Basic question about Head-Up Display and Near Eye Display

[nikosb]

I have seen built-in HUD in both a BMW and a Mini Cooper. In the first the windshield is used as the combiner while in the second they use a separate piece of glass located on the dashboard. The systems work by reflecting and magnifying a real image at least once in order to create a virtual image at some distance beyond the windshield. In both cases the virtual image appears to be projected close to the hood of the car, which would be 1m or 1.5m beyond the windshield.

Some head mounted displays (HMD) or near eye displays (NED), for example, the Vuzix glasses claim that the virtual image is projected at infinity. How come a head mounted display which is so tiny can project the virtual image at infinity and a built-in HUD which is at least 10times bigger can’t project the image more than a couple of meters? Do you know if HMD or NED have inherent advantages when projecting a virtual image far away related to the proximity of the optics to the pupil of the eye?

 

[A.T.]

How come a head mounted display which is so tiny can project the virtual image at infinity and a built-in HUD which is at least 10times bigger can’t project the image more than a couple of meters?

Aren't the built-in car HUDs in the lower part of the view, so the background (the road just ahead) is also just a couple of meters away? Or the eyes tend to focus on the static car front when looking down, rather than the fast moving road, so that was chosen as the projection distance.

Do you know if HMD or NED have inherent advantages when projecting a virtual image far away related to the proximity of the optics to the pupil of the eye?

Built-in aircraft HUDs project to infinity.

 

[Drakkith]

How come a head mounted display which is so tiny can project the virtual image at infinity and a built-in HUD which is at least 10times bigger can’t project the image more than a couple of meters?

Good question. I found the following on page 13 of this article: http://cdn.intechopen.com/pdfs-wm/24834.pdf

3.3.3 Distance of the virtual image
The distance of the virtual image must be between 2.5 and 4 meters from the eyes of the driver, in order to increase the time of recognition and accommodation.

3.3.4 Accommodation effect
When superimposed objects are located within the observer FOV, the object that has a shorter distance from the observer’s focus, tends to dominate the accommodation answer, this situation is shown in HUD systems when using an external combiner instead of the windshield.

Note that focusing the virtual image at infinity would require that the driver accommodate for that distance, blurring the nearby scene, which includes nearby road hazards.

 

[nikosb]

Good comments from all of you.

Aren't the built-in car HUDs in the lower part of the view, so the background (the road just ahead) is also just a couple of meters away? Or the eyes tend to focus on the static car front when looking down, rather than the fast moving road, so that was chosen as the projection distance.

It depends on the car, in the Mini Cooper that I tried the virtual image appeared on the static car front when looking down even after adjusting the position of the HUD. At that point I was wondering why it is even transparent.

Built-in aircraft HUDs project to infinity.

For obvious reasons, the enemy aircraft is miles aways.

 

[nikosb]

Good question. I found the following on page 13 of this article: http://cdn.intechopen.com/pdfs-wm/24834.pdf
Note that focusing the virtual image at infinity would require that the driver accommodate for that distance, blurring the nearby scene, which includes nearby road hazards.

Good article, the proposed HUD explained in the paper is not implemented in any car and it would project the image at 2.5m from the driver or about 1.5m beyond the windshield. Projecting a virtual image at infinity is probably too much and it would have the blurring effect that you describe. However projecting the image at 4m or7m in front would be much more practical and safe. Here is another article

http://informationdisplay.org/IDArchive/2015/MayJune/FrontlineTechnologyNextGeneration.aspx

that explains that traditional HUDs currently project the image at 2.4m in front of the driver but the next generation HUDs should project the image between 7.5m-20m in front of the driver. Those systems obviously don't exist.

So, how come the AR glasses or other head mounted displays are capable of projecting a virtual image at infinity in such a small package? What's the trick they use?

 

[A.T.]

It depends on the car, in the Mini Cooper that I tried the virtual image appeared on the static car front when looking down even after adjusting the position of the HUD.

Then a projection at infinity or at 20m doesn't make sense.

For obvious reasons, the enemy aircraft is miles aways.

The point was, that built-in HUDs can project at infinity, if it makes sense.

HUDs currently project the image at 2.4m in front of the driver but the next generation HUDs should project the image between 7.5m-20m in front of the driver.

The next generation has a bigger FOV, so it covers more of the distant background, hence needs to be projected further away.

So, how come the AR glasses or other head mounted displays are capable of projecting a virtual image at infinity in such a small package? What's the trick they use?

Built-in HUDs can project at infinity as well, it's just not always sensible:

- When the background is not far away, projection at infinity is pointless.

- With a small HUD you see the image at infinity only when your head stays within an small region laterally (unless you track the head movement and adjust the projection).

 

[Andy Resnick]

How come a head mounted display which is so tiny can project the virtual image at infinity and a built-in HUD which is at least 10times bigger can’t project the image more than a couple of meters?

Because the location of the image plane is independent of the diameter of the lens.

Do you know if HMD or NED have inherent advantages when projecting a virtual image far away related to the proximity of the optics to the pupil of the eye?

Your eye is relaxed when viewing images at infinity, so it is more comfortable to view the image for extended periods of time. Plus, when you are driving, you are generally focused at infinity, so the projected HUD image will also be in focus when you are operating the vehicle.

 

[Drakkith]

However projecting the image at 4m or7m in front would be much more practical and safe. Here is another article

http://informationdisplay.org/IDArchive/2015/MayJune/FrontlineTechnologyNextGeneration.aspx

that explains that traditional HUDs currently project the image at 2.4m in front of the driver but the next generation HUDs should project the image between 7.5m-20m in front of the driver.

The HUD referenced in the article overlays a lot of information onto the existing terrain, as you can see in the pictures. Perhaps the focal distance needs to be increased in order to compensate?

So, how come the AR glasses or other head mounted displays are capable of projecting a virtual image at infinity in such a small package? What's the trick they use?

They don't use any tricks. Focusing the HUD at infinity is a near-trivial exercise for an experienced engineer. They simply choose not to focus at infinity in HUD's.

 

[nikosb]

Because the location of the image plane is independent of the diameter of the lens.

Yes,, but in buit-in HUDs the distance between the real image and the magnifying optics is 10 times or even more larger than in a AR glass. The location of the image is dependent on that distance.

 

[Andy Resnick]

Yes,, but in buit-in HUDs the distance between the real image and the magnifying optics is 10 times or even more larger than in a AR glass. The location of the image is dependent on that distance.

No, it depends on the (effective) focal length. Larger diameter optics allow more throughput and better detail rendering, but the location of the image plane depends only on the distance between the source and lens and focal length of the lens. Because the source-lens distance and lens focal length are related for constant image distance, there are many established design principles to enable a reduction in source-lens distance while maintaining good aberration control in a short focal-length lens.

 

[nikosb]

No, it depends on the (effective) focal length. Larger diameter optics allow more throughput and better detail rendering, but the location of the image plane depends only on the distance between the source and lens and focal length of the lens. Because the source-lens distance and lens focal length are related for constant image distance, there are many established design principles to enable a reduction in source-lens distance while maintaining good aberration control in a short focal-length lens.

If the distance between the real image and the magnifying optic(s) is 30cm, then magnifying that 5times will give a virtual distance of 1.50m. In a compact setting as AR glasses the distance between the real image and the magnifying optics is probably 3cm. To get the same virtual distance the real image would have to be magnified 50 times. Does it work that way? What established design principles enable a reduction in source-lens distance while maintaining good aberration control in a short focal-length lens?

 

[Andy Resnick]

<snip> Does it work that way?

Yes.

What established design principles enable a reduction in source-lens distance while maintaining good aberration control in a short focal-length lens?

Kingslake's book is a good start. One basic idea is to split a single element into a cemented doublet, the designer gains 2 degrees of freedom (1 additional surfaces + 1 thicknesses) to control monochromatic aberrations and a degree of freedom to control chromatic aberration (choice of glass). Also, Smith's "Modern Lens Design" has the optical designs and performance characteristics for about 300 lens designs- take your pick.