Items with no label
3335 Discussions

Realsense D435/D415 Range Accuracy Graph

JMitc7
Beginner
3,593 Views

Is there a graph that displays the theoretical range resolution of the D435/D415 cameras at various distances. I have attached a graph of the depth resolution at various distances of another off the shelf stereo cam. Is there an equivalent graph for the intel stereo sensors.

I plan to use this sensor to measure distances around 10m and I am wondering how accurate the measurements from 10m will be.

Thanks

0 Kudos
9 Replies
RTasa
New Contributor I
1,199 Views

According to the Intel interent gods I don't have access to that PDF

0 Kudos
MartyG
Honored Contributor III
1,199 Views

The message below should have an offline version of this PDF attached.

0 Kudos
易水
Beginner
1,199 Views

Can not find the PDF you said, just show Not Authorized to Access. Could you give another link? Thx

 

0 Kudos
MartyG
Honored Contributor III
1,199 Views
0 Kudos
易水
Beginner
1,199 Views

Hi, thanks for your rapid reply, I read the pdf above, actually I'm concerned about the depth error with subject to distance,however the PDF only gives distance below 4000mm.

In this link(https://forums.intel.com/s/question/0D50P0000490SE2SAM/which-is-better-for-depth-applications-d415-or-d435), WTatt said this is a graph which has depth error-curve "Out at 10 meters". I searched a lot and didn't find the discussed graph, do you know about that and can you show me the link, thanks again!

0 Kudos
MartyG
Honored Contributor III
1,199 Views

Both the D415 and D435 can depth-sense to a range of 10 meters by default. A good general rule to follow though is to assume that the accuracy of the depth measurements will drift noticably after about 3 meters from the camera.

 

I have never seen a graph that projects RealSense 400 Series depth error up to 10 meters. WTatt was talking theoretically about how depth error increases over distance. Because the curve on the graph follows a linear math formula, it is quite easy to imagine how the curve would look if you continued projecting it on its angle of ascent after the end of the existing graph is reached.

 

If you need a graph that represents 10 m of distance, it should be possible to plot it yourself using the math formula that is provided.

0 Kudos
易水
Beginner
1,199 Views

Ah, sorry I misunderstood. But watching the trend of curve, ​d435 has error 120mm at 5m distance. According to the test by jb455(still in that link), d435 has performance of <10% up to 15m, which is quiet difference. Is the curve from PDF conservative? Or maybe just due to different test environment.

0 Kudos
MartyG
Honored Contributor III
1,199 Views

The difference is due to differences in the hardware design between the D415 and D435. I believe that the D435's wider field of view is the primary factor in the higher error factor over distance. The advantages that the D435 has over the D415, other than the wider field of view, are a smaller default minimum sensing distance (0.2 m instead of D415's 0.3 m) and a faster Global Shutter that enables it to better track fast motion than the D415 model.

 

The D415 model, though it has a larger minimum distance, smaller field of view and slower Rolling shutter, provides better image quality than the D435.

 

The values presented in the tuning guide are the ideal performance values to aim for, meaning that it is hard to get better values from practical use than the ones depicted in the charts. So they are the best-case scenario rather than a conservative estimate.

 

0 Kudos
Reply