iPhone 17 Pro front camera's 1:1 square photos show lower resolution than expected

random_notrandom wrote:

Update:

Apple Support conducted multiple screen shares with me, first collecting screen recordings and then escalating my case to a Senior Advisor. The Senior Advisor connected me with another specialist (assuming Level 3 support), who also did a screen share, gathered additional information, and performed tests.

They confirmed that when using the 1:1 Aspect Ratio mode, the iPhone performs a center crop of the 18 MP sensor, which results in a 13 MP image file. According to them, in the current iOS version, this is by design. The front-facing 18 MP sensor can only capture full 18 MP images in Wide or Vertical mode... not in Square. They suggested that a future iOS update may introduce a true full 18Mp resolution square capture option that uses the full sensor without center cropping.

Finally, they recommended submitting feedback at apple.com/feedback.

So it is not a square sensor?

MikriViolinista wrote:

My suspicion, and I don't believe Apple will disclose this, is that this may be a hardware limitation. As camera lenses project a circle image, you need an image circle whose radius is at least slightly larger than the full square area of the sensor to avoid vignetting at the corners. This would require a larger lens. So with the forced crop you avoid that, and you don't need the image circle to be that large with the vertical or horizontal aspect ratio. If that is the case (and admittedly this is just speculation) the question is why isn't it still a bit larger since there would be room for a larger square image than the 13MP crop.

You may have hit the nail on the head.

I've seen similar topics come up on digital photography Web sites – where some users of dedicated cameras argue that sensors should have a 1:1 aspect ratio.

The supposed benefit of a square sensor is you wouldn't need to rotate your camera to portrait orientation to capture as many pixels as possible when photographing a subject that is taller than it is wide.

However, things are not that simple.

• If you keep the image circle projected by the lens entirely within the bounds of the square sensor, then you cannot take advantage of the full resolution of the square sensor for square photographs. You must use a smaller square area that fits within the image circle.
• If you make the image circle larger, so that the square sensor is completely within it, then all of your lenses become bigger and heavier. And it may turn out that there's not much advantage to building say, a camera system with an APS-sized square sensor and an oversized image circle, as opposed to simply going up to a system with a full-frame 3:2 sensor and an image circle that falls within the full-frame sensor area.

So even though there are, or were, some medium-format cameras with square sensors, the benefits of square sensors have not proved convincing enough to usher in their adoption in most standalone cameras that home users might buy. Let alone to usher in the adoption of square sensors and oversized image circles.

I don't know how Apple made their decisions as to how the front camera on the iPhone 17 Pro should work, but I would not be surprised if they debated similar considerations and came to similar conclusions.

As it turns out, the center stage camera doesn’t quite use a square sensor. It has a square aspect ratio, but the corners are cut to fit a larger sensor behind the lens. This means that both 4:3 aspect ratios have a vertical edge, and then there’s pretty much a diagonal between them, so it’s octagonal.

The 3672x3672 square is the biggest square that fits within that shape.

MikriViolinista wrote:

My suspicion, and I don't believe Apple will disclose this, is that this may be a hardware limitation. As camera lenses project a circle image, you need an image circle whose radius is at least slightly larger than the full square area of the sensor to avoid vignetting at the corners.

Why would they not admit that. Cameras with round lenses and square plates have existed for 150 years

GSMArena – Apple iPhone 17 Review has a photo.

They call the sensor a "unique octagonal shape sensor." ("We aren't sure exactly why, but the missing corners mean photos taken in 1:1 aspect ratio are of actually lower resolution than both 4:3 and 3:4 ones.")

Before I saw the picture, the mention of an octagonal sensor brought to mind discussions that I had seen where people proposed that sensors should be hexagonal. (Silicon wafers are round, so a bunch of hexagonal sensors packed together in a honeycomb pattern would waste less of the expensive area than a bunch of square ones.)

Here, the sides are not of equal lengths, and I'm not sure how much cutting off the corners would have helped when it came to manufacturing.

Compression are resolution (more specifically, file dimensions in terms of pixel count) are two different things though.

An image can be heavily compresses and high pixel count, and lightly compressed with a low pixel count. Both of them contribute to image quality but in different ways.

My suspicion, and I don't believe Apple will disclose this, is that this may be a hardware limitation. As camera lenses project a circle image, you need an image circle whose radius is at least slightly larger than the full square area of the sensor to avoid vignetting at the corners. This would require a larger lens. So with the forced crop you avoid that, and you don't need the image circle to be that large with the vertical or horizontal aspect ratio. If that is the case (and admittedly this is just speculation) the question is why isn't it still a bit larger since there would be room for a larger square image than the 13MP crop.

If you put a compass with a point in the center the square and adjust the diameter to the blue vertical lines and draw your circle. The circle represents the projection from the rear of the lens. The small part of the corners would be subject to a small vignette that computational photography could easily adjust for in a lens profile.

When I shoot large format I sometimes shoot with lenses that don’t project a large enough circle. But it can be “fixed” in darkroom or Photoshop after I scan my film. Apple could do it and it would be fine for selfies etc.

Do you have Center Stage enabled?

Use Center Stage to keep you centered in the camera frame - Apple Support

I'm guessing that the only way Center Stage can work is by recording from a smaller area than the full sensor area. Automatic zooming and rotating would be accomplished in part by changing the area of the sensor from which the final picture or video was made.

What you want is called Open Gate. The rear cameras now shoot Open Gate, but the front camera does not. You can learn more about Open Gate here,

https://www.panasonic.com/uk/consumer/cameras-camcorders/lumix-expert-advice-learn/lumix-expert-advice/what-is-open-gate-video-and-why-is-it-a-big-deal.html

The article above references a 3:2 sensor while the rear cameras on iPhone are 4:3 and the front camera is 1:1>

Just adding my vote that "open gate" capture of the full square image would likely be a "win" for millions of users, even if they dont know it just yet. ha. Classic scenario is a Youtube creator who wants to primarily film landscape long format but then wants some shorts for tiktok, YT shorts etc... having that extra vertical resolution will give the necessary headroom so the vertical crop doesnt zoom in the face

Btw you can partially get this via Black Magic app and choosing Open-Gate. hold phone horizontally and film youll get a 4032x3024 vid image that is about 50% taller. this maths out to 12mp.. so not the full 18mp sensor, but it is larger than 4k

Center Stage involves automatic zooming and rotating to "center" attention on a subject.

The iPhone cannot physically move itself and rotate itself to achieve this zooming and rotating. That leaves it with the option of generating output based on a CROP of the full mage, with the CROP rectangle position automatically changing to simulate zooming, panning, and rotating.

It stands to reason that a cropped image might have fewer megapixels than one based upon recording every pixel value captured by the whole sensor.