Illustration: Multi-Aspect Four Thirds Versus Standard Four Thirds

I've gotten some emails expressing confusion regarding the difference between the multi-aspect ratio Four Thirds sensor in the Panasonic GH1 and the standard Four Thirds sensor found in all other Micro Four Thirds cameras. This illustration ought to clear it up.

On the left (first column), you can see the relative sizes of the GH multi-aspect ratio sensor (top) and the smaller standard Four Thirds sensor (bottom), each represented by white boxes.

In the second column, one can see that the portion of the GH1 sensor used for 4:3 aspect ratio (top) is the same size as the entire standard Four Thirds sensor (bottom). Thus the image captured will be the same when in 4:3 mode, regardless of whether one uses a standard Four Thirds or multi-aspect ratio sensor camera.

In the third and fourth columns, 3:2 and 16:9 respectively, the situation changes. Here, the GH1 uses the larger sensor to keep the lens diagonal angle of view unchanged regardless of aspect-ratio. In contrast, the standard Four Thirds sensor simply crops out a portion of the 4:3 capture to make the 3:2 or 16:9 image. As a result, the diagonal angle of view is less wide, and the megapixel count drops to a greater extent.


Click for larger version

Posted by Amin

Comments (23)

Loading... Logging you in...
  • Logged in as
Amin,

How does the GF1 fit in here? Is the sensor similar to the GH1?
2 replies · active 809 weeks ago
Hi Chuck,

The GF1 uses a standard Four Thirds sensor like the G1 and E-P1. I don't know why only the GH1 has the multi-aspect sensor so far. I think it would be a great thing to implement in all Micro Four Thirds cameras going forwards.
agreed, as I'm not a fan of the 4/3s aspect ratio =) But i suspect the GF1 didn't get it because of cost and/or size. How much more space does that slightly bigger sensor take?
Hard to tell's avatar

Hard to tell · 809 weeks ago

Linh:

From DPReview:
GH1: 4/3 " (18.00 x 13.50 mm, 2.43 cm²)
G1: 4/3 " (18.00 x 13.50 mm, 2.43 cm²)

From Panasonic: G1 GH1
Image Sensor Size17.3 x 13.0mm17.3 x 13.0mm
http://www2.panasonic.com/webapp/wcs/stores/servl...
1 reply · active 809 weeks ago
If you look further on the Panasonic pages, it's pretty clear. They clearly state that the GH1 has 14MP total and the GF1 and G1 have 13.1 total. They list multi-aspect ratio as a feature of the GH1 and have an illustration of how the GH1 makes use of the imaging circle at various aspect ratios, while they don't list it as a feature of the GF1 or G1.
by your picture of sensor sizes i would say that GH1 has got same sensor size as 4/3 but uses smaller portion of it so i think that 4/3 version is better
1 reply · active 809 weeks ago
No, that is not the case. As shown by the white boxes on the left, the GH1 has a larger sensor size than other 4/3 cameras. At 4:3 aspect ratio, the portion of the GH1 sensor used is the same size as the entire standard 4/3 sensor. At 3:2 or 16:9, the portion of the GH1 sensor used is larger than the portion of the standard 4/3 sensor used at those aspect ratios.
Amin,

What are the dimensions of the sensors? The white boxes don't have any dimensions. And what is the source for the size of the white boxes? Panasonic and dpreview say the sensors are the same size for the G1 and GH1, even if they don't agree on the dimensions. My original link may have had an error http://www2.panasonic.com/webapp/wcs/stores/servl...

And more pixels doesn't necessarily make the sensor bigger.
1 reply · active 809 weeks ago
The white boxes show the relative sizes, not the actual sizes, which I don't know. Panasonic says that the diagonal angle of view does not change between the 4:3, 3:2, and 16:9 aspect ratios when using the GH1, whereas it certainly does change when using any of the other cameras. This can be deduced from the fact that the horizontal pixel count does not increase when switching to 3:2 or 16:9 on a GF1 or G1. Meanwhile, the imaging circle is the same regardless of which Micro Four Thirds camera on which one mounts the lens. From that information alone, one can produce the representation I presented.

Basically it comes down to this: The difference is zero if you shoot 4:3, fairly minimal if you shoot 3:2, and starts to become more significant at 16:9. At least that is what I've noticed in use and I think is represented as such if you look at the figure shown above.
I decided to work out the numbers. if 4/3 is 18.0 x 13.5 mm and the same pixel density, then to get the 16/9 using Panasonic pixel count (their count gives a slightly smaller diagonal—about 1 percent), the sensor would be 19.6 x 13.5 mm with the corners cut off. So the published sizes don't make sense since the sensor should be this ratio for the GH1 pixel counts. (The pixel count for 16/9 is stated to be 4352x2448 and for 4/3 4000x3000.)

A drawing shows the situation a bit better here: http://www.dpreview.com/reviews/panasonicdmcgh1/. Although I think the rotating gif on the bottom of the page is wrong.

I should add that all of this isn't terribly important, the results are what counts. And as you originally pointed out the GH1 doesn't crop from a 4/3 sensor, but uses others parts of a sensor with more pixels resulting in more pixels in the 3:2 and 16:9 formats than those formats in the other u4/3 cameras.
1 reply · active 808 weeks ago
robotslave's avatar

robotslave · 808 weeks ago

The numbers, worked:

Start with a diagonal of 21.63mm, as per the m4:3 the white paper. This gives us:

4:3 - 17.3 X 12.98mm
3:2 - 18.0 X 12.0mm
16:9 - 18.85 X 10.6mm

That's effective imaging area, actual sensor size will be a bit bigger, and they're doing the sensible thing and just using a larger rectangular sensor, not chopping off the corners of the wafer (http://www.imaging-resource.com/PRODS/DMCGH1/ZSEN... so it's in the ballpark of 19.5 X 13.5mm, or 8 or 9 percent larger than the usual 4:3 sensor.

Imaging Resource almost correctly calls it an "oversize sensor" instead of "multi-aspect sensor"; the accurate, though much more cumbersome term would be "oversize sensor that can safely have defects in a few small regions in the corners".
How does it do with square photos?
4 replies · active 809 weeks ago
Square photos are reported to be 2992 x 2992 (1:1) which would seem to be a narrower field of view.
Forget the narrower field of view comment.

Seems more likely that 3000x3000 pixels are saved to 2992 x 2992, but I have no idea how all of this really works. But downsizing images is common in most digital cameras. But I don't see why it wouldn't save 3000 x 3000, it's a smaller file than the other formats. Time to learn more how all this works.
I was hoping that the image would be large enough to use the entire circle of coverage from the lens. I looks like the GH-1 just crops the 4/3 image to make it square. So, it doesn't offer any advantage over other 4/3 cameras.
Chez Wimpy's avatar

Chez Wimpy · 809 weeks ago

Right, for 1:1 it is no better. In fact this is the only AR (to date) that would benefit from a "taller" sensor... but I suppose the number of 1:1 shooters is so small that Panasonic couldn't be bothered with the engineering overhead (non-trivial since the sensor suddenly becomes larger in a second dimension). My friend with the GH1 just set his at 3:2, and has left it alone. I would probably have done the same (had the GF1 not been stuck with the G1 sensor!).
By the way, does anyone know how come the Fuji F30's 3:2 mode outputs 3024 horizontal pixels, while in the 4:3 mode it produces 2848 - with the horiz. field of view staying exactly the same between the two modes?
It seems not plausible to me that it would have a multi-aspect sensor, and instead of using the greater FOV of the 3:2 mode, it would be compensating by a tiny automatic "zoom-in"...
1 reply · active 809 weeks ago
... the best would be to check the raw file from Fuji S6500fd which uses the same sensor :)
improbable's avatar

improbable · 809 weeks ago

Are there any tools yet to access the rest of this data? As far as I know the raw files are the same size in all formats, so all the pixels are getting included.

Most obviously I'd like to be able to change aspect ratio after the fact.

But also, when I get the horizon skew, I'd like to be able to rotate the picture. In 2x3 this should be possible, within the image circle, with no change in angle of view; in other formats it will still be better than cropping within the crop...

I have much more hope of seeing the first than the second, as presumably this should just involve some minor hack of the files. And I'm sure I'll hear about it here when this becomes possible!
1 reply · active 808 weeks ago
improbable's avatar

improbable · 808 weeks ago

Sorry I think I spoke too soon -- the raw files are in fact different sizes for the different formats. (At least on the LX3.) So it seems only the relevant pixels are being recorded at the moment of taking.

This is a pity, I'd like to have raw files with everything. But the camera would have to save more data from each picture, and perhaps that is why they didn't do it that way.
robotslave's avatar

robotslave · 808 weeks ago

The GH1 has the multi-aspect sensor because it was designed for video, and video means 16:9

The multi-aspect sensor is not and will not be used in still-only m4/3 cameras because:

1) it is significantly more expensive than the 4:3 sensor. It has more surface area and almost certainly higher defect rates than a normal 4:3 sensor. If Panasonic ever release body-only versions of the g1 and gh1, there will be a price difference large enough to deter those who don't absolutely require video, due almost entirely to the higher cost of the multi-aspect sensor. Cost, however, is a less important reason than:

2) m4/3 still lenses are and will continue to be designed for a 4:3 aspect ratio sensor. Defects like vignetting and soft corners will affect more of the frame in the wider aspect ratio, and may show up where they were mostly absent on a 4:3 sensor, even in high-end lenses.

The 4:3 and m4:3 systems were designed as just that - *systems*. You can't change a key parameter of the system and expect everything else to work as well as before without re-engineering every component that was designed, built, and tested to that parameter.
2 replies · active 797 weeks ago
robotslave's avatar

robotslave · 808 weeks ago

I take that back about defects showing up where none were before, I don't know what I was thinking. Defects will be more noticeable in the wider aspect ratios, though.
As shown in the figure, defects will not be more noticeable at 16:9 on a GH1 than at 4:3. Either way, the extreme corners are placed the same distance from the center of the imaging circle. On a G1/E-P1/GF1/E-P2, defects will be less noticeable at 3:2 and 16:9 than at 4:3 because the wider modes crop out the extreme corners in those modes.

Post a new comment

Comments by

 
Copyright 2007 | Andreas08v2 by GeckoandFly and TemplatesForYou | Design by Andreas Viklund
TFY Burajiru