-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Lens musings - Prosumer vs. dSLR
- Thread starter Tul
- Start date
J
John Kim
Guest
Lemme try to put numbers to what he's saying.
Say an f/2.0 lens puts 800,000 units of light onto a square millimeter.
An f/5.6 lens would put 100,000 units of light (three stops fewer) onto a square millimeter.
A 10D has 6.5 million pixels in 22.7x15.1mm, or 19,017 pixels per square mm. An f/5.6 lens would put 100000/19017 = 5.26 units of light onto each pixel.
A 828 as 8 million pixels in 8.8x6.6mm, or 137,741 pixels per square mm. An f/2.0 lens would put 800000/137741 = 5.81 units of light onto each pixel.
If the sensor sensitivity is equal for both cameras, this means a 10D with an f/5.6 lens can produce roughly the same quality pictures (in terms of noise) as the 828 with an f/2.0 lens.
Yes you'd have to crank up the 10D's ISO by three stops to accomplish this, but that's really the point he's making. You have to remember that ISO in digital cameras is an artificial construct. Changing the ISO just cranks up the gain in the picture. It's like shooting 100 ISO film and underexposing, scanning it, and stretching the levels curve in Photoshop to make the picture look like it was shot properly exposed with 800 ISO film. The digicams with smaller sensors get less light per pixel for a given EV, so have to do more stretching, and pay for it in image noise.
Essentially he's correct. We shouldn't be comparing cameras using the same ISO. We should be comparing using the same image noise levels (which roughly correlates to pixel pitch). Viewed that way, a 10D with an f/5.6 lens is about equivalent to the 828 and its f/2.0 lens, both in exposure (normalized for noise) and DOF.
Say an f/2.0 lens puts 800,000 units of light onto a square millimeter.
An f/5.6 lens would put 100,000 units of light (three stops fewer) onto a square millimeter.
A 10D has 6.5 million pixels in 22.7x15.1mm, or 19,017 pixels per square mm. An f/5.6 lens would put 100000/19017 = 5.26 units of light onto each pixel.
A 828 as 8 million pixels in 8.8x6.6mm, or 137,741 pixels per square mm. An f/2.0 lens would put 800000/137741 = 5.81 units of light onto each pixel.
If the sensor sensitivity is equal for both cameras, this means a 10D with an f/5.6 lens can produce roughly the same quality pictures (in terms of noise) as the 828 with an f/2.0 lens.
Yes you'd have to crank up the 10D's ISO by three stops to accomplish this, but that's really the point he's making. You have to remember that ISO in digital cameras is an artificial construct. Changing the ISO just cranks up the gain in the picture. It's like shooting 100 ISO film and underexposing, scanning it, and stretching the levels curve in Photoshop to make the picture look like it was shot properly exposed with 800 ISO film. The digicams with smaller sensors get less light per pixel for a given EV, so have to do more stretching, and pay for it in image noise.
Essentially he's correct. We shouldn't be comparing cameras using the same ISO. We should be comparing using the same image noise levels (which roughly correlates to pixel pitch). Viewed that way, a 10D with an f/5.6 lens is about equivalent to the 828 and its f/2.0 lens, both in exposure (normalized for noise) and DOF.
Sundance138872
Leading Member
--
That's true, but only if the light is insufficient for the 10D to use lower ISO and still get the picture.
If the shooting takes place on a normally lit day (outdoors off course) both camera would have enough light to operate at their lowest ISO setting, but this would only benefit the 10D.
This is getting a little hairy, and english is not My native language so I hope this makes any sense....
Sundance
Jason Hutchinson
Forum Pro
Yes, it's all about pixel area and how much light is gathered. However the most important measurements isn't the area of one pixel (in fact rather short sighted), it's the area of the entire sensor. The pixel size is only important in finding out how much noise is visible at 100%, but if you want to know how much noise is evident in your photo, you need to look at the entire pixel capturing area. A sensor with only 1 million 1D sized pixels would be much more noisy in a high ISO photo than the 4 million pixel 1D sensor.
Thought of another way, even if the 10D had pixels the size of a Sony 828, it would still have far lower noise at the photo level.
Going back to the example of the 1 million 1D sized pixel, the total energy captured is 1/4th what a 4MP 1D is, so it would need to be amplified 4x to get the same print/display size as a 1D. Get it now?
Jason
Godfrey
Forum Pro
R
Rob Patterson
Guest
I'd like to give some more information regarding my experience that led to this statement. The Camera with the f/1.8 lens of 50mm focal length was a Pentax. Model number forgotten. Year possibly 1968 or 1969. The camera with the f/1.2 lens of 50mm focal length was a Canon. Model number forgotten. Same time frame.
I would roll my own film for these 35mm cameras. Forty rolls per day were shot between the two. Everyday for approximately 3 years.
Because of this, I can in no way forget the performance of the two cameras when loaded with the same ASA speed film and used to shoot the same subject under the same lighting conditions.
No ends ifs or buts about it. When these two cameras were both set to the same aperature, the Canon yielded faster shutter speeds.
I have asked why. Question still not answered.
Aeronautical engineers with all of their technical expertise have determined that bees can't fly. We all know differently. Even those of us who flunked basic 3rd grade math. Which I didn't. I had trouble with Algebra. Did fine in Geometry and Trig.
Do know that a larger clear aperature will transmit more light. That's why I first built a Newtonian reflector with a 4.25 inch diameter mirror (clear aperature). Then I built a 6 inch. Could see more faintly lit deep sky objects. Then built an 8 inch and could see even more faintly lit deep sky objects.
So why did the 50mm f/1.2 yield significantly faster shutter speeds than the 50mm f/1.8. Bees do fly so maybe this question can't be answered?
More likely a difference in one cameras reported speed and true speed. Another reason could be coatings on the optics. If one is fully-multicoated, it will transmit considerably more light than a completely uncoated one...Especially if it has many lens elements.
This isn't rocket science, f/5,6 is f/5,6, nomatter the max aperture...How else could people use light meters than output only an aperture and shutter speed, valid for any lense?
This isn't rocket science, f/5,6 is f/5,6, nomatter the max aperture...How else could people use light meters than output only an aperture and shutter speed, valid for any lense?
Jason Hutchinson
Forum Pro
Actually that's B.S., commonly used on forums.
Yes, and you're USING it at wide aperture. Of course it's going to be different.
LOL. You act like optics is some black science that we've yet to understand. In fact it hasn't changed much in the past 50 years. I don't know what you want to hear, but when I shoot my 50f/1.4 and 50f/1.8 at the same aperture they have exactly the same shutter speed. Sure, the exposure is different in the corners at wide aperturers. There are lots of reasons why the exposures could have been different, but there is no way to guess. I suggest you do some more reading (or testing if you prefer hands on) on these topics.
Jason
Hm, peculiar.
My old Pedrotti&Pedrotti textbook on optics gives a slightly different formula. Doing a google search turned up at least one site that supports the formulas I'm using ( http://www.mapril.net/dof/dof.htm ).
That's hardly a guarantee for correctness, though I'll have to check further into this...
My old Pedrotti&Pedrotti textbook on optics gives a slightly different formula. Doing a google search turned up at least one site that supports the formulas I'm using ( http://www.mapril.net/dof/dof.htm ).
That's hardly a guarantee for correctness, though I'll have to check further into this...
Tom Crowning
Senior Member
Sorry, but what you say is absolutely wrong. Physics and tests of various sensors with different pixel size shows that larger pixels mean less noise.
Regards,
tc
Some digital cameras, some lenses, 2 eyes
http://www.tom-crowning.com
Regards,
tc
--
Some digital cameras, some lenses, 2 eyes
http://www.tom-crowning.com
Brian D. Schneider
Forum Pro
--One of the cameras meters was not reading correctly. the only way this test could mean anything is if the 2 lenses could be used in the SAME camera.
Brian Schneider
Hmm, I'll agree that a higher S/N-ratio could be extracted, but I wouldn't say it's more noisy...
Yes, because the 10D would still have larger pixels.
Ah, yes, when comparing at same printed size, of course the optical amplification will decrease resolution. Noisier? by lower resolution, yes. But this is rather irrelevant to this discussion as this is a different kind of 'noise'.
There the 828 would win out because of resolution - If you enlarged your photo enough to make the pixels look like sqaure blocks...The 828 would have smaller blocks.
Jason Hutchinson
Forum Pro
Likely you're confused, although it's difficult to tell where because you haven't specified about which you're speaking of. Likely your confusion is speaking about noise/pixel, instead of the more important noise/photo.
If you doubt a 10D sized sensor with pixels the size of a 828 would have less noise, simple simulations with a 10D, D60 or about any camera could demonstrate this. I'll skip the demonstration unless you specify that you doubt this.
The reason why people say "bigger pixels" mean less noise, is because they assume you're talking about similar pixel counts which more importantly implies larger sensors. A good example is the fact that the 1Ds has less noise than a 1D at most, if not all of the available ISOs of the 1Ds. I recommend you try this with identical targets, and you'll see.
Jason
Well, you could also look at it from a lens point of view...How fast (expensive) a lens do you need, etc..
Jason Hutchinson
Forum Pro
No, you misread. I specifically stated a 10D with the same sized pixels (2.7u) so by definition the 10D would have identical pixels. In this case, the
I'm not sure what you're talking about here, other than to say yes I'm comparing the same print size. What else matters when comparing two cameras? Yes it's different than comparing noise at the pixel level, but it's also more useful for 90% of cases.
I suspect the only reason you mentioned this is because of the above error with the pixel size. If they're the same pixel size, they have the same absolute resolution, although the 10D sized sensor would have 4x the resolution relative to the frame (the measurement Phil performs.)
Jason
Tom Crowning
Senior Member
Jason Hutchinson wrote:
[...]
That's just optophysics, but if you don't want to believe me you don't have to
Regards,
tc
--
Some digital cameras, some lenses, 2 eyes
http://www.tom-crowning.com
[...]
If you have 2 sensors with the same size and the same number of pixels, the one with the bigger pixels has less noise because each pixel can gather more photons and therefore less amplification (which produces what we call 'noise') is necessairy.
That's just optophysics, but if you don't want to believe me you don't have to
That does NOT imply larger sensors, that's where you are wrong. If you look closely at a sensor you'll notice that there's a lot of space between each pixel, mainly hidden behind the micro lenses.
You ignore the fact that the 10D and the 1Ds have different circuits and a completely different way to reduce noise!
Regards,
tc
--
Some digital cameras, some lenses, 2 eyes
http://www.tom-crowning.com
R
Rob Patterson
Guest
Yes, I did mention this. Reflectance of the elements lens elements from one lens to another will effect light transmission between the two at the same aperature.
They aren't exactly valid for each and every camera. The Pentax had no light meter. The Canon did although I prefered to use the handheld light meter for both. The Canon with the f/1.2 required bracketing for less exposure for the same shutter speed. The solution was to simply shoot at a faster shutter speed with the Canon at any given aperature.
R
Rob Patterson
Guest
I don't B.S. The fact that bees can't fly was something I remember hearing. It may in fact not be true. I've visited many, many forums since the days of 9600 baud modems. I've never seen it mentioned at all. Not once!
The light transmission to film will be different also. Depending on the effective aperature.
I have done more reading. There are a lot of factors that'll effect light transmission that reaches the sensor whether that sensor be film or CCD or CMOS or...
The two lenses I mentioned showed significant differences in shutter speed needs to achieve proper exposure. I feel it's going to vary from lens to lens.
R
Rob Patterson
Guest
The Pentax didn't have a meter. The old Canon did but I didn't like it's performance. I used a handheld meter for both.
The conclusion that I must draw from this thread at this time is that there can be some drastic differences in light transmission from lenses of different manufacturers of the same focal length. Probably something that won't commonly be encountered but the possibility remains.
Those two lenses I cited required noticeable shutter speed differences or aperature settings to achieve the same exposure.
This was in the days far before digital photography was even a gleam in the eyes of the designers. The film from both cameras was developed in chemicals where temperature was controlled within two or three degrees. Not enough for variances in exposure with black and white. Color yes. But i wasn't shooting in color then.
I wish I had the two cameras and the two lenses now accompanied by a good drum scanner. Then again maybe not. I'll just leave it as is. It's no longer important.
Keyboard shortcuts
- f
- Forum