full body scan using intel realsense sr300

[imaje]

hi i was just checking sdk examples for intel realsense camera sr300 which contains only one body scanning example but its full body scanning part is not working .or may b i am not using it properly .

anybody can provide help regarding this .

Any where i can get guide for complete sdk details for developers ?

[MartyG]

The SR300 is not best suited for full body scanning, sadly, as it is best at scanning small objects. It is the R200 camera model that is better at large objects and full human bodies, and the Person Tracking sample that is packaged with the RealSense SDK is a R200-only program .

Having said that, the RealSense documentation claims that the Person Tracking feature can work with the SR300 as well as R200.

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?doc_pt_person_tracking.html Intel® RealSense™ SDK 2016 R2 Documentation

In practice though, I have only seen one example where Person Tracking has been successfully used with the SR300.

https://vimeo.com/166142848 RealSense SR300 Person Tracking Preview from .NET on Vimeo

It would be useful if you could confirm for us what you want to use the SR300 camera for - full body scanning (making a 3D model of the body), or full body tracking (tracking the position of joints on the body). Thanks!

[imaje]

thanks Marty for your response i need it for full body scanning (making a 3D model of the body), i am working on some virtual dressing idea i have both cameras but i have seen r200 examples are not showing in updated sdk .can u tell me how can we use both sdks together. waiting for ur reply any helpful link about it

[MartyG]

Support for the R200 camera has been removed from the '2016 R3' RealSense SDK onwards, supporting only the F200 and SR300. That is why there are no R200 samples in the R3 SDK. Developers who wish to use the R200 must use the '2016 R2' SDK.

At this time, there is not a clear advantage in using R3 over R2, other than R3 taking up less installation space due to it being split up into a series of modules instead of all being in one package like R2 is. So my advice would be to use R2 for both the R200 and SR300.

[imaje]

so can we use sr300 examples on r200 camera is it possible because i have seen their are many features which are not provided by r200 but are available in sr300 then how we can use them also .i have seen a post on some site they were saying we can use both using librealsense library

https://software.intel.com/en-us/blogs/2016/01/26/realsense-linux-osx-drivers Intel® RealSense™ Driver for Linux* & OS X* | Intel® Software

[MartyG]

You can use both the R200 and SR300 cameras with Librealsense. Some features that Librealsense has, the official RealSense SDK lacks though. Likewise, the RealSense SDK has features that Librealsense lacks. The choice comes down to which SDK can best meet your needs with its feature list.

The samples provided with the RealSense SDK in its Sample Browser application are divided into Common, R200 and SR300 sections. Samples in the R200 section only work on the R200. Samples in the SR300 section work only on the SR300. And samples in the Common section work for both cameras.

[imaje]

here they have discussed it in detail

https://github.com/IntelRealSense/librealsense/issues/138 https://github.com/IntelRealSense/librealsense/issues/138

will it be a good solution that first we scan the object and get 3d model using r200 and switch between cameras for different tasks so that they dont interfere each other when pointing to the same target

[imaje]

and they suggested also this technique

https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/shake27n27Sense.pdf https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/shake27n27Sense.pdf

[MartyG]

I can't speak for multi-camera use on Librealsense, as I do not have experience of that. With the RealSense SDK though, you can attach multiple cameras to the same PC but only one of those cameras can be active at a time. So you use camera IDs to cycle through each camera, activating it and taking some data and then activating the next camera in the sequence until you get back to the first camera again. So you could arrange a number of R200s in a 360 degree circle around a subject that you want to scan and cycle through each camera in a loop.

[MartyG]

The new RealSense ZR300 camera has technology that helps to avoid interference when using multiple ZR300's. Its data sheet document says:

[imaje]

so what you will recommend me zr300 or use sr300 and r200 together ?

[MartyG]

The ZR300 is more advanced camera technology than the other two cameras but it currently requires Librealsense, as the RealSense SDK does not support it. So it would depend on whether you can achieve your project goals with Librealsense. The ZR300 is also likely to provide superior scanning results than those of the R200.