5D / 7D - Full Frame vs. 1.6x - Pros and Cons

[egproductions]

It would be a waste of money to use EF lenses on a crop sensor because you lose all of the extra field of view that you were paying for in the first place. The faster EF lenses wouldn't help either because the crop sensor does not have the same low-light capabilities of the full frame sensor. Best thing to do with APS-C cameras is use the APS-C lenses that were designed for them. Unless you have the extra money for more expensive primes from Zeiss, Cooke, etc...in which case you probably wouldn't be looking to buy the semi-pro 7D model.

Just the opposite. it's a waste of money to buy EF-S lenses. Yes they are cheaper to manufacture but their resale value is not great. EF lenses are almost an investment. You can sometimes sell them for more than you bought them for years down the road.

Secondly, using an EF lens on a cropped sensor will discard the edges of the lens which are optically inferior to the middle of the glass.

"The faster EF lenses wouldn't help either because the crop sensor does not have the same low-light capabilities of the full frame sensor."

All the more reason to buy faster lenses

 

[ydgmdlu]

Sensor size and the number of photodiodes on the sensor are related so it does contribute to low-light capabilities. Full-frame sensors have fewer, larger photodiodes on the chip, which in turn helps the chip gather more light.

I think you need to read up on Canon's white paper on Full-Frame CMOS chips.

http://www.usa.canon.com/uploadedimages/FCK/Image/White Papers/Canon_CMOS_WP.pdf

Do a little research yourself sometimes before just quoting what someone else has said.

Thank you for being presumptuous about what I know and don't know. I guess that you might not have been around these forums long enough to know that I've said EVERYTHING that Barry said in his post months ago when people were just starting to talk about the GH1 and its low-light abilities.

Here is an example of one of my posts: http://www.dvxuser.com/V6/showthread.php?p=1637000

Sensor size is not a factor. Photosite size is the factor. The correlation between sensor size and low-light performance is coincidental.

 

[Ian-T]

Um…vdgmdlu is correct on this one. It’s quite possible that the low light capability on this new cam will match the 5Dll. Too early to tell.

 

[KeithAndrews.TV]

So basically it is correct then to say that chip size does influence low-light capability. The reason being is that based on current offerings by Canon, Nikon, etc... no one is manufacturing a full-frame chip with hundreds of tiny sensors, they are manufacturing them with fewer, larger sensors. Not coincidence, that's just how they are being made.

The opposite stands true for APS-C sensors. They make those with many smaller sensors. I think people are being a little too nitpicky with wording on these forums.

 

[egproductions]

When everyone is speaking of full frame and APS-C sensors, they are really talking about the 7D and the 5D. It's true that it's too early to tell which one is better in low light but given that they have both been produced in a relativly short time from eachother, I think it's safe to say that the 5D will be able to handle low light better maybe 1 stop.

 

[egproductions]

It's like saying that desktops are more powerful than laptops. It's generally true unless you are talking about a 10 year old desktop and a core2 duo laptop

 

[KeithAndrews.TV]

Just the opposite. it's a waste of money to buy EF-S lenses. Yes they are cheaper to manufacture but their resale value is not great. EF lenses are almost an investment. You can sometimes sell them for more than you bought them for years down the road.

Secondly, using an EF lens on a cropped sensor will discard the edges of the lens which are optically inferior to the middle of the glass.

"The faster EF lenses wouldn't help either because the crop sensor does not have the same low-light capabilities of the full frame sensor."

All the more reason to buy faster lenses

Considering this is a forum based around video production, I would have to disagree that it would be worth the extra money to use faster EF lenses on an APS-C camera. Although the edges of a lens are traditionally the weakest, you would need a lot more detail than what HD can resolve in order to see that loss. It's a great idea for still photography shooting 5k res, but you wouldn't see the difference in HD.

Also, you have to pick your battles. You either gain light sensitivity and lose field of view from the crop, or you keep the original field of view intended and maybe have a slower lens. It would really depend on what lenses you were comparing.

 

[KeithAndrews.TV]

When everyone is speaking of full frame and APS-C sensors, they are really talking about the 7D and the 5D. It's true that it's too early to tell which one is better in low light but given that they have both been produced in a relativly short time from eachother, I think it's safe to say that the 5D will be able to handle low light better maybe 1 stop.

Oh it's absolutely true that down the road, maybe even with the 7D, that the low-light capabilities of these cameras, whether full-frame, or not, will be very close to one another, or even the same. I'm just basing my information from my experiences, and what I have gathered from original sources.

 

[Barry_Green]

So basically it is correct then to say that chip size does influence low-light capability.

Only maybe.

For example, you can get a 1/3" Z1U, or a 1/3" VX2000. The VX2000 is miles better in terms of sensitivity. Laughably better. But they're the same chip size!

The difference, again, and the only difference that matters, is pixel size.

All other things being equal, a 14-megapixel full-frame sensor will be more sensitive than a 14-megapixel APS-C sensor. But a 10-megapixel APS-C sensor would be way more sensitive than a 28-megapixel full-frame sensor.

The argument you're applying (bigger sensor size = better sensitivity) is akin to someone declaring that coats are toxic to plants. When asked to explain, said individual says "well, I put my plants in the closet, and they all died. The only things in my closet were coats, therefore I deduce that coats are deadly to plants."

Fewer, bigger pixels = better imagery in all ways (except one: resolution). Fewer, bigger pixels = better sensitivity, better dynamic range.

So, just do the math -- figure out the sensor size (22mm for Canon 7D APS-C, 36mm for "full frame") and divide by the # of horizontal pixels (5000-something for Canon 7D, don't know on the 5D) and figure out which one has the bigger pixels. If the 7D has bigger pixels, it should be more sensitive than the 5D, even though the 5D has the bigger sensor!

I think people are being a little too nitpicky with wording on these forums.

If you're meaning ydgmdlu and me, we're being picky about accurate terminology. Can't help it, and not gonna change.

 

[ydgmdlu]

So basically it is correct then to say that chip size does influence low-light capability. The reason being is that based on current offerings by Canon, Nikon, etc... no one is manufacturing a full-frame chip with hundreds of tiny sensors, they are manufacturing them with fewer, larger sensors. Not coincidence, that's just how they are being made.

The opposite stands true for APS-C sensors. They make those with many smaller sensors. I think people are being a little too nitpicky with wording on these forums.

No, it is coincidental in the sense that larger chips seem to perform better in low light only because they allow for larger pixels (as well as lenses with larger absolute apertures). In the end, what counts is the pixel size, not the chip size. You're using confusing terminology.

An APS-C chip can have identical low-light performance to an FF chip, if the sensor and image processor technology is the same, and if the sizes of the pixels are the same.

With the 7D, the pixels may or may not be smaller than the 5D's. We do know that Canon claims that the pixels are much closer together than in previous CMOS generations.

 

[Taylor Rudd]

I think it's safe to say that the 5D will be able to handle low light better maybe 1 stop.

While the pixel density of the 5D is probably less than that of the 7D, I doubt the 5D is twice as sensitive as the 7D.

 

[Taylor Rudd]

So basically it is correct then to say that chip size does influence low-light capability.

Indirectly. Pixel density is what affects low-light capabilities. What is true in this situation - though not for any other reason than marketing - is that bigger sensors mean more pixels, simply because more pixels sell.

I would rather the 7D be a 10 or 12 mp camera and have higher light sensitivity. If the sensor was only 1920x1080, this thing would be off the hook.

Less pixel density = higher light-sensitivity.

 

[Barry_Green]

Okay, math on the 7D: largest image it takes is 5184x3456, so assuming that's the active pixel data on the sensor, that leaves a pixel size of 4.25 microns in each dimension, or 18 square microns.

On the 5D, the largest image it takes is 5616x3744; that leaves a pixel size of 6.4 microns in each dimension, or about 41 square microns.

Lots more imager area, not a lot more pixels, means each individual pixel is over twice as big. Which means, basically, that you can probably expect the 5D to be a bit over a stop more sensitive than the 7D. If all other things are equal.

 

[Rakesh Jacob]

If you're meaning ydgmdlu and me, we're being picky about accurate terminology. Can't help it, and not gonna change.

This is why we love you!
You had me at "At the National Association of Broadcasters convention in 2002, Panasonic made two startling announcements." :love4:

EDIT: Seriously the DVX Handbook was the first and only book I've read on video cameras and filmmaking and the knowledge has served me well thus far!!!! (everything else I know I learned on the streets like I did about sex and how to make crack, etc...)

 

[Taylor Rudd]

Which means, basically, that you can probably expect the 5D to be a bit over a stop more sensitive than the 7D. If all other things are equal.

I stand corrected!

What I said about pixel density remains true, right Barry? By density, of course, I mean how many pixels are found in an X by Y area. This correlates with your Micron measurements of the 5D and 7D.

5D has larger microns, therefore has less pixel density, therefore has higher light-sensitivity. Correct?

 

[ydgmdlu]

However, bear in mind that Canon claims that the pixels on the 7D are closer together, which means that they could still be comparable in size to the 5D's.

So the answer is still: Gonna have to wait until someone (i.e. Phillip) actually tests and does a side-by-side comparison before we know for sure.

 

[KeithAndrews.TV]

No, it is coincidental in the sense that larger chips seem to perform better in low light only because they allow for larger pixels (as well as lenses with larger absolute apertures). In the end, what counts is the pixel size, not the chip size. You're using confusing terminology.

An APS-C chip can have identical low-light performance to an FF chip, if the sensor and image processor technology is the same, and if the size of the pixels are the same.

With the 7D, the pixels may or may not be smaller than the 5D's. We do know that Canon claims that the pixels are much closer together than in previous CMOS generations.

Well I certainly appreciate the accurate wording from everyone, although one persons sensor is another persons chip is another persons pickup device is another persons imager, etc... The point is unless we were to all use the same exact terminology, maybe that designated by the manufacturer, we are going to spend all day correcting one another and probably not even be talking about the same thing.

I was just making a generalization that based on the larger sensors of today, which in turn offer more room for larger pixels (again, current offering from the manufacturers), the larger sensors offer better light gather capabilities. That's it. Of course that may not be true 10 years from now, or even 1 year from now. Just basing the information on what has been released to date.

 

[Barry_Green]

I would rather the 7D be a 10 or 12 mp camera and have higher light sensitivity.

Me too!

If the sensor was only 1920x1080, this thing would be off the hook.

No, there'd be quite a bit of hook there, because the res would be too negatively impacted. Remember that this is a single-sensor design, utilizing a Bayer pattern. Therefore the maximum resolvable detail is only going to be around 75% of the native pixel density. On a 1920x1080 sensor, you'll get a pretty good 720p's worth of detail out of it, but not much more.

Now, a 3K sensor at APS-C size... that would be the premium bomb-diggity. Ultimate sensitivity, ultimate resolution, movie-film-style DOF, miniscule noise, maximum latitude, and massive 61-micron pixels... now you're talking!

 

[Barry_Green]

What I said about pixel density remains true, right Barry? By density, of course, I mean how many pixels are found in an X by Y area. This correlates with your Micron measurements of the 5D and 7D.

Exactly. The only factor that we haven't really accounted for is the insensitive registers between pixels... that becomes a proportionately bigger factor the smaller the pixels get.

5D has larger microns, therefore has less pixel density, therefore has higher light-sensitivity. Correct?

Well, yes except the terminology. Microns are a set unit of measurement, like "millimeter" or "inch". The 5D's pixels are larger, meaning they measure more microns across, and that does indeed mean that the relative density of pixels is lower, with each pixel being over twice as large, and should therefore translate into higher sensitivity.

With the wild-card being that we don't know if all other things are equal... maybe some technical wizardry has been implemented which makes the new 7D's sensor effectively more sensitive? Remember that all these discussions are crucially related to "all other things being equal."

 

[KeithAndrews.TV]

Me too!


No, there'd be quite a bit of hook there, because the res would be too negatively impacted. Remember that this is a single-sensor design, utilizing a Bayer pattern. Therefore the maximum resolvable detail is only going to be around 75% of the native pixel density. On a 1920x1080 sensor, you'll get a pretty good 720p's worth of detail out of it, but not much more.

Now, a 3K sensor at APS-C size... that would be the premium bomb-diggity. Ultimate sensitivity, ultimate resolution, movie-film-style DOF, miniscule noise, maximum latitude, and massive 61-micron pixels... now you're talking!

You're an engineer right?! Why don't you draft one up and submit it to Canon :grin: