Talk:Image sensor format

semi-log chart
Under the sensor area chart; "The x axis is a discrete set of sensor format sizes used in digital cameras, not a linear measurement axis".

Just a technicality, but the dots shouldn't be joined together to make a line in this case. A column chart would be more appropriate. — Preceding unsigned comment added by 203.129.23.146 (talk) 23:58, 9 March 2013 (UTC)

Totally agree. Noticed the same thing and stopped by to point it out. Apparently it was generated with gnuplot, the relevant code is on the photo's Wikimedia Commons page. If someone is capable with gnuplot and wants to remake it as a bar graph, that would be ideal. Rjbranch (talk) 22:49, 14 August 2020 (UTC)

@Gsarwa: Unscientific source wrongly presented and badly written
No point is exactly true, as you write it. And your style: Is it a different joke? Comments: See all discussions above: This article is about image sensors and therefore mostly image noise, not pixel values. The same area in small pixels can be more effective then in large ones. 77.12.172.81 (talk) 13:42, 15 July 2013 (UTC)
 * IMAGES are taken by photographers, not pixels: So partly wrong, and even more: too complicated written.
 * Again, factually wrong, use talk page to provide evidence for each point
 * Remember: Those who add text must provide evidence: Your text is not covered by the quite bad source.

I insert here the disputed section:

Image sensor format, sensor resolution and pixel size
Different sensor format with different image sensor resolution will give different pixel density with different pixel size and without considering the lens and the processor will give different image quality, because of: So, the important thing to produce good image quality is not the sensor size (image sensor format) or pixel density, but the pixel size. Bigger image sensor format is possible to give more Mega Pixel with bigger pixel size.
 * More bigger pixel size, more data can be saved in a single pixel, so 2 microns pixel pitch is inferior than 6-8 microns pixel pitch.
 * A close distance between two pixel can cause interference between the two pixel and make reducing overall image contrast, a sufficient distance between two pixel which is separated with a septum can avoid the problem.
 * In a small pixel size some camera makers frequently boost the light by increasing the ISO and make digital noise, more ISO will make more noise.

I think we could help User:Gsarwa turn this into English if the source were decent, but it's an amateurish mess, as the IP points out, and further badly interpreted by G. Yes, bigger sensors and bigger pixels have advantages, and the current algebraic treatment of that is rather difficult to appreciate, even for an expert. So if someone gets motivated and wants to find better sources and write something about this, that would be useful. Dicklyon (talk) 15:40, 15 July 2013 (UTC)
 * Thanks. By the way, the section "Sensor format and lens size" isn't very good, either.
 * Probably combine both with the attractive "noise" word somewhere in the section-title, a "Summary" or "Overview" of the whole, quite scientific article on top of all sections? 77.12.172.81 (talk) 16:41, 15 July 2013 (UTC)

Incorrect sensor sizes for Nokia 808 Pureview and Nokia Lumia 1020
According to Nokia 808 Pureview white paper, the sensor has a total number of pixels of 7728 x 5368 and the pixel size is 1.4 microns. This translates into sensor size of 10.82 x 7.52 mm. Aspect ratio is 1.44 and not 4:3 (1.33).

According to Nokia Lumia 1020 white paper, the sensor has a total number of pixels of 7712 x 5360 and the pixel size is 1.12 microns. This translates into sensor size of 8.64 x 6 mm. Aspect ratio is again 1.44 and not 4:3 (1.33).

Diagonals, area, crop factors can be found here.90.157.178.15 (talk) 13:47, 17 July 2013 (UTC)

Max8
I'd like to propose Max8 aka "Super-Duper8" to be added to the table that not only in the past has been used for a number of shorts and feature films, but even a Super8 camera (including XLR input, C-mount for interchangeable lenses, as well as realtime video out, like on a professional 35mm film camera) newly designed from scratch to come from an as-of-yet unnamed company in Norway will be working with Max8 by default, whereupon the aspect ratio with Max8 is changed from 4:3 to 1.59:1 (which is only 1% away from 16:10).

On that note, especially regarding the other figures required for the table here I'd like to point out a discrepancy: In this English-language image, the exact image dimensions of Super8 are given as 4.01mm × 5.79mm, however the German Wikipedia gives image area dimensions of 4.22 mm × 5.69 mm instead. The weird thing is, neither dimensions work with the 4:3 ratio.

In any case, the exact width, diagonal, area, and crop factor of Max8 could be derived from the exact figures for Super8 (whatever they are) and the aspect ratio of 1.59:1. I'm not sure what you mean by "Stops (Area)" though. Does a smaller area require more light, or something? With that field obviously related to 35mm with 0, it seems similar to a field on a similar sensor size table at the German article for Crop factor that says how much percentage of image area the particular format has in comparison to 35mm. --93.232.167.181 (talk) 11:35, 22 August 2013 (UTC)

Pixel counts
It would be interesting to include pixel counts (e.g. 1024 x 1024) for the various formats listed in charts and mentioned in the prose. (That's what I was looking for when I came to this article.) -- Beland (talk) 18:41, 27 December 2013 (UTC)
 * Pixel count and format are largely independent, as pixels come in a wide range of sizes. Dicklyon (talk) 03:49, 28 December 2013 (UTC)

Why is a 'one inch' sensor called that?
... the dimensions (and area) are much less than an inch (or square inch)? Apologies if I missed it in this rather scattered article! quota (talk) 14:41, 21 June 2014 (UTC)


 * It's inherited from video camera formats. It's the size of the active imaging area that would fit on a vidicon or plumbicon tube whose glass envelope has an outer diameter of 1 inch.  The diagonal is about 2/3 of the indicated dimension.  This explains 1-inch and half-inch formats, but why did they genalize to things like 1/2.5 instead of 0.4-inch?  Probably because marketing like bigger numbers, even for smaller formats.  Dicklyon (talk) 02:26, 22 June 2014 (UTC)
 * It's mentioned in the first paragraph of the section Image_sensor_format. Perhaps it deserves its own section and better sourcing.  Here is a decent source on video tube formats.  Dicklyon (talk) 02:33, 22 June 2014 (UTC)

Depth of field
There should be reference to the multiple form factors in film and that some concepts that were developed with film now dictate how those concepts are to be used in sensor photography.

Someone needs to review the section on sensor size and depth of field as light sensitivity is not dependent on sensor size and the article states that it does, rewarding bad science with credibility.

RE: Depth of Field 1) IT IS PLACED HIGH IN THE ARTICLE IN A POSITION OF PRIORITY. This lends it credibility not deserved. It also gives more importance to depth of field than exposure.  At the very least it should not imply differences in sensitivity and DOF based on sensor format, but attribute the differences to crop factored apertures.    Ideally it should emphasize that the final image view is based on a cropped image from a larger format with the actual focal length and not some crop-adjusted length. 2) The fact that fstop is not a true measure of lens opening but only a ratio of focal length to opening is not properly covered. 3) It is highly questionable scientifically, presenting an opinion not supported by fact that iso ratings vary by sensor size.   That would be similar to claiming that if film were used that the same film should have a different rating with a different film format.   Is 8x10 panatomic-x (using classic films) more sensitive to light than half-frame of the same film?   It also would imply changes in S/N that do not occur as would be expected using those numbers. 4) Any mention of fstop rating, when considering depth of field, should be based on the true focal length and not a crop factor adjusted length since the lens length does not ever actually change. 5) Table of sensor formats and sizes in this article has a column labelled stops. This column repeats the same fallacy and again implies that 8x10 film is more light sensitive than smaller format film.

Any statement that is made should be evaluated as different film formats rather than different film sizes. If it isn't valid for film, it isn't valid for sensors. The specific numbers may change, but the concepts must remain valid.

My writing skills are not sufficient to re-edit this section, but drastic re-editing is required.

Namssurt (talk) 16:57, 20 July 2014 (UTC)namssurt

The Call To Action: Re-write for Coherency
I just felt compelled to make some edits. It's evident that much of this (mongrelised) page should be rewritten for coherency. I guess I'll be back with an actual account at some stage if no one takes this up. 121.45.207.17 (talk) 11:39, 27 November 2015 (UTC)

Type 1/4 sensor wrong
Should be: Size: 1/4", Diag: 4.5mm, Height: 3.6mm, Width: 2.7mm

See and also Google the specs for the Sony ICX098BQ. — Preceding unsigned comment added by G01d4 (talk • contribs) 20:10, 20 September 2016 (UTC)

What about other film sizes?
The table includes several film sizes for reference, naturally everyone use 135 also to compare the equivalent focal length. The 127 film size was used in cameras which take squared pictures. The 126 film cartridge was also very popular. Both formats where used in point and shot instant cameras whose users usually ignored the focal length, shutter speed and f/aperture. Also some movie film sizes are ignored like the 8mm and Super 8mm formats. Those are important references for movies.

Apart of citing such films for reference, it would be important to mention the equivalent pixels of different ISO sensitives, to take into account for example, how a Kodachrome Super 8 film compares with a 1080p video recorded with an 1/3" sensor. — Preceding unsigned comment added by 189.233.103.27 (talk) 06:43, 26 September 2016 (UTC)

Max8
For almost two decades, camera repair shops around the world have been offering the service of Max8 modification of existing Super8 cameras, for about a decade, Pro8mm has been selling Max8-modified cameras, since 2015, Logmar are selling Super8 cameras that were designed for Max8 from scratch, and this spring, Kodak will start shipping a new Super8 camera designed for and natively shooting in Max8. As the figures and dimensions significantly deviate from that of traditional Super8, it should really be about time to make an own spot for it in our table here at the section of Table of sensor formats and sizes.

The image height is the same as with traditional Super8, but it's wider so that the image is not 4:3 but 16:10. The information is especially notable because Kodak will be selling their new camera with an interchangeable C-mount, so a spot in this table would be needed to know what lenses are compatible and what angle of view one can expect. Most likely, most tradtional Super8 lens attachement converters won't be compatible anymore because their image circle is too small for Max8. --2003:71:4E6A:B409:F0C2:BF33:943F:4CCD (talk) 21:31, 23 January 2017 (UTC)

Initial Statements
“In digital photography, the image sensor format is the shape and size of the image sensor. The image sensor format of a digital camera determines the angle of view of a particular lens when used with a particular camera.”    A “format” is just a shape to me, a relation like 4:3 or 24/36. The actual size is something else. – And the sensor format, e.g. its proportion, doesn’t determine the angle of view. Just zoom. It’s the optics. And what do you mean by: “a particular lens when used with a particular camera”? A severabiliy clause? I’d say: “The format of the sensor determines – together with the shape (format) of its pixels – the native output format of a digital picture, using all its sensor points (pixels).” — Fritz Jörn (talk) 18:22, 29 January 2017 (UTC)
 * The aspect ratio of a rectangular image is an important thing, and perhaps "image format" ought to be used as a synonym for it. However, the terms are not actually used that way in the business, and I don't think it's the job of Wikipedia to make the outside world choose another term to mean what "image format" and "image sensor format" mean now. Jim.henderson (talk) 13:56, 2 February 2017 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 6 external links on Image sensor format. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20120513162446/http://www.zeiss.com/C12578620052CA69/0/58D501E36518AFC9C12578D2004104E1/%24file/cln_39_en_tessar.pdf to http://www.zeiss.com/C12578620052CA69/0/58D501E36518AFC9C12578D2004104E1/%24file/cln_39_en_tessar.pdf
 * Added archive https://web.archive.org/web/20150113004959/http://www.imageval.com/public/Papers/EI%205678-01%20Peter%20Catrysse.pdf to http://www.imageval.com/public/Papers/EI%205678-01%20Peter%20Catrysse.pdf
 * Added archive https://web.archive.org/web/20100812233952/http://www.phaseone.com/en/Digital-Backs/P65/~/media/Phase%20One/Products/Documents/Phase-One-digital-back-overview.ashx to http://www.phaseone.com/en/Digital-Backs/P65/~/media/Phase%20One/Products/Documents/Phase-One-digital-back-overview.ashx
 * Added archive https://web.archive.org/web/20121028001406/http://www.robgalbraith.com/public_files/Canon_Full-Frame_CMOS_White_Paper.pdf to http://www.robgalbraith.com/public_files/Canon_Full-Frame_CMOS_White_Paper.pdf
 * Added tag to http://www.spotimaging.com/resources/glossary/image-sensor-format.php
 * Added archive https://web.archive.org/web/20130125090640/http://www.dpreview.com/glossary/camera-system/sensor-sizes to http://www.dpreview.com/glossary/camera-system/sensor-sizes
 * Added archive https://web.archive.org/web/20120119113015/http://www.21stcenturyshoebox.com/essays/formatindependence/ to http://www.21stcenturyshoebox.com/essays/formatindependence/

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 14:46, 9 April 2017 (UTC)

Dynamic Range
I went through the calculations for the read noise and dynamic range section and it took me a while, but I finally found it to all be correct.

However, it is very non-intuitive as it suggest that smaller sensors have much better dynamic range. All the while I have thought that shot noise and dynamic range are not affected by sensor size given a DOF-constrained situation and my intuition is confirmed by DxOMark's data.

From experimental observations (just look at DxOMark's data), given similar sensor technology (eg. Nikon D750 and Nikon D5500):

1. Doubling the sensor size roughly increases the dynamic range at ISO 100 by one stop. Thus, for a fixed exposure, the dynamic range appears proportional to the sensor size.

2. Doubling the sensor size and doubling the ISO at the same time roughly gives the same dynamic range. Thus, for the DOF-constrained situation, (i.e. at ISO_large = ISO_small × CF²), it appears that dynamic range is constant.

So why does theory deviate from experimental data so wildly? And even if the theory is correct, why does sensor/camera manufacturers seem to equate larger sensor as being better? It seems weird that camera manufacturers transitioned to full frame sensors when in theory, crop frame sensors should have a better dynamic range (ignore legacy lenses for the sake of discussion). Am I missing something?

(Also, let's not quote photography sites on sensor technology, most of them are photographers, not engineers and have no idea what they are talking about. eg. DPReview, apparently a trusted photography site, took a long time to understand the concept of equivalent fstop)

EDIT: I read up more on the topic and it appears that experimentally, read noise is not proportional to pixel pitch (and given that the number of pixels is constant, read noise is not proportional to sensor size). If this is true, then my initial intuition is correct and the assumption made in the calculations is wrong.

LeongQiJin (talk) 03:55, 5 June 2017 (UTC)


 * To first order, read noise should be independent of pixel size and sensor format if you have a good kTC-noise-cancelling pixel and readout architecture. If that's the case, then the sensitivity limit is independent (to first order) of the format size.  See my "outside-the-box" approach for help understanding this.  So that's the floor, the low end of the dynamic range, which is what you usually care most about.  But the dynamic range also considers the high end, limited by the "full well" capacity of the pixel (measured in electrons, typically), and that's generally higher for larger pixels.  So bigger sensors tend to have more dynamic range, and this can result in way less shadow noise if it's employed via shooting at a much lower ISO setting.  Larger formats necessarily work at lower ISO settings anyway, even when set to get the same number of photons per pixel, same DOF, same exposure time, as the outside-the-box analysis should make clear; the extra high-end dynamic range lets you go further toward lower ISO when you want to take advantage of high dynamic range. Dicklyon (talk) 23:16, 14 August 2020 (UTC)
 * The discussion in the article is flawed in assuming "amount of light per unit area" is constant. With a bigger format, the light is spread over a larger area.  I'll work on that (a few years late, sorry). Dicklyon (talk) 23:20, 14 August 2020 (UTC)
 * I just took that out for now. If someone wants to work on it with me, that would be cool.  Dicklyon (talk) 06:14, 15 August 2020 (UTC)