If GoPros did not have fish eye lenses, what aspect ratio of video would cover the same field of vision?

[avg123]

Gopros have a fish eye lens to fit a larger field of vision in the 16:9 aspect ratio video

 

If it did not have fish eye lens, what aspect ratio of sensor/video would it need to cover the same field of vision?

 

[LAwLz]

That's not really how it works.

The video would still be 16:9 no matter what lens you use, because that's the ratio of the active part of the sensor.

 

Changing the lens would not alter which part of the sensor is active. Changing to a less wide-angle lens would cause things in the middle of the lens to become bigger though, and you'd see less stuff in the peripherals. How much it would change depends on which lens you'd use.

 

Here is a picture that explains it visually. In this picture, the GoPro's lens would be the one labeled "14mm". Switching the lens out to the 35mm would make everything in the center look bigger, thus making you see less in the peripheral view. It wouldn't change the size of the image itself though, since that's based on the sensor.

[avg123]

3 hours ago, LAwLz said:

That's not really how it works.

The video would still be 16:9 no matter what lens you use, because that's the ratio of the active part of the sensor.

 

Changing the lens would not alter which part of the sensor is active. Changing to a less wide-angle lens would cause things in the middle of the lens to become bigger though, and you'd see less stuff in the peripherals. How much it would change depends on which lens you'd use.

 

Here is a picture that explains it visually. In this picture, the GoPro's lens would be the one labeled "14mm". Switching the lens out to the 35mm would make everything in the center look bigger, thus making you see less in the peripheral view. It wouldn't change the size of the image itself though, since that's based on the sensor.

Which of these is equivalent to human eye and what aspect ratio sensor would we need to cover the field of view in 14mm

[LAwLz]

44 minutes ago, avg123 said:

Which of these is equivalent to human eye

It's usually very hard to map digital equipment to human eyes. It is also hard to say which one of those images would most accurately map to our human eyes because the field of view (how much you see on the sides) changes depending on sensor size. A smaller sensor with the same lens will show a narrower field of view than a larger sensor with the same lens. Without knowing the sensor used in that comparison we don't know what the field of view was. Since the field of view is dependent on both the sensor size and the lens it is hard to try and isolate one or the other.

 

I am not really sure why you are asking this, but it is not an easy thing to answer and there are a lot of "ifs" and "buts". It sounds like you want to know about the field of view and how it maps to the human eye.

The field of view of a person with two eyes can typically be defined in a few different ways.

Our "binocular vision", which is the field of view that both our eyes see at once, is roughly 120 degrees. 

We also have "uniocular vision", which is parts we see with just one eye. Your left eye can see things to the left that your right eye can't see. Our uniocular field is roughly 40 degrees.

In total, we humans can see somewhere around 200 degrees without eye movement. With eye movements (which we constantly do) we have an even larger field of view.

But most of our field of view is just used for detecting danger. The actual area of our field of view where we do things as perceive depth is far smaller. It's around 110-120 degrees.

 

 

 

Please note that I am only talking about horizontal FOV here. Vertical FOV is a whole other can of worms.

To skip over a lot of details, humans have around 60 degrees vertical FOV (not counting peripheral vision).

 

 

So if you want a camera lens that somewhat matches what a human sees, an argument could be made that you want a lens that provides around 120 degrees horizontal FOV and around 60 degrees vertical FOV.

I am not sure if this is intentional or not, but if you set the GoPro Hero 7 (the first GoPro I could find) to 16x9 with 0% zoom, you get 69.5 degrees vertical FOV and 118.2 degrees horizontal FOV. So that matches the human vision very well.

 

If you were to change the sensor you'd also have to change the lens in order for it to stay the same FOV.

[avg123]

2 hours ago, LAwLz said:

I am not sure if this is intentional or not, but if you set the GoPro Hero 7 (the first GoPro I could find) to 16x9 with 0% zoom, you get 69.5 degrees vertical FOV and 118.2 degrees horizontal FOV. So that matches the human vision very well.

 

If you were to change the sensor you'd also have to change the lens in order for it to stay the same FOV.

But Gopros have fish eye distortion, human eyes don't. What camera lens/sensor combination would give a image equivalent to human eye without distortion?

[LAwLz]

1 hour ago, avg123 said:

But Gopros have fish eye distortion, human eyes don't. What camera lens/sensor combination would give a image equivalent to human eye without distortion?

That's a good question, but sadly there isn't much we can do about that except apply distortion correction.

I am not 100% sure, but I think our brain applies "barrel distortion correction" to our vision and filters out the distortion we should be seeing, which is similar to what a camera with a fisheye lens captures. Our brain does a ton of "post-processing" to the light our eyes capture. As I said earlier, it is very hard to translate "computer vision" to "human vision" because what we see with our eyes is just an image our brains come up with loosely based on input from our eyes. The number of things the brain "fixes" to make our vision work is crazy.

 

Another reason why humans see a "flat" image instead of a "convex" like from a fisheye lens image might be because our retinas are concave, and the image sensor is flat. But that's purely speculation on my part.

 

 

If you want to as accurately capture what our eyes see then I suggest using a setup that results in roughly a horizontal POV of ~120 degrees and a vertical POV of 60 degrees and then applying lens correction to "flatten" the image.

But unless you have a very good reason for wanting this (like creating an immersive experience) I would argue that cameras shouldn't try and replicate what humans see. It's usually a wasted effort that doesn't result in good images/video anyway. It's usually better to just go with whatever looks nice, regardless of whether or not it matches what the human eye would see.

[NinJake]

1 hour ago, LAwLz said:

The number of things the brain "fixes" to make our vision work is crazy.

Like everything we see is actually upside down.

 

Image reference:

Spoiler

To somewhat chime into OP question even though I'm by no means an expert in photography or the human eye... I'd wager you'd want a smaller focal length (like 10-24ish mm for example) on the largest sensor you can get. (If needed, you would then use the video footage and crop what is needed to get the appropriate FOV you're looking for.) But this is with a caveat, being the best way to replicate would actually be to use some sort of a fisheye lens, even though that goes directly against your request. I mainly just wanted to chime in that fun fact!

[rrats]

it's very subjective. first of all, cameras tend to crop outputs a lot, so by having the photo the same aspect ratio as sensor you would have a lot to work with AND generally bigger lenses with bigger sensors give a tiny advantage to the picture fov.