIH endorsed. Categories include:

• 

  Prototype: Initial submission, not yet reviewed.

• 

  Community Use: Suitable for public settings, not clinical care.

• 

  Clinically Reviewed: Tested by the VA in clinical environments.

• 

  FDA Emergency Use Authorization: Temporarily authorized during the emergency.

• 

  Warning: Requires regulatory oversight due to associated risks.

Yes, but users should exercise caution. 3D printing results vary widely depending on materials, printer settings, and post-processing. Designs are offered as-is. Neither NIH 3D nor partner organizations are liable for outcomes.

Not necessarily. Desktop 3D printers—especially FDM printers—often lack the precision needed for fluid resistance or proper filtration. Only facilities that follow best practices in quality control may produce PPE at clinical-grade levels.

An IFU explains how to correctly assemble, use, and maintain a device. IFUs in this collection may not meet formal regulatory standards and are provided for informational use only.

• Face Mask: General use; not necessarily fluid-resistant.
• Surgical Mask: FDA-regulated, protects against large droplets.
• N95 Respirator: Offers tight fit and filters 95% of airborne particles; regulated by NIOSH and FDA.

Per 42 CFR 84.181, N95 respirators must filter at least 95% of 0.3-micron particles.

Generally no. Most PPE is designed for single-use. Exceptions exist (e.g., goggles, elastomeric respirators) but must follow strict decontamination procedures.

They may be unsuitable for children, people with facial hair, or those with respiratory or cardiac conditions. Only use N95s in sterile settings if they lack exhalation valves.

Caution is advised. Healthcare workers should:

• Check seal integrity
• Confirm breathability
• Use proper disposal methods
• Understand the limitations of 3D-printed filtration

No. This program is no longer accepting new submissions. Historical review procedures are outlined below for archival reference.

• Clear documentation and images
• Printing and assembly instructions
• Material recommendations
• Use case description (clinical or community)

• Fluid resistance (ASTM F1862)
• Flammability (16 CFR 1610)
• Air exchange (MIL-M-36945C)
• Bacterial filtration (ASTM F2101)
• Sub-micron filtration (ASTM F2299)

• Clarity of usage instructions
• Ease of printing and assembly
• Fit and durability testing
• Adequate airflow and wearer comfort

These required:

• Emergency Use Authorization (EUA) from the FDA
• NIOSH approval (42 CFR 84)
• OSHA fit testing (29 CFR 1910.134)

Reviewers provided feedback and allowed for resubmission after revisions. Unsuccessful designs were not promoted.

Scoring considered:

• Community demand
• Feasibility of the design
• Completeness of documentation

NIH 3D hosted the collection but did not conduct reviews. Clinical evaluations were led by the VA Innovation Ecosystem and partners.

For non-regulated community-use items, formal labeling is not required but encouraged. Labeling best practices:

• Product name
• Use type (clinical/community)
• Material info
• Link to IFU

During the emergency, the FDA permitted improvised PPE when no approved alternatives were available, per the Enforcement Policy for Face Masks and Respirators.