- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
I've encountered the following behavior:
Changing MvR values going from zero up, there is a point there the depth image get's some kind of deformation (may be it is being folded). As a result the depth reading of interest becomes wrong (~ 5mm less than it actually is).
Turning the camera off and on and changing the MvR value solves the issue.
I am working with facial landmark detection feature on.
I have 2 main questions about it:
1. Is it a known behavior? Why does it happen? Is there any work around that does not include to restart the camera?
2. Is there any command that does a "soft restart" => forcing the camera to recalculate/ retake the depth data?
Thank you,
Alex
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Alex,
Thanks for your interest in the Intel® Realsense™ Platform.
Does the depth data gets this kind of deformation with every app? Because the facial detection is just an Alpha maturity app, so maybe by changing application this behavior might change.
We will be waiting for your results, have a nice day!
Best Regards,
-Jose P.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Jose,
Thank you, for the fast reply.
Unfortunately my company is interested only in the facial landmarks feature, therefore I won't be able to test it on other apps. But in my opinion the facial detection rides on top of the data it gets from the depth measurements – so it shouldn't be application dependent.
Anyways, is it possible to reset the depth data without turning on and off the camera?
Best regards,
Alex.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Alex,
Would you be so kind to provide me the code or sample that you are using, so that I can attempt to replicate your issue.
We will be waiting for your reply, have a nice day!
Best Regards,
-Jose P.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Jose,
I am using SDK 10. The code is based on the face-tracking example supplied together with the SDK (DF_FaceTracking_vs2010.sln). The MvR (motion vs range) is changes via SetIVCAMMotionRangeTradeOff command externally via graphic interface we have prepared for this purpose.
I've snooped around in the documentation, and I saw that exposure setting of MvR less or equal to 8 has interpolated HVGA as VGA, could it be related?
+ how about soft reset of the camera?
Alex.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Alex,
Would you be so kind to tell me what do you mean with SDK 10? At the moment I'm only aware of the SDK R2 and the SDK R3. Also, I would appreciate it, if you could share the link where you found the statement: "exposure setting of MvR less or equal to 8 has interpolated HVGA as VGA"
We will be waiting for your reply, have a nice day!
Best Regards,
-Jose P.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Jose, after much googling I am guessing that Alex is using the RealSense DF_FaceTracking sample with Visual Studio 2010.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi guys,
sorry for the confusion, I am using SDK R2(V10).
The link to the document I've been quoting:
http://www.mouser.com/pdfdocs/intel_realsense_camera_sr300.pdf http://www.mouser.com/pdfdocs/intel_realsense_camera_sr300.pdf
Even though it is not an official Intel site, the rest of the data in the document was quite accurate.
You keep ignoring my question about the possibility of soft reset, is it because there is no such command?
Alex.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
It occurred to me that you may be referring to R2 by its other version number - 10.0.26 - but I thought it seemed too obvious. My teacher in my school class when I was 6 years old told my mother that I got hard math questions right but easy ones wrong because I over-thought the problem and thought the easy ones must be trick questions. I guess I haven't totally escaped from that way of thinking. My apologies!
The official Intel data sheet link is below. It's the same document though.
https://software.intel.com/sites/default/files/managed/0c/ec/realsense-sr300-product-datasheet-rev-1-0.pdf https://software.intel.com/sites/default/files/managed/0c/ec/realsense-sr300-product-datasheet-rev-1-0.pdf
One user's approach to a soft reset a couple of years ago was to stop and then restart the camera DCM service.
https://software.intel.com/en-us/forums/realsense/topic/560608 Getting "Init failed"? Here's how to restart the Depth Camera Manager
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
The problem with stopping and restarting the DCM service along with stopping the streaming and starting it again is that it takes more than one second to get back to the initial frame rate. Is there anything that can keep the frame-rate, just restart the calculations?
Really appreciate your patience ,
Alex.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I found two possible references of interest in the R2 documentation.
Reset - reset the camera's "pose" data.
https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?reset_pxcsceneperception.html Intel® RealSense™ SDK 2016 R2 Documentation
ResetTracking - reset the tracking context.
https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?resettracking_pxchandconfiguration.html Intel® RealSense™ SDK 2016 R2 Documentation
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thank you!!!
I am not sure it will be applicable as is to the face tracking, but it looks like the right direction.
Thank you!!!
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page