Square sensor

rutt

Here's what I'd really like. A camera with a square sensor. We pay for the lens, why not use all of it? No more shoot time decisions about orientation. This was a fantastic feature of old Hasselblad and the like and I'd love to see it in digital.

Why not, do you think?

bham

Hey many of the medium formats have digital backs now so you can have that. I guess you are talking about it on a DSLR. Interesting idea. Maybe that will be the new rage in a year or two, square, as now the rage is FF.

wholenewlight

rutt wrote:

Here's what I'd really like. A camera with a square sensor. We pay for the lens, why not use all of it? No more shoot time decisions about orientation. This was a fantastic feature of old Hasselblad and the like and I'd love to see it in digital.

Why not, do you think?

Me too!

About 25 years ago I went to a week long course at the "Hasselblad University" (actually a touring Hasselblad training program for camera salespeople and Hasselblad photographers). It was taught by the professor, author and photographer, Ernst Wildi.

I still remember him extolling the virtues of square format composition. And he's still doing it online. Link

Bob Bell

wholenewlight wrote:

Me too!

About 25 years ago I went to a week long course at the "Hasselblad University" (actually a touring Hasselblad training program for camera salespeople and Hasselblad photographers). It was taught by the professor, author and photographer, Ernst Wildi.

I still remember him extolling the virtues of square format composition. And he's still doing it online. Link

I know the 35mm is sacred so if you are going to really mess with it, make the sensor a circle. 36mm in diameter, that way you can orient the body any way you want and still have a decent ratio for larger prints.

Blurmore

If ya got the resolution...
You could do mask off the viewfinder so you were always on a sqaure. I have heard non confirmed stories about famous fashion and portrait photographers who were used to shooing 6x6 doing this to their 35mm SLRs.
Masking a 35mm negative would prevent you from getting a maximum enlargement unless the shot was dead on. But masking say a 1DSMkII would leave you with enough res to enlarge and possibly even crop for over 20 inch prints. I think this is the main drawback, the V/H format allows camera manufacturers to get the maximum sales impact of xmegapixel without expotenially increasing sensor size and cost. Imagine if sensor manufacturers had to go from say a 3cmx3cm 3000x3000 pixel sensor to a 3cmx3cm 6000 x 6000 pixel sensor to double resolution. The best way I've found to get 6000x 6000 pixel images is to scan them from my medium format positives :P

mercphoto

rutt wrote:

Here's what I'd really like. A camera with a square sensor. We pay for the lens, why not use all of it? No more shoot time decisions about orientation.

Think about the geometry for a moment. If you put a square inside a circle you are not using all of the lens either. In reality it is no different than putting a rectangle inside a circle -- both leave big parts of the circle untouched. The square uses a taller height at the expense of a narrower width. I really see no benefit to a square sensor. And no, not worrying about orientation at shot time is not a benefit in my opinion.

Bob Bell

Blurmore wrote:

You could do mask off the viewfinder so you were always on a sqaure. I have heard non confirmed stories about famous fashion and portrait photographers who were used to shooing 6x6 doing this to their 35mm SLRs.
Masking a 35mm negative would prevent you from getting a maximum enlargement unless the shot was dead on. But masking say a 1DSMkII would leave you with enough res to enlarge and possibly even crop for over 20 inch prints. I think this is the main drawback, the V/H format allows camera manufacturers to get the maximum sales impact of xmegapixel without expotenially increasing sensor size and cost. Imagine if sensor manufacturers had to go from say a 3cmx3cm 3000x3000 pixel sensor to a 3cmx3cm 6000 x 6000 pixel sensor to double resolution. The best way I've found to get 6000x 6000 pixel images is to scan them from my medium format positives :P

Someone is selling focus screens with etched squares, and ratios like 3:4, 4:5 etc... When I see fashion photogs on TV it looks like they are using 645 bodies with digital backs like on that model show with tyra banks, different photographer every week but the same type of camera.

mercphoto

Bob Bell wrote:

Someone is selling focus screens with etched squares, and ratios like 3:4, 4:5 etc... When I see fashion photogs on TV it looks like they are using 645 bodies with digital backs like on that model show with tyra banks, different photographer every week but the same type of camera.

There are times when I really, really would like a 4:5 indication on my focus screen. Seems like a no-brainer to me but Canon doesn't do it. Oh well.

As per the model shows, I'm sure they are using 645's with digital backs because of the very high resolution that has, not because its a square sensor.

Blurmore

Bob Bell wrote:

Someone is selling focus screens with etched squares, and ratios like 3:4, 4:5 etc... When I see fashion photogs on TV it looks like they are using 645 bodies with digital backs like on that model show with tyra banks, different photographer every week but the same type of camera.

When I said "were" I mean like 15 years ago now when Hasselblad EL series cameras were the gold standard. Most rigs I see today are Mamiya 645 AF bodies with digital backs. I still see people wielding Hasse's on TV during shoots but it is usually a chromed out 500 with chrome lenses, and conspicously set up for the behind the scenes look. I'd say that Hasse use in general in professional work is limited almost exclusively to film use. I've yet to see a H1 live and in person or on TV.

wxwax

mercphoto wrote:

Think about the geometry for a moment. If you put a square inside a circle you are not using all of the lens either. In reality it is no different than putting a rectangle inside a circle -- both leave big parts of the circle untouched. The square uses a taller height at the expense of a narrower width. I really see no benefit to a square sensor. And no, not worrying about orientation at shot time is not a benefit in my opinion.

I keep expecting Rutt's thundering reply to your post, merc. He's keeping me waiting. :pissed

:poke

wxwax

mercphoto wrote:

There are times when I really, really would like a 4:5 indication on my focus screen.

Could you help me understand why?

mercphoto

wxwax wrote:

I keep expecting Rutt's thundering reply to your post, merc.

You can't argue with geometry! A rectangle inside a circle doesn't use all the circle. But neither does a square. And lenses are circles.

As per why I want 4:5 indication on my focus screen, the simple reason is the stupid framing industry cannot figure out which aspect ratio they are in love with. I take a photo and it makes a perfect 4x6 (in other words, everything I see in the viewfinder makes it on the print). But it is too much for an 8x10. The question is, as I take that picture and I'm looking through the viewfinder and composing my shot, what part of what I am seeing in the viewfinder will fit on an 8x10 print? If I guess wrong I have a problem. So why guess?

wxwax

Isn't that what matting is for?

mercphoto

wxwax wrote:

Isn't that what matting is for?

Not at all. Matting takes one size print and puts it in a larger size frame. How does this help me when I try to crop an 8x12 native image into an 8x10 final print and not cut the heads and feet off people?

marlof

35mm is not sacred. Says this Olympus four thirds user. Where the sensor is using most of the image circle, and full frame is not related to the old 35mm format. I'm also contemplating getting the Panasonic LX1, so that'll be a whopping 16:10. I like to expand my horizons, and yes, I'd like a square sensor as well.

DoctorIt

mercphoto wrote:

Think about the geometry for a moment. If you put a square inside a circle you are not using all of the lens either. In reality it is no different than putting a rectangle inside a circle -- both leave big parts of the circle untouched.

Yes, but the square clearly maximizes the inscribed area.

(nerd-out alert)

mercphoto

DoctorIt wrote:

Yes, but the square clearly maximizes the inscribed area.

Yeah, I did the math too over lunch.

Apect Ratio and Percent of Circle Covered

---

1:1 64% (square)
3:2 59% (35mm)
5:4 63% (an 8x10 image)
4:3 61% (four-thirds)

So the sqare has a bit more covered than any of the other formats, with a 5/4 ratio being very close behind.

Still, we almost always want a rectangular image, not a square image. So shoot with a square sensor and you are going to crop it to a rectangle, so why not start with a rectangle in the first place?

Take a square image and crop out an 8x10 and you use nearly exactly the same lens area as cropping an 8x10 out of a 35mm 3:2 image. Thus little value added to a square sensor.

rutt

Whew, I was just waiting for someone else to do this. I was about to prepare a Pythagorean proof.

DoctorIt

rutt wrote:

Whew, I was just waiting for someone else to do this. I was about to prepare a Pythagorean proof.

Glad I could help. I live for nerd-outs. I was gonna do it in 3d but the boss walked in and had to cut my losses. I had him going for about a minute that the graph was actually a velocity profile for my latest test fluid. Whew.

NHBubba

wxwax wrote:

Isn't that what matting is for?

Matting gets expensive if you need it custom cut. All the pre-cut matting is to matte a print into the next-size larger frame, as merc outlined.

Although it can be done: I've found that I can print to 10x15" and get a custom cut matte to fit the print to a 16x20" frame. That works acceptably well, but increases the price both for the print and for the matte. I'm experimenting w/ cutting my own mattes now.. but so far w/o much luck.

Andy

Excellent discussion, thanks guys! I'm enjoying this read.

rutt

mercphoto wrote:

Still, we almost always want a rectangular image, not a square image. So shoot with a square sensor and you are going to crop it to a rectangle, so why not start with a rectangle in the first place?

Take a square image and crop out an 8x10 and you use nearly exactly the same lens area as cropping an 8x10 out of a 35mm 3:2 image. Thus little value added to a square sensor.

[imgr]http://rutt.smugmug.com/photos/48909551-M.jpg[/imgr]

1. I love square images.
2. I think I shoot portrait mode more than landscape.
3. Cropping a portrait orientation out of a landscape original really does start to lose some pixels.
4. I'll choose the crop at post time, please.
5. I can see why Merc doesn't want this, given his workflow.

BenA2

Fabulous Idea

Bob Bell wrote:

I know the 35mm is sacred so if you are going to really mess with it, make the sensor a circle. 36mm in diameter, that way you can orient the body any way you want and still have a decent ratio for larger prints.

Wow, I love the idea of making the sensor a circle. To me, that would be a real advancement in digital camera features. It really is ideal. Expensive? Yes. Obviously it's a bigger chip for a given diameter. But, of course, you could still have crop-factor circles for low-cost.

The beauty is, it gives you so many great options. The only challenge I see is in composing the shot. I think you would need some sort of adjustable masking system in which the user would set the aspect ratio and the viewfinder would mask accordingly (OK, I concede that could be a big challenge in SLR viewfinder design).

For shooting in RAW, the camera would still expose the whole circle, but the as-shot crop dimensions would be carried as meta-data in the RAW file. Then the crop could be kept, expanded, contracted, or rotated in RAW conversion. When shooting JPEG, you would have the option of only saving the cropped shot as specified at shoot time.

Somebody really has to role with this idea.

NHBubba

Round sensors would DEFINITELY cost quite a bit more. Almost all semiconductor manufacturing is set up for rectangular dies. Round sensors would result in wasted silicon space.. unless they chose to toss some of the support electronics into the wasted corners.. but I would think that would push yeild down too.

I think square sensors are far more likely..

DoctorIt

All square native images would definitely make my smugmug galleries look better in thumbnail view.


:hide

rutt

NHBubba wrote:

I think square sensors are far more likely..

But don't hold your breath.

Why? That was my original question. Merc came closest to answering this in spite of his suspect geometry.

mercphoto

rutt wrote:

But don't hold your breath.

Why? That was my original question. Merc came closest to answering this in spite of his suspect geometry.

Cost is going to be the biggest reason. More silicon area in the sensor is just the beginning. And forget circular sensors. The waste involved yanks the price up higher still.

You're correct Rutt that my workflow would not benefit from the ability to choose orientation in post. My workflow wants as much as possible done at capture time, including orientation of image. I'm currently uploading 1,200+ race images from yesterday. Rating, culling and keywording was enough work!

As per my suspect geometry maybe I missed something. My math was based on the assumption that most portraits will end up as a 4:5 aspect ratio. An 8x10 or 11x14, for example. As such, both the 3:2 sensor of 35mm and the square sensor you want will need to be cropped. The math shows a small advantage over cropping the square image than cropping the 3:2 image. I'm also assuming the 3:2 image is cropped the long way, not the short way, which may not be valid given your shooting preference.

As per why I want a focus screen that indicates the 4:5 aspect ratio I'm not sure how my comment started a side-discussion about matting, because matting doesn't address my issue at all. Maybe this will explain better what I was wanting and why:
http://www.robgalbraith.com/bins/content_page.asp?cid=7-7890-8180