Sensor size question

Sanglier86 · Apr 16, 2024

Hi, I'm trying to find out the exact size in mm of the Samsung ISOCELL HM2 108mp CMOS sensor. I've only found the fractional size online (1/1.52)

It seems you can use this size to calculate the dimensions using the equation from this page.

CMOS Sensor Size Comparison Guide and Lens Calculator

Wondering about digital camera (CMOS) sensor sizes? This article provides information about camera sensor format sizes, general tradeoffs, and lens equivalency.

commonlands.com

The thing I'm struggling with is how do I use the fraction in the equation?

Many thanks

Mako2011 · Apr 16, 2024

Sanglier86 said:
Hi, I'm trying to find out the exact size in mm of the Samsung ISOCELL HM2 108mp CMOS sensor. I've only found the fractional size online (1/1.52)

1/1.52” = 24.4mm by 38.6mm on the long side

Sanglier86 · Apr 16, 2024

That seems a bit too big, how did you arrive at that size?

Mako2011 · Apr 16, 2024

Sanglier86 said:
That seems a bit too big, how did you arrive at that size?

Their website specifies the size in inches so easy to convert. It's also not much larger than standard 8k video camera sensor size

--
My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Quarkcharmed · Apr 16, 2024

Sanglier86 said:
Hi, I'm trying to find out the exact size in mm of the Samsung ISOCELL HM2 108mp CMOS sensor. I've only found the fractional size online (1/1.52)

It seems you can use this size to calculate the dimensions using the equation from this page.

https://commonlands.com/blogs/technical/cmos-sensor-size

The thing I'm struggling with is how do I use the fraction in the equation?

Many thanks

The specs from https://semiconductor.samsung.com/image-sensor/mobile-image-sensor/isocell-hm2 say it's 12000 x 9000 pixles, 0.7nm pixel pitch.

it's 8.4 x 6.3 mm, diagonal 10.5 mm

Now we can check that with the formula from your link that gives the diagonal in mm.

( 1 / 1.52 * 25.4 ) / 1.5875 = 10.526 mm

--
https://www.instagram.com/quarkcharmed/
https://500px.com/quarkcharmed

tbcass · Apr 16, 2024

Mako2011 said:
Sanglier86 said:

Hi, I'm trying to find out the exact size in mm of the Samsung ISOCELL HM2 108mp CMOS sensor. I've only found the fractional size online (1/1.52)

Click to expand...

1/1.52” = 24.4mm by 38.6mm on the long side

That can't be right because Full Frame is 36x24mm. 1/1.7 is 7.6x5.7mm. 1" sensors are 13.2x8.8 so 1/1.52 would be somewhere in between. Looking at the ratio tells me that the sensor is 2/3" or 8.6x6.6.

SensorSizes - Full-frame DSLR - Wikipedia

Bob A L · Apr 16, 2024

Mako2011 said:
Sanglier86 said:

Hi, I'm trying to find out the exact size in mm of the Samsung ISOCELL HM2 108mp CMOS sensor. I've only found the fractional size online (1/1.52)

Click to expand...

1/1.52” = 24.4mm by 38.6mm on the long side

I'll have to look it up, but that is not right I'm pretty sure. That is bigger than FF if that's what it is. Phone companies try their best to deceive folks on the size of sensors.

tbcass · Apr 16, 2024

Bob A L said:
Mako2011 said:

Sanglier86 said:

Hi, I'm trying to find out the exact size in mm of the Samsung ISOCELL HM2 108mp CMOS sensor. I've only found the fractional size online (1/1.52)

Click to expand...

1/1.52” = 24.4mm by 38.6mm on the long side

Click to expand...

I'll have to look it up, but that is not right I'm pretty sure. That is bigger than FF if that's what it is. Phone companies try their best to deceive folks on the size of sensors.

Re: mm: Open Talk Forum: Digital Photography Review (dpreview.com)

DanDim81 · Apr 16, 2024

On a web-site as important as this (I've seen several other less interesting ones) there should be a summary of all the dimensions of photographic sensors and smartphones!

Otherwise the administrators and users of this web-site (given the strong authority in this area) could request all the manufacturers of these photographic products and cameras to show at least "Basic" data (as well as the "real size of the sensor" ).

Very often superficial data is provided to those who purchase these products (and the most important data are ignored), even on higher range products. I believe it is a right for those who work in the sector or purchase these products, what do you think about it?

Richard Butler · Apr 16, 2024

That's definitely wrong, in that that's larger than "full-frame."

Full-frame is Type 2.7" in the mad sensor industry terminology, for what it's worth.

So far as I know you can't directly calculate the dimensions from the Type designation, not least because the types themselves are classes that include sensors of different (similar) sizes. The difference between the largest and smallest Type 1/1.7 chip is appreciable.

The best info you've got comes from the Samsung page.

It says that the resolution is 12,000 x 9,000 pixels and that each pixel is 0.7μm. We don't know how precise either figure is, but the fact they say a 3x3 grid of pixels acts as a 2.1μm pixel suggests it's not 0.745 or anything like that.

12,000 x 0.0007 = 8.4mm

9,000 x 0.0007 = 6.3mm

That's a tiny bit smaller than the old Type 2/3 (8.8 x 6.6mm) sensor format that Fujifilm used to use, which is consistent with the Type 1/5.2 designation they give.

Richard - DPReview.com

Sanglier86 · Apr 16, 2024

Great, thanks for your help

Bob A L · Apr 16, 2024

There was a quite public deception campaign a year or two ago when a major phone mfg. advertised their sensor as 1" diagonal and even included a drawing of the sensor with this dimension included in the advertisement. I was quite disturbed by this and made it a point to respond any time I saw the information published. It didn't do any good though. All anyone would have needed to do was visualize a rectangle with a 1" diagonal measurement and compare that with the physical size of the phone with multiple cameras and see that it could not have been true.

Richard Butler · Apr 16, 2024

Bob A L said:
There was a quite public deception campaign a year or two ago when a major phone mfg. advertised their sensor as 1" diagonal and even included a drawing of the sensor with this dimension included in the advertisement. I was quite disturbed by this and made it a point to respond any time I saw the information published. It didn't do any good though. All anyone would have needed to do was visualize a rectangle with a 1" diagonal measurement and compare that with the physical size of the phone with multiple cameras and see that it could not have been true.

We've been publishing articles trying to make clear how misleading this naming system is for over 20 years now (the original Wikipedia page on the topic drew heavily from one of Phil's articles on here). And if even people within phone companies get it wrong, it suggests it's horrendously misleading.

Frustratingly it's so well embedded in the sensor industry (which isn't especially public-facing) that it's essentially impossible to fix.

We've adopted a system of using the industry terminology Type x/X but removing the spurious 'inch' designation and then staying the approximate dimensions. Unfortunately, despite conducting a public vote, no one seems happy with the result. Also, sadly, some of the other sites, such as Imaging Resource, that said it would follow suit are no longer active.

Richard - DPReview.com

TechStrider · Apr 16, 2024

Some useful information here

You used "exact" in your query.

I am not sure if calculated values are sufficient for you, but this verifies some of the responses above using formulas.

ISOCELL - Wikipedia

en.wikipedia.org

Joseph S Wisniewski · Apr 16, 2024

Richard Butler said:
Bob A L said:

There was a quite public deception campaign a year or two ago when a major phone mfg. advertised their sensor as 1" diagonal and even included a drawing of the sensor with this dimension included in the advertisement. I was quite disturbed by this and made it a point to respond any time I saw the information published. It didn't do any good though. All anyone would have needed to do was visualize a rectangle with a 1" diagonal measurement and compare that with the physical size of the phone with multiple cameras and see that it could not have been true.

Click to expand...

We've been publishing articles trying to make clear how misleading this naming system is for over 20 years now (the original Wikipedia page on the topic drew heavily from one of Phil's articles on here). And if even people within phone companies get it wrong, it suggests it's horrendously misleading.

Frustratingly it's so well embedded in the sensor industry (which isn't especially public-facing) that it's essentially impossible to fix.

Yep. Try to get the lighting industry to stop measuring round LEDs, fluorescent tubes (and LED tube replacements) and small bulbs in "T" units. One T = 1/8 inch, you have a T8 fluorescent tube or a T-1 3/4 LED. Yeah, that's 1/8 * 7/4. Figure out the right drill bit for that.

(At least round LEDs are now generally sold as 3mm and 5mm, roughly T1 and T1 3/4. Don’t count on being able to drill a decent hole until you measure the actual part, though)

Have you ever had a plug get stuck in a really old piece of audio equipment because the equipment has 1/8" (3.175mm) "mini jacks" and modern "mini plugs" are 3.5mm? Or the 3.5 won’t go in at all, so you try the 2.5mm and it goes in but doesn’t really latch or connect solidly. I'm old enough to remember when this was a very frequent problem. It contributed to the "only 1/4 inch phone plugs and jacks are reliable" myth because you could rely on "phone plugs" being a pretty consistent 1/4" or 6.35mm no matter where they were made, while mini plugs might not mate reliably if one side was metric and the other imperial or worse, some weird compromise size.

And don’t get me started on the mystery of wire gauges (and the further mystery of needle gauges) or pipe and tubing sizes.

Richard Butler · Apr 16, 2024

Joseph S Wisniewski said:
Yep. Try to get the lighting industry to stop measuring round LEDs, fluorescent tubes (and LED tube replacements) and small bulbs in "T" units. One T = 1/8 inch, you have a T8 fluorescent tube or a T-1 3/4 LED. Yeah, that's 1/8 * 7/4. Figure out the right drill bit for that.

(At least round LEDs are now generally sold as 3mm and 5mm, roughly T1 and T1 3/4. Don’t count on being able to drill a decent hole until you measure the actual part, though)

Interesting: I'd just assumed the T units were simply the names of different fittings, not actual measurements. Thankfully no one put 'inch' in the name to add confusion.

Joseph S Wisniewski said:
Have you ever had a plug get stuck in a really old piece of audio equipment because the equipment has 1/8" (3.175mm) "mini jacks" and modern "mini plugs" are 3.5mm? Or the 3.5 won’t go in at all, so you try the 2.5mm and it goes in but doesn’t really latch or connect solidly. I'm old enough to remember when this was a very frequent problem. It contributed to the "only 1/4 inch phone plugs and jacks are reliable" myth because you could rely on "phone plugs" being a pretty consistent 1/4" or 6.35mm no matter where they were made, while mini plugs might not mate reliably if one side was metric and the other imperial or worse, some weird compromise size.

Growing up in a primarily metric country I didn't realise that both existed. I thought the large ones were 1/4" (~6.35mm) and the (newer) mini ones were always 3.5mm, with 1/8" as an approximate alias, to avoid having to constantly switch between talking about different scales.

Joseph S Wisniewski said:
And don’t get me started on the mystery of wire gauges (and the further mystery of needle gauges) or pipe and tubing sizes.

Likewise although mountain bike sizes are 'measured' in inches ('named' might be more accurate) and road bikes are sized in cm, thankfully most of the parts use metric bolts and nuts, regardless.

I was fascinated, on going to a US hardware store for the first time, to be confronted by a board with a vast array of metric and US customary nuts and bolts glued to it: so you can check which you're dealing with, before accidentally buying a bucket of not-quite-compatible fastenings.

A rather elegant workaround for a situation that would have ideally been planned around.

Richard - DPReview.com

MarkInSF · Apr 16, 2024

Richard Butler said:
Bob A L said:

There was a quite public deception campaign a year or two ago when a major phone mfg. advertised their sensor as 1" diagonal and even included a drawing of the sensor with this dimension included in the advertisement. I was quite disturbed by this and made it a point to respond any time I saw the information published. It didn't do any good though. All anyone would have needed to do was visualize a rectangle with a 1" diagonal measurement and compare that with the physical size of the phone with multiple cameras and see that it could not have been true.

Click to expand...

We've been publishing articles trying to make clear how misleading this naming system is for over 20 years now (the original Wikipedia page on the topic drew heavily from one of Phil's articles on here). And if even people within phone companies get it wrong, it suggests it's horrendously misleading.

Frustratingly it's so well embedded in the sensor industry (which isn't especially public-facing) that it's essentially impossible to fix.

We've adopted a system of using the industry terminology Type x/X but removing the spurious 'inch' designation and then staying the approximate dimensions. Unfortunately, despite conducting a public vote, no one seems happy with the result. Also, sadly, some of the other sites, such as Imaging Resource, that said it would follow suit are no longer active.

Richard - DPReview.com

Thanks, as always. I've always appreciated your efforts to make sensor sizes clear in your reviews. Your standard graphic showing sensor sizes of all the common types and where a particular camera falls is very helpful. Keep up the good work, even if others aren't.

Leonard Migliore · Apr 16, 2024

Richard Butler said:
That's definitely wrong, in that that's larger than "full-frame."

Full-frame is Type 2.7" in the mad sensor industry terminology, for what it's worth.

So far as I know you can't directly calculate the dimensions from the Type designation, not least because the types themselves are classes that include sensors of different (similar) sizes. The difference between the largest and smallest Type 1/1.7 chip is appreciable.

The best info you've got comes from the Samsung page.

It says that the resolution is 12,000 x 9,000 pixels and that each pixel is 0.7μm. We don't know how precise either figure is, but the fact they say a 3x3 grid of pixels acts as a 2.1μm pixel suggests it's not 0.745 or anything like that.

How is the lens going to focus on sub-micron pixels? They're 1 or 2 waves wide, depending on the wavelength of the light. I haven't done optics for a while, but I'd guess that you would get more than a micron of diffraction even at f/1.

Joseph S Wisniewski · Apr 17, 2024

Leonard Migliore said:
Richard Butler said:

That's definitely wrong, in that that's larger than "full-frame."

Full-frame is Type 2.7" in the mad sensor industry terminology, for what it's worth.

So far as I know you can't directly calculate the dimensions from the Type designation, not least because the types themselves are classes that include sensors of different (similar) sizes. The difference between the largest and smallest Type 1/1.7 chip is appreciable.

The best info you've got comes from the Samsung page.

It says that the resolution is 12,000 x 9,000 pixels and that each pixel is 0.7μm. We don't know how precise either figure is, but the fact they say a 3x3 grid of pixels acts as a 2.1μm pixel suggests it's not 0.745 or anything like that.

Click to expand...

How is the lens going to focus on sub-micron pixels? They're 1 or 2 waves wide, depending on the wavelength of the light.

1/2λ resonant pixels?

Leonard Migliore said:
I haven't done optics for a while, but I'd guess that you would get more than a micron of diffraction even at f/1.

The airy disc is 1.27μm at f/1.0.

But keep two things in mind. First, you can't reach a resolution of 1.27μm with a pixel pitch of 1.27μm. You have to oversample a bit to realize that whole 1.27μm resolution.

Second, you're talking a Bayer pattern sensor, so the nearest distance between two pixels diagonally is sqrt(2)*pitch, so even if you didn't need oversampling, you'd still want a 1.27μm/sqrt(2) = 0.90μm pixel pitch.

PhotoTeach2 · Apr 17, 2024

Sanglier86 said:
Hi, I'm trying to find out the exact size in mm of the Samsung ISOCELL HM2 108mp CMOS sensor. I've only found the fractional size online (1/1.52)

This may not be the case, but that may not have an exact relationship to the actual size if they are using the (conventional) GLASS-TUBE OD diameters, (aka Vidicon/Plumbicon tubes).

You then have to subtract (unknown) glass-wall thickness and then fit a rectangle within its flat ID area.

That is why we refer to the 1" sensors as 1"-type since they are nowhere close to an actual 1".

Sanglier86 said:
It seems you can use this size to calculate the dimensions using the equation from this page.

https://commonlands.com/blogs/technical/cmos-sensor-size

The thing I'm struggling with is how do I use the fraction in the equation?

Many thanks

Sensor size question

New member

Community Leader

New member

Community Leader

Senior Member

Forum Pro

Forum Pro

Forum Pro

Member

Managing Editor

New member

Forum Pro

Managing Editor

Well-known member

Forum Pro

Managing Editor

Senior Member

Forum Pro

Forum Pro

Forum Pro

Similar threads

Keyboard shortcuts