Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

How do I improve the resolution of Realsense Viewer? #6104

Closed
nbonwit opened this issue Mar 21, 2020 · 6 comments
Closed

How do I improve the resolution of Realsense Viewer? #6104

nbonwit opened this issue Mar 21, 2020 · 6 comments

Comments

@nbonwit
Copy link

nbonwit commented Mar 21, 2020

I am imaging Lego bricks at about 200-300mm. I want to get more resolution from the viewer. I want to get sum-mm resolution (e.g., .1 or .01mm).

I have Controls->Depth units set to 0.0001 (i.e., 0.1mm). I have Depth Table ->Depth units set to 1000.

When I hover the cursor over the depth map, it only shows 3 digits and the color map is very broad. I want to visualize the bumps on the Lego bricks but don't know how to do it.
image

image

@agrunnet
Copy link
Contributor

You are adjusting the depth setting. You should be adjusting the “visualization” instead. There are adjustments for that in the viewer. You can set colors and ranges.

Returning to the depth settings, I do want to add some comments. When viewing something near with the D415 (I presume that is what you have) you will benefit from reducing the depth units from default 1mm (1000um) to 0.1mm (setting of 100). And in order to get to this near distance you also need to do one or both of the following 1. Reduce resolution from 1280x720 to 480x240 which will reduce the MinZ from about 46cm to about 15cm. 2. Under advanced mode you change the disparity shift from zero to higher number.

@agrunnet
Copy link
Contributor

Regarding the on-screen readout in lower left of your image, the precision of that number is not adjustable in the app.

@nbonwit
Copy link
Author

nbonwit commented Mar 23, 2020

Thanks for all your tips and information. Regarding the visualization, this is how I have it set.
image

As far as I can tell, min/max only clips data but doesn't affect color ranges. Do I have that correct? Am I missing something?

Yes, I have a D415 camera and have been adjusting the disparity shift. If I set the Depth units to 100, you're saying I get 0.1mm resolution in the reported data, so if I set it to 10, then I get 0.01mm resolution? I realize there is a limit, but I'm assuming the data just gets reported to the visualizer with more decimal places, right?

Also, why does reducing the resolution affect the MinZ? What is the physics behind this? I don't understand that part. I would think that reducing the resolution would affect other things like spatial resolution, FOV, image transfer rate, etc. So I don't understand how it would affect MinZ.

@agrunnet
Copy link
Contributor

The reason the resolution affects the MinZ is that we have a hard limit on the number pixels we search when compare left and right images. This limit is 126 pixels. But if there are fewer pixels in the image it means that you can effective search a longer in terms of fraction of image.

Yes the visualization knobs are the right ones. To see the effect you may have to change the color scheme.

Regarding depth units, yes the smaller the number the smaller the depth steps that get reported. This is explained in more detail in the white paper talking about the best known methods for tuning performance. The paper also explains the MinZ.

@nbonwit
Copy link
Author

nbonwit commented Mar 23, 2020

Thanks for explaining about MinZ and depth units. I will review the tuning methods paper again on these points.

@RealSenseCustomerSupport
Copy link
Collaborator


Hi @nbonwit, do you have a chance to try agrunnet's suggestion and review the white paper? Do you need further help on this?
If we don't hear from you in 7 days, this issues will be closed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants