The photogrammetry process (converting photos into 3D models) looks at series of photos and reconstructs the camera location, orientation, lens type and lens distortion. It does that by comparing and matching pixels/points across photos in the entire scene and match/ triangulate on visual features. To get the most out of the process, and to get the best possible 3D models, there are few essential guidelines that need to be followed:
ENVIRONMENT (Lighting conditions)
Environments with strong casting shadows should be avoided. Shadows don’t work well for the reconstruction process as there aren’t many features that can be matched for reconstruction.
Indoor spaces:
– The object must be placed in a room with diffuse light that does not produce high contrast shadows.
– Do not use the camera flash. Flash can make every photo different in terms of lighting exposure, which is not recommended for the photogrammetry process.
Outdoor spaces:
– Avoid shooting at midday under bright sun. Best times are early in the morning, later in the afternoon, or on overcast days.
EQUIPMENT
– Good results can be achieved with any camera if the photos are taken in the right way. However, plastic lenses can only go so far in terms of sharpness. For getting high detailed results, a good lens makes a difference.
– Recommended are fixed lens. Fixed lens have no focal change which is good for the photogrammetry process. A recommended lens is 50 mm.
– Whenever possible, photos should be taken with tripod and a camera clicker. In order to make the entire photo sharp, sometimes it will be required to set a wide depth of field. This means that a very small amount of light will be coming into the camera so the shutter speed will have to be set on long time interval. Long shutter intervals will more likely result in blurry photos with the tiniest shake of the camera. The tripod and the clicker will assure in getting sharper photos.
PHOTO CAPTURING STRATEGIES & SCENE SET UP
– The targeted object should always be in the middle of the picture and fills most of the photo frame. This means that more pixels from the image will be used from the engine to do the matching process, which eventually will result in more detailed 3D models.
– “Rich” background can help. If capturing a single object, avoid placing on a monochromatic surface.
– Big don’t is moved objects in the scene during the capture process, or people walking in the background. Movements in the photos are one of the biggest reason for distortions in the final reconstructed model.
– There should be multiple overlapping photos, so every small feature is captured at least three times in different photos. Taking photo every 5% degrees from the same distance and height of the object is a good practice. For high detailed models, there can be multiple captured heights of the object, ending with 50-70 photos for small, 70‐100 for medium sized object like sculptures, and 180‐200 photos for buildings.
– It is not enough to have many photos if there isn’t sufficient overlap between the frames. More overlap will result in tighter, more accurate 3D model.
– The diagram below well illustrates the shooting strategy for interior spaces. Instead of capturing photos from the middle of the room and rotating around, move along the walls and take pictures in the other direction. With that more angles are captured that helps with the calculation process.
– When working with exteriors, avoid standing in one spot and shooting ‘panorama like’ spherical photos. Always move even slightly when taking the next photo, no matter if it’s interior or exterior space.
PHOTO QUALITY
– The photos must be in focus and sharp throughout the whole scene. It’s essential that the entire pictures are sharp and not just the object we are trying to reconstruct in 3D. Pixels are compared in the entire photo. If there are few blurry photos among the set, it’s key to remove them before submitting the project for reconstruction.
– Lighting exposure settings of the camera should be kept the same during the entire acquisition as much as possible.
– Use ISO 100 whenever possible (lower ISO the better). Try to get the required quality just by setting up the aperture and the shutter speed. ISO 100 creates the least amount of noise in the photos.
– Before starting the capture process, take some test pictures and zoom-in until the very last pixel to check how sharp your photo is and if the right camera settings were chosen.
GOOD SAMPLE PHOTOS:
VIDEO PLAYLIST WITH SAMPLE 3D MODELS
Many more sample models can be found on the ReCap YouTube channel.
3D PRINTED SAMPLE MODELS GENERATED FROM PHOTOGRAPHS
KNOWN TECHNOLOGICAL LIMITATIONS OF PHOTOGRAMMETRY
The photogrammetry process does not work on shiny, transparent or highly glossy objects. This is because these objects reflect the surrounding which makes the same point on one photo to be in a different color on another photo, making it impossible for the algorithms to calculate the scene.
Enjoy using the photo service on ReCap 360 and please send any questions, suggestions and wishes on the ReCap Photo forum
More deep dive conversations about the this photo capture process can be found under the Reality Solutions Webinar playlist.
For more upcoming news related to Reality Capture make sure to subscribe to our blog.
Pingback: Autodesk-forge,Autodesk-realitycapture:非常混乱的3D模型!!!正在生成 - 实战宝典
where are the pictures?!
Gracias!!