Sorry if this has already been discussed. I did a search on your forums and couldn't find an answer.

I am thinking of moving into MF and so I've been trying to understand the technology a little better before making a decision. In my reading, the subject of pixel pitch seems to come up quite a bit. One of the arguments in favor of MF is that the larger sensors allow for larger pixels and those larger pixels give MF an advantage over DSLRs. So I was curious to see what the published differences were. For example, I found the following:

Canon EOS-1Ds Mark III - pixel pitch 7.2 um
Hasselbald H3DII-39 - pixel pitch 6.8 um

Maybe I am not understanding this metric. I looked into an explanation of it and found Wikipedia - Pixel Pitch. This article seems to show a geometric relationship between pixels on the sensor and this relationship is what defines which defines the pixel pitch. So it isn't exactly the dimension of a pixel which is being referred to when pixel pitch is used.

So I am left confused about why the Hasselbald has a lower value for pixel pitch than the Canon. Wouldn't that give the Canon a better image quality with all other factors being the same (which they aren't).

I think the CCD has less electronics on the sensor (microlenses, etc), so the receiving area of the pixel becomes greater. Therefore the CCD can produce stronger signal compared to the CMOS sensor.

I believe they are not the same metric.
One is talking about the rgb triad of pixels a la wiki article the other is talking of 1 pixel...
I have never seen the MF Back makers talk about pitch but single pixel size in microns.

I just read a technical article from Hasselbald's site which explained some of the differences between 35mm systems and MF systems. The article (titled "The bigger the better") confirms what I found concerning pixel size. It states that the pixel size of MF is not larger than that of 35mm systems. The better image qualities of the MF is due to the increased sensor size and the affects this has on the camera/lens optics.

I've read in several threads on this site statements such as "since the photo sites are larger ..." when discussing advantages of MF. From what I have found I don't think "larger photo sites" is correct. There are other reasons for the MF's better images.

Here is a quote from the Hasselbald article:

"With a pixel size of 6.8 μm (H3DII-39 and -31) or 6.0 μm (H3DII-50), resolution hits the Nyquist limit at 74 or 84 line pairs per millimeter, respectively. The pixel size of a 35 mm DSLR is similar, 6.4 μm being a typical size, so the limits imposed by the sensor are similar as well: 78 lp/mm is the maximum spatial frequency that a typical 35 mm sensor with 21 megapixels can resolve."

I am in no way trying to be negative or argumentative. I just want to understand the technology.

Here are a couple examples, first at full frame.

Click to view attachment

Canon 5D

Click to view attachment

Hasselblad H3DII 22

Click to view attachment

Canon 5D 100% crop

Click to view attachment

Hasselblad H3DII 22, not quite 100%

Most statements about pixel pitch are wrong (except this one ).



That's one example of an incorrect statement about pixel pitch. Larger pixels are only better if they have higher full well capacity, less read noise, or higher QE after scaling for spatial frequency. In the case of upgrading from 35mm DSLR to MF, none of those are the case (in fact the DSLR pixels are better by a wide margin). Even when upgrading from a digicam to DSLR, the pixel size imparts no advantage except at high ISO. (The sensor size, on the other hand, makes the difference). Similarly, the MF advantage lies elsewhere.



No, it's *sensor* size, not pixel size, that is the most important factor.

Do you mean something other than pixel count?

Cheers,
Bernard

He probably means sensor area, the larger the surface area of the sensor the more photons captured in less time equals higher image quality. So even though pixel pitch/size and quality change my P45+ has higher image quality than my 5DII which has higher image quality than my G10 which is better than my cell phone.
Marc

Those numbers are slightly wrong and out of date. The Nikon D3X has cell size (a.k. "pixel pitch" or "photosite spacing") of 6 microns, the Canon 1DsMKIII is only slightly larger, about 6.4 micron, the newest DMF sensors of 50MP and up have 6 micron cell size, the previous generation MF sensors have 6.8 (Kodak) and 7.2 micron (Dalsa) spacing. The older 22MP 48x36mm Dalsa sensor still is use has 9 micron cell size.

But the pattern is indeed that DMF offers cells (photosites) of roughly the same size, but more of them. And as others will explain, using a larger sensor with more photosites of about the same size can improve IQ (including reducing the effects of noise as seen on equal sized prints) even if the individual photosites have about equal performance on lab. tests.

And maybe increased resolution is still one major reason for using a larger format! Higher dynamic range and lower noise at high ISO are not the only measures of improved image quality, despite the current strong emphasis on them in forum discussions.

Lens coating > Lens elements > Aperture blade design > internal body coating > microlens > Anti aliasing filter (found in 35mm but not in medium format) > IR filter > sensor photo well > sensor read-out (heat-sinking and/or active cooling very important here) > cables to A/D converter > A/D converter > (read-out of black calibration file from sensor recorded as adjunct to the image) > debayering algorithm, deconvolution / detail finding algorithm, noise reduction based on black calibration file > noise reduction based on image data > sharpening.

Yes. If ALL other factors are the same than a photo site with larger sensitive surface will have higher quality. However as you can see there are a LOT of other factors.

As a real world example from the P25/P20/P21 with 9 micron sensors to the P40+/P65+ with 6 micron sensors the dynamic range, shadow color accuracy, tonal gradation smoothness, and maximum possible detail at 100% have gone UP rather than down. This is because every other area of the image-quality chain has been improved.

Doug Peterson (e-mail Me)
__________________
Head of Technical Services, Capture Integration
Phase One, Leaf, Leica, Canon, Apple, Profoto, Eizo & More
National: 877.217.9870  |  Cell: 740.707.2183
Newsletter: Read Latest or Sign Up
RSS Feed: Subscribe

Well said!

The correct word for "debayer" is "demosaic" I think ;-).

Erik

True.

Sensor area in itself does not translate directly into image quality. The factors impacting image quality are the usual suspects: DR, type of noise, sharpness per pixel after suitable capture sharpening and the other factors mentioned by Doug below... all of these qualities can be measured at pixel level until proven otherwise.

It seems therefore reasonnable to define image quality as the combination of pixel quality and a number of pixels.

Since the poster claims that he does not consider MFDB individual pixels to be superior, the logical conclusion appears that the only value he sees in MF is more pixels.

I was just trying to confirm whether he was in agreement with my interpretation or not.

Cheers,
Bernard

In a "relative" sense, it does. All else being equal, the larger the sensor, the better the image quality. The same can't be said for pixel size. Sensor size matters more than any other single factor (except perhaps lens).



Yes, they are measured at the pixel level.



Agreed.



I think of it a different way. Larger formats can be superior even if they have *smaller* pixels than smaller sensors. It's not the pixel size that matters, it's the total overall performance.

It's like an engine. If people talked about engines the way they do about cameras, they would concern over the "horsepower per cylinder", biasing themselves to engines that have a small number of cylinders. The more important information is the total horsepower of the engine, which, as you said, is HP/cyl multiplied by the total number of cylinders.

Bernard, if by poster you are referring to me, the let me respond. I don't know if MFDB pixels are superior or not. That is part of the reason for the initial posting ... to investigate this question. So I am not in a position to state that the only value in MF is more pixels. In fact, from postings in this thread and other reading that I've been doing, I believe there are many factors contributing to image quality, one of which is the number of pixels.

An article that I found on the Hasselbald site addresses the questions raised by this thread. A couple of quotes are as follows:



Summary points:



The full article read by clicking on the link The bigger the better

I appreciate the posting in this thread and have learned from them. Thanks.

Hi,

Thanks for the pointer to the article from "Victor". A very major effect is the difference in the lens. The 35 mm lens is simply not really good, while the Hasselblad lens being quite good. Also, the crops here are at or close to actual pixels from the MFDB so the DSLR images are "uprezzed" to match. This is not cheating, no way you can argument with more pixels.

One interesting point is that there is a lot of aliasing in the hair of the model in the MFDB images. If you look somewhat carefully you see discontinuous strains, staircase effects, etc. This shows that the lens outresolves the sensor even at full aperture.

One issue I have with this comparison is that it's done near full aperture. It puts the DSLR at extra disadvantage because it has a than very good lens. Focusing errors are also aggravated by using maximum or close to maximum aperture, and the subject doesn't really call for small depth of field. This subject could have been shot at f/8 and the difference would be smaller. Or they could use a better lens.

This points a bit in the direction that DSLRs need better lenses.

Best regards
Erik