Pictofit / Content Service / 1.5 / Input Data Specifications

Input Data Specifications

Depending on the type of product, input data tagged with specific types has to be provided so that we can transform the data into a smart 2D or 3D object. This page lists the currently supported types and their requirements.

Garment 2D

Capturing photos is simple and quick - the ghost mannequin photo process is well established in the industry. The product type for this request is GARMENT in this case. Mind the following requirements:

  • 2 Megapixel resolution (or higher) with no/little compression (JPG or PNG)
  • Bright uniform and symmetric lighting
  • The mannequin needs to fill all parts of the garment (e.g. sleeves, legs) to achieve “3D fill” and avoid deflated, wrinkled parts
  • All parts need to be visible in the frontal shot (e.g. sleeves not behind body)
  • These shots are required to achieve the desired result:
  • (optional) A reference image of a model wearing the product

Examples

The following tables illustrate an example data set for a garment. Each configuration needs at least one frontal image showing the garment which must conform to the specs. Additional reference images can be added to give the quality control team more information about the fit and apperance of the product.

Example Garment Frontal Image FRONTAL
Example Garment Frontal Image REFERENCE_ADDITIONAL

Counterexamples

The following images show examples which don’t conform to the specifications and will be rejected or charged with additional cost.

Counter Example 1 Counter Example 1
Counter Example 1 Counter Example 1
Counter Example 1 Counter Example 1

Garment 3D

Our service can generate 3D models from 360 degree shots of garments. These images can be taken with any camera setup. A turntable speeds up the process significantly but images can also be taken by walking around. The following steps explain how it should be done. The required product type here is GARMENT_3D.

Reactive Reality's automated photo studio.
  • Dress a mannequin with the garment you want to capture. Be aware that details like wrinkles will be cap-tured as well and be part of the 3D model. The following requirements are essential:

    • The arms of the mannequin should be rotated away from the torso by approximately 35 degrees (A-Pose) to make sure that all parts of the garment are visible.
    • For short-sleeved/sleeveless shirts, the mannequin should not have arms.
  • Capture around 100 images from all viewing angles and from different levels. Only what is visible in the images can be converted into a 3D model. The following requirements are essential:

    • The photos need to show the product from multiple viewing angles (see figure below) by either walking around the product with a camera or a turntable setup with adjustable camera/product elevation. The horizontal angle between the images should be around 10 degrees and the vertical angle between different levels should be around 30 degrees.
    Visualization of how images should be taken.
    • Preferably a DSLR camera should be used to capture the images. Resolution should be at least 12 MP, preferably 24 MP. Manually set aperture, shutter speed, ISO, white balance and focal length and use the same configuration for all images of a product for best results. Disable all automatic features of the camera.
    • The object must be entirely in focus (This can be especially challenging with DSLR cameras. Choose high f/stop values for a greater depth of field).
    • Ensure bright uniform lighting, that the object is standing free of other objects and high contrast between the object and the background
  • Transfer the data to Reactive Reality’s Content Service. The images must not be altered/edited, EXIF info should be included and no special numbering or sequencing is required. The follow data is required for the captured product:

    • Image set (file type of the images needs to be ADDITIONAL_VIEW)
    • Height of the mannequin as part of the configuration’s metadata json as 'mannequin-height' : '175'. The unit is metric centimeters.

Avatars

Avatars are also generated through the Pictofit Content Service. The different SDKs provide components for capturing an avatar and also for transfering the data to the web service. Please use these components instead of sending the data to the web service directly.

© 2014-2020 Reactive Reality AG