350D has a higher res sensor than 1Ds II?

[smasraum]

I think it's actually preferable to have larger "pixels" (because I can't think of the correct term) on a larger sensor, but I've done the math and the 350D (and therefore, I assume the 20D too) have more pixels per area of photo than the 1Ds Mark II.

The 350D has 155 pixels per mm of sensor while the 1Ds mkII has only 138 pixels per mm.

If the CMOS of the 350D was scaled to be full size 36x24 like the 1Ds the camera would take pictures that were 5604 pixels wide instead of 4992 like it does now.

So if you took a shot on the 350D and then took the same shot on the 1Ds mkII and cropped it to the same physical size as the 350D you'd end up with a lower res shot.

Not really important, but interesting (to me at least). :rolleyes:

[Dante King]

whoa horse. how can this be true when the 1dsm2 has more than double the pixels on a frame that is not double the size.

[Ikinaa]

whoa horse. how can this be true when the 1dsm2 has more than double the pixels on a frame that is not double the size.

the size of an APS-C sensor is 22.2 x 14.8 = 328 mm^2
the size of a FF sensor is 36 x 24 = 864 mm^2
so a FF is 2.6 times larger than a APS-C sensor

[smasraum]

Ah, but you've gotta think in multiple dimensions. Something like if the earth was a smooth ball and the ball grew by 1 inch in radius the volume would actually double, or something like that. (I can't remember the details, but they sound that ridiculous)

The CMOS on the 1Ds mkII is 864mm square. The CMOS on the 350D is 328.56mm square. So the sensor has less than half the area with almost half as many pixels.

ratio of area 2.62:1
ratio of pixels 2.086:1

Do the math, the 1Ds mkII has a 36mm wide sensor and will shoot pictures that are 4992 pixels wide, that's 138 pixels per mm. The 350D has a 22.2mm CMOS and will shoot pics 3456px wide for 155 pixels per mm. For the 1Ds mkII to have the same density it'd have to be an 18.7mp camera. the pics would be 5604x3736 instead of 4992x3328.

Like I said, the quality due of the images is still probably greater on the 1Ds mkII for a bunch of reasons, but I thought the math worked out in a way that most people wouldn't expect. It surprised me.

[felix21685]

interesting read at 12 at night :)

[robertwgross]

You might want to sort out the various terms. Resolution versus pixel density.

---Bob Gross---

[goatee]

But surely one of the reasons that DSLRs take higher quality images than P&S with higher MP counts (and vastly higher pixel density) is that as pixel density increases noise increases.

[smasraum]

But surely one of the reasons that DSLRs take higher quality images than P&S with higher MP counts (and vastly higher pixel density) is that as pixel density increases noise increases.
I can't say for sure, but I've read in a post here that a larger sensor ie dslr will take better pics than a p&s sensor which is usually smaller. Something about larger pixels over a larger area working better or something, I can't remember. Like I said, the 350D is not necessarily going to take better pictures because of this, it just seemed like some interesting math.

Bob, yep, seems the term pixel wasn't really the correct term, but I couldn't remember the correct term so I went with it. It's late, I'm tired but can't sleep, the math seemed interesting.

Would you please take the info that I presented and offer a more correct and in depth explanation?

[neil_r]

You guys really do have way to much time on your hands....

I hope none of you loose sleep over this :D

N

[robertwgross]

Would you please take the info that I presented and offer a more correct and in depth explanation?

I'll decline. I don't think it is worthwhile, and I have 75 images left to edit.

---Bob Gross---

[Lotto]

I think Steve is correct, and I read the same math on POP magazine once. The example is like this: On a stationary tripod, use a same 50mm lens shooting a beer can 10 ft away (we could all use a bear at 2 AM:D ), the 20D (or 350D) would have more pixles counts on the image of beer can itself than the 1DsII.

If we move the 1DsII closer to get same frame crop... WAIT, I will stop right here before I confuse myself:eek:

[MDJAK]

All right, I'm convinced. I'm going to run and sell my 1Ds-MKII immediately for the 350D. What a moron I must have been in the first place for buying one. Heck, I should have stuck with my Kodak Instamatic.:D

[2new]

Kodak Instamatic... haven't heard that in quite a while :)

[CyberDyneSystems]

It is true that the 350D has more "photosites" per mm than the 1Ds MkII.

"Pixel" is commonly used to replace the term "Photosite" when referring to Image sensors.. though technically not correct.. it is accepted,.

...so accepted in fact that the term Bob used to describe the 350D sensor's "advantage" over the 1Ds is "Pixel density" when in fact we are really talking about "Photosite Density" :mrgreen: (Before typing this however I have never heard the term "Photosite density)

Anyway.. yes "Pixel Density" is the term for the factor that is often forgotten when cameras of differing sensor sizes are compared.. and in the end Pixel density is a key ingredient to understanding all the issues that come up when discussions on sensor sizes,. "X-factors" etc come up.



It is importatn to remember that small P&S cameras with there diminutive sensor sizes have MUCH higher pixel density (photosites per mm) than any DSLR.. and yet no one would argue that they have "better" or "higher resolution" images.


Higher Pixel Density is not without it's sacrifices.. image noise, quality of per "photosite" color and light gathering abilites is degraded as the photosites get smaller... (as they must to fit more in the same space)

On the other hand.. higher Pixel Density is one of the advantages of the 20D and 350D over say a 1D of any type.. for those using telelphoto lenses..

It's all well and good to like a 20D because it offers the 1.6 X-Factor to our 300mm lens.. saving us lot's of $$$ on larger lenses,.. but if the 20D's "Pixel density" weren't so high relative to the 1D cameras.. if it were the 300D's 6MP sensor,. then the advantage would not exist,.. as a similar crop from the 1Ds MkII would have more pixels/photosites in the same area.

[smasraum]

MDJAK, I'll make the ultimate sacrifice. You send me your well used 1DsmkII and $100 and I'll send you my practically new 350D and we'll call it even. (unless you have a 70-200 F4L lens in which case you can substitute that for the $100)

:D

[MDJAK]

I always knew you Texans were generous folk.:)

And I'm unhappy to have to report/admit, my 1Ds-MKII is not yet well used. I don't think I've taken more than 2,000 shots with it. Unless that constitutes well used.

Also, I don't have the 70-200 f4, only the 70-200 f/2.8 IS. Will that do?

[smasraum]

Well, 2000 shots is probably 1800 more than me, but we'll let you slide on the well used. I guess the 70-200 2.8 IS will do, but you'll have to send $50 with that one. :lol:

[AJSJones]

Resolution is the problem word here!

Some think of the original meaning : "linear resolution" i.e." how fine a detail" can be recorded (film/lens combos used to be - I guess still are - evaluated with a view to seeing how many line-pairs/mm could be recorded - no-one thought about total line pairs! Because you could have film of different sizes, it was a meaningless concept.)

Along came digital and the megapixel wars and a 6MP camera *obviously* had more resolution than a measly 3MP camera. So now, many think of the *number of details" that can be recorded when they hear "resolution"

So the 20D has the ability to resolve finer details in an image focused on its sensor, but it does not record as many details when compared to the 1Ds2 (or any other Canon DSLR, for that matter - except the 350D) given the same image

[muscleflex]

I think Steve is correct, and I read the same math on POP magazine once. The example is like this: On a stationary tripod, use a same 50mm lens shooting a beer can 10 ft away (we could all use a bear at 2 AM:D ), the 20D (or 350D) would have more pixles counts on the image of beer can itself than the 1DsII.


haha! it IS definitely late where you are - no one would want a BEAR at 2am! it'd eat you alive!! haha!

[MDJAK]

Well, 2000 shots is probably 1800 more than me, but we'll let you slide on the well used. I guess the 70-200 2.8 IS will do, but you'll have to send $50 with that one. :lol:

If I throw in my 24-70 F2.8 Canon, will you take that instead of the fifty? I'd rather not have to throw in cash.:)

[Persian-Rice]

CDS basically nailed it on the head. The advantage of a lower pixel density isn't really the better quality at higher resolutions. When you have more photosites on a sensor, the individual sensitivity of each photosite to light is diminished and luminance and chrominance signal accuracy is affected. As current processors in cameras improve, this issue is slowly being wiped away.

When the 20D was announced, many, including myself, were concerned that noise might be an issue. However, Canon had made huge strides in the signal boosting efficiency of the camera and in turn, actually improved accuracy across the board. In terms of image quality, they are the same, the 1Ds might be more prone to noise, but the image processor fixes that issue. So in terms of overall image quality, there is no difference.

This is where I think digital is still weak. That across the board, basically all the images are the same and the final look is achieved through processing. The 1Ds, as Canon's top camera should be able to reproduce images much cleaner and with better colours right at the sensor, but it doesn't, that what I think sensor manufacturers will be improving on in the near future. The sensor will be able to capture far more and leave processing and repairing to a minimum, right now, they are doing it all through processing.

[smasraum]

Lots of good info here. Thanks.

I suppose I'll take the 24-70. But mostly just to do you a favor. :)

[MDJAK]

I guess that's why you come from "friendswood."

And would you mind if I take it with me tomorrow when I leave for 12 days in Hawaii? If you insist, I'll send you the whole rig today and have to use my son's S70, my middle daughter's S50, or my youngest daughter's coolpix 7900.
I'll also be stuck with my Sony 3ccd video camera.

[Longwatcher]

Former Imagery analyst speaking here:

The original statement that started this thread was a correct statement
The 350D and 20D do have higher theoretical resolution sensors then the 1DsMkII over the same area. Given taken at same distance with same lens (which they can do)

However, since both are starting to hit the lens resolution limits, the slightly larger photosites of the 1DsMkII will frequently result in a slightly higher final resolution. There are a host of other factors that go into determination of final resolving ability of the camera system that it could boggle the mind.

But you are only talking resolution; not quality, not area of coverage, not dynamic range, noise, color, or a host of other factors that differentiate the 1DsMkII from the 20D and the 350D.

Just my experience,

[NeoteriX]

This is where I think digital is still weak. That across the board, basically all the images are the same and the final look is achieved through processing. The 1Ds, as Canon's top camera should be able to reproduce images much cleaner and with better colours right at the sensor, but it doesn't, that what I think sensor manufacturers will be improving on in the near future. The sensor will be able to capture far more and leave processing and repairing to a minimum, right now, they are doing it all through processing.

I think the final look always has to be achieved through processing -- especially with the prevalence of the Bayer Interpolation style sensors. When we're only capturing one color at each pixel, and then "intelligently" combining them together during the processing, the faithfulness of the reproduction -- the cleanliness and better colors -- will always be dependant on processing. Improvements to the sensitivity and resistance to interference of the photosits will surely help image quality, but the in camera processing/repairing you speak of seems to me to "be" digital imaging.

[ed2day]

Also keep in mind that the full frame sensor captures a larger image than the smaller sensor. Both in physical dimensions (sensor size) and number of pixels. So to get an equivalent image from the smaller sensor it must be magnified a greater amount (in the case of prints) or interpolated and resized(in the case of a digital image, ie comp monitor). Both result in loss of resolution.

There's no doubt that a bigger sensor (why stop at full frame??) will collect more "information". Information can be limited by either resolution or signal/noise ratio and the designer can trade one off against the other.

[Persian-Rice]

NeoteriX, I agree that all images must go through processing and that is something that can't be worked around. The processing, which really acts a compiler and translator of data, in many cases takes flawed and improper data and tries to fix it. This is basically down to poor sensor design, or rather, a lack of an affordable technology. My thinking is that if they do produce better sensors and record cleaner and more accurate data, processing will not be needed to correct errors, just enhance good, clean, and correct data.

The problem is that sensors aren't able to get it right the first time around, and they rely on a repairman to fix this bad data. Why not have sensors give you the best signal possible from the start.

Take sharpening as an image for example, or working in 16 bit. Why do we leave sharpening to the end? why do we convert to 8 bit at the end? Because we want to avoid any loss of quality before we fix things, we want to keep as much good data as possible. It's the same thing here, we need better data to work with, not take bad data and try to fix it. Why not take good data and enhance it to make it even better.