Sigma Announces 18-35 f/1.8 DC HSM Art for APS-C

[CarlTN]

There has been a lot of nonsense on this thread. It's pointless contributing now because the ones spreading the nonsense have entrenched themselves so deeply that they have no prospect of coming out of this with any dignity or credibility.

Let's not say these people have no dignity...that's personal and uncalled for. They're simply sharing their opinion.

However, I agree...the discussion has turned rather silly it seems. And there's nothing "epic" about this lens. "Epic" is such a silly slang word these days, overused by teenage pop culture...it's time to retire it.

via google: ep·ic

/ˈepik/

Noun

A long poem, typically derived from oral tradition, narrating the deeds and adventures of heroic or legendary figures or the history of...

http://www.merriam-webster.com/dictionary/epic

The word really does not apply. There actually have been other fast zooms, such as Olympus' f/2. Just because this is the first f/1.8 zoom, does not make it "epic".

I mean, the above exchanges seem to have to do with light intensity, etc. A crop sensor uses only part of a full frame lens' image circle. Since the lens in question, is designed for a smaller image circle, then it can't really be compared to a lens that is designed to output a larger image circle to a larger sensor. Can it? Sure, you can attempt to calculate, and approximate...but does it really matter as much as those above think it does? No.

Let's just all go full frame, and forget crop sensors, especially for anything less than telephoto focal lengths (such as over 100mm).

 

[AdamJ]

There has been a lot of nonsense on this thread. It's pointless contributing now because the ones spreading the nonsense have entrenched themselves so deeply that they have no prospect of coming out of this with any dignity or credibility.

Let's not say these people have no dignity...that's personal and uncalled for. They're simply sharing their opinion.

Yes, I shouldn't have put it that way. Apologies.

(Although, describing what f/1.8 means - whether on FF or APS-C - ought to be a matter of fact, not opinion).

 

[9VIII]

Exactly. And you have to decide if we're talking about focal length or angle of view. You can't freely interchange the two concepts, because they are two different things. You can compare one to the other, but within limits.

Sensor dimensions taken from: http://en.wikipedia.org/wiki/Full-frame_digital_SLR
FoV/Focal Length formulas: http://paulbourke.net/miscellaneous/lens/

Calculations:
(Using an 81mm lens here to be more accurate, 36/22.2 = 1.62 crop factor)

APSC sensor: 22.2mm width, focal length 50mm = 0.4369 radians = 25.0 degrees
Full Frame sensor: 36mm width, focal length 81mm = 0.4373 radians = 25.1 degrees

Focal length and Field of View are the same thing. The only reason they might not match in practice is because manufacturers don't quite tell the truth and the advertised focal length is slightly different from what they say (if it's 52mm, they're going to advertise it as a 50mm anyway)

A Nikon 1 10mm f/2.8 lens will have an equivalent focal length (read: angle of view) of a 27mm lens on FF, but it will be nevertheless a 10mm lens. Assuming transmission is the same, the exposure would be the same for this lens and a 10mm F/2.8 lens on FF.

A Nikon 1 10mm f/2.8 lens will have a pupil diameter of 3.57mm.

A 27mm f2.8 lens will have a pupil diameter of 9.64mm.

There are 2.7^2=7.3 times as many photons are coming through that 27mm lens as through the 10mm lens at the same f stop. Since the same proportion of those photons falling on the sensor is the same [same aspect ratios fitting in a circular aperture], the full frame sensor is receiving 7.3 times as many photons at the same f-stop.

The fact that one sensor is larger or smaller is completely irrelevant -- both sensors are covering the same proportion of the image circle (because we're talking about 35mm equivalent focal lengths here -- if you want to argue proportion of image circle covered then you need to do the comparisons using the same focal lengths -- see below)

The Nikon 1 lens can be smaller because the sensor is smaller. The total amount of light gathered by a FF will be more, but the amount of light hitting the sensor / surface of the sensor would be the same.

See above -- the amount of light (ie photons) hitting the sensor is most definitely not the same. The physical aperture size and the field of view are the only 2 things that control how many photons go through the lens.

You could mount the FF lens on the Nikon 1 and it would be the same - because the extra light will fall off the sensor.

Huh? A Nikon 1 sensor is 2.72x2.72 = 7.44 times smaller -- 87% of photons that would be hitting a FF sensor are being ignored. Assuming sensors are equivalent technology, at the same shutter speed the simple fact is that the Nikon 1 will have to amplify the signal coming off the sensor 7 times more than the full frame camera to get a proper exposure. How is this possibly equivalent?

That's why exposure is not affected.

More photons hitting the sensor = more electrons being excited = less amplification necessary to properly expose an image = less amplification of background sensor noise.

Note that I deliberately haven't mentioned ISO settings at all -- ISO settings are just calibrated labels to measure how much sensor signal amplification the camera needs to perform to properly expose a scene to a certain level, they are not an inherent property of the sensor itself. Go to dxomark and compare the SNR on a Canon 5D2 and a 60D -- the 60D is far noisier despite having a newer sensor, because ISO 1600 on a 60D actually means a lot more signal amplification than ISO 1600 on a 5D2. The ISO number itself is irrelevant, what is relevant is how much noise is amplified along with the useful signal.

If you now transpose this concept to this lens, you will understand that exposure-wise a f/1.8 lens is a f/1.8 lens disregarding of the system. This, again, is because f-stops depend on focal length and focal length is a parameter of the lens and not of the camera.

Yes, a f/1.8 lens is a f/1.8 lens regardless of the system. A f/1.8 lens on a full frame will perform better than a f/1.8 lens on crop with equivalent focal lengths though, and that's what the original question was.

Imagine that tomorrow I introduce a new camera system for amateur astronomers called "Super 100x35mm", sensor dimensions 360mm x 240mm, same ~4000x2800px resolution as a crop system. I stick 35mm-equivalent 50mm focal length lens on it, which for this system would be a 500m f/2.8. The pupil diameter of this lens is 178mm -- this pupil is 10 times wider than a full frame lens. According to the logic above, the fact that this pupil & sensor has 262 times more light gathering capability than a crop 32mm f/2.8 lens is irrelevant; "exposure-wise" all f/2.8 lenses are equivalent? Does this really sound sensible to you?

If that's the case, why are astronomy telescopes so wide?

• It's not to get a narrow FOV -- you could get http://pan-starrs.ifa.hawaii.edu/public/home.html FOV using a 650mm lens on a 35mm system.
• It's not resolution, you could just move the focal plane farther back to fit in more sensors -- eg 36cm could give you 1.7gigapixels with off the shelf components http://ambivalentengineer.blogspot.com.au/2012/08/argus-is.html
• It's because a wider aperture gives you more light = better low light performance

The original question was whether a crop camera with a given lens at f/2.8 performs the same as an equivalent focal length f/2.8 full frame. The answer is no, for equivalent focal lengths at the same f-stop the full frame sensor receives a lot more light and so gives less noise. In light of that, rs is right -- the lens might be nice for someone like me who only owns a crop camera but if you care terribly about low light performance then you should be on full frame.


Native sensor ISO? Let's not open *that* can of worms

Then it gets interesting when you look at the Metabones Speed Booster. I have to wonder if those kinds of abilities have been an option all along in crop sensor lens design.

 

[Radiating]

I can just imagine Canon & Nikon engineers staring at their screens in disbelief. Sigma engineers have managed to crack some code to lens making. I have the Sigma 35mm f/1.4, just tested the new Sigma 30mm f/1.4 and can't wait for this. Amazing lenses Sigma, now please give me a nice 50mm lens because I am not at all happy with my Canon 50mm. At least my Canon 85mm is great.

It's actually not difficult to make a lens like this any more than it is to make a 24-70mm f/2.8 lens for full frame.

Both will gather approximatly the same amount of light.

The amazing thing here is that Sigma has the motivation to make a lens like this. Canon have been sitting on their butts making ridiculous products for the recent past. T5i? WTF. 8-15mm L fisheye WTF?

Canon have dozens lenses that are 13 years old and a complete embarassment. I think Sigma just saw a moment of weakness and decided to obliterate the competition with these new products.

 

[AdamJ]

I've decided to offer some help to Sigma with an FAQ set for potentially confused customers.

 

[GaryJ]

I've decided to offer some help to Sigma with an FAQ set for potentially confused customers.

+1

 

[rs]

I've decided to offer some help to Sigma with an FAQ set for potentially confused customers.

+1

Think of the metabones speed booster. Imagine now that they created one which made a FF lens create an imaging circle to match the APS-C crop sensor found in a Canon camera - a 1.6x telecompressor. That way you could mount, say, a 24-70/2.8 on crop with the metabones adapter, and get an identical FoV that the lens achieves on FF.

The 24-70/2.8 would be turned into a 15-44/1.75, right?

Now, we all know f1.75 is faster than f2.8. No-one is disputing that. If you mount this lens on the crop sensor camera with the telecompressor, it will allow for more than a stop faster shutter speeds at equal ISO's. It is an f1.75 lens, and no-one can argue with that. But where does this metabones get this extra speed from? Its not magic - its just it compresses the larger image circle into a smaller one - that extra light from that larger FF image circle is now condensed down into a smaller, more intense imaging circle, and is then received by a smaller sensor. However, in total its only the same amount of light/photons coming in through the lens which hits the sensor. The FF sensor and the APS-C sensor with a telecompressor both receive an identical number of photons, but the APS-C sensor has brighter light presented to it - more light per area - in other words its just over a stop brighter.

ISO's are rated to make exposure calculations work. What one camera does to achieve ISO 1600 isn't the same as another camera at ISO 1600 - especially with different size formats. The larger sensor as a whole has more photons hitting it at a particular aperture, so it needs to amplify the resulting electrical signal less for any given ISO. And even some cameras with the same sensor have to do different amplification, such as the Sony NEX 7 and Sony SLT A77.

If you can't see that, haven't you ever wondered why FF sensors are typically just over a stop better than crop sensors when it comes to noise? This faster aperture of the Sigma simply allows the noisier sensor to work at lower ISO's to finally fight back. Use a 18-35/1.8 at 35mm f1.8 1/100th of a sec ISO 10000 on crop, or a 24-70/2.8 at 56mm f2.8 1/100th, ISO 25600 on FF and you'll find its the same framing, depth of field, exposure and noise. (well, it would be if the Sigma was slightly brighter at f1.75)

Regardless of how you understand this difference between full frame and crop, and either agree or disagree with me, this Sigma lens really ups the game for crop users. From the specs point of view (and samples images), it looks great.

 

[AdamJ]

Think of the metabones speed booster. Imagine now that they created one which made a FF lens create an imaging circle to match the APS-C crop sensor found in a Canon camera - a 1.6x telecompressor. That way you could mount, say, a 24-70/2.8 on crop with the metabones adapter, and get an identical FoV that the lens achieves on FF.

The 24-70/2.8 would be turned into a 15-44/1.75, right?

Now, we all know f1.75 is faster than f2.8. No-one is disputing that. If you mount this lens on the crop sensor camera with the telecompressor, it will allow for more than a stop faster shutter speeds at equal ISO's. It is an f1.75 lens, and no-one can argue with that. But where does this metabones get this extra speed from? Its not magic - its just it compresses the larger image circle into a smaller one - that extra light from that larger FF image circle is now condensed down into a smaller, more intense imaging circle, and is then received by a smaller sensor. However, in total its only the same amount of light/photons coming in through the lens which hits the sensor. The FF sensor and the APS-C sensor with a telecompressor both receive an identical number of photons, but the APS-C sensor has brighter light presented to it - more light per area - in other words its just over a stop brighter.

I confess this the first time I've heard of a metabones speed booster but I get the idea and, yes, I'm with you so far.

ISO's are rated to make exposure calculations work. What one camera does to achieve ISO 1600 isn't the same as another camera at ISO 1600 - especially with different size formats. The larger sensor as a whole has more photons hitting it at a particular aperture, so it needs to amplify the resulting electrical signal less for any given ISO. And even some cameras with the same sensor have to do different amplification, such as the Sony NEX 7 and Sony SLT A77.

No, you've broken the sequence of logic from your metabones analogy. You can only say that the amount of light hitting the FF sensor is more because the FF sensor is bigger. The amount of light hitting each pixel is exactly the same, regardless of sensor size.

If you can't see that, haven't you ever wondered why FF sensors are typically just over a stop better than crop sensors when it comes to noise? This faster aperture of the Sigma simply allows the noisier sensor to work at lower ISO's to finally fight back. Use a 18-35/1.8 at 35mm f1.8 1/100th of a sec ISO 1000 on crop, or a 24-70/2.8 at 56mm f2.8 1/100th, ISO 1600 on FF and you'll find its the same framing, depth of field, exposure and noise. (well, it would be if the Sigma was slightly brighter at f1.75).

.

This is a compound error from the previous incorrect statement. Sensor size is not a factor in this.

 

[rs]

This is a compound error from the previous incorrect statement. Sensor size is not a factor in this.

My original argument which has been hotly debated was merely that as nice as this Sigma 18-35/1.8 is on a crop camera, it's not quite as nice* as a 24-70/2.8 II is on a full frame camera. I was comparing two complete systems, sensor and all. I know that a 24-70/2.8 is no match for this Sigma when they're both mounted on a crop body.

*by nice, I mean the 24-70 on FF goes wider, longer, is capable of a vaguely narrower DoF and capturing vaguely more light.

My argument is simply the total quantity of light a system can capture is more than just aperture - it is a combination of aperture and sensor size. The entrance pupil size when both systems have an equivalent FoV is a simple way of quantifying that.

But if you disagree with me and think that an f1.8 lens on 1.6x crop captures more light than a lens with an equal field of view at f2.8 on full frame, then fair enough. I've tried explaining this concept in many different ways, and you still don't get it. So I give up.

 

[Wildfire]

But if you disagree with me and think that an f1.8 lens on 1.6x crop captures more light than a lens with an equal field of view at f2.8 on full frame, then fair enough. I've tried explaining this concept in many different ways, and you still don't get it. So I give up.

Why do you care how much light the lenses capture? That's irrelevant. What I care about is the final exposure.

With an f/1.8 lens, the final exposure will be brighter than with an f/2.8 lens no matter what (even if the f/2.8 lens is "capturing" more light, it doesn't "capture" enough to be brighter than the f/1.8 lens in-camera). This is true regardless of whether you use one lens on crop and one on FF, or both on FF, or both on crop.

My original argument which has been hotly debated was merely that as nice as this Sigma 18-35/1.8 is on a crop camera, it's not quite as nice* as a 24-70/2.8 II is on a full frame camera.

That is your subjective opinion. The fact of the matter is that the lenses will have their different uses and there will be situations where each lens is better than the other. Because the Sigma is an f/1.8 lens, it may be quite possible that the 24-70 does not create a bright enough exposure for a photographer in a low-light situation, so the Sigma will be much "nicer" to use in that situation.

 

[rs]

Why do you care how much light the lenses capture? That's irrelevant. What I care about is the final exposure.

Because the Sigma is an f/1.8 lens, it may be quite possible that the 24-70 does not create a bright enough exposure for a photographer in a low-light situation, so the Sigma will be much "nicer" to use in that situation.

What about using the FF system at a higher ISO? As the FF sensors larger area allows it to capture 2.56x more light, you can use an ISO 2.56x higher (just over a stop), without suffering from any more noise than the crop sensor. ISO 10,000 on a typical APS-C sensor gives the same noise as ISO 25,600 on a typical FF. If you do choose to make use of the higher ISO's made available to you, the final exposure is the same, and the f1.8 crop system offers no low light advantage over an f2.8 full frame system.

 

[AdamJ]

What about using the FF system at a higher ISO? As the FF sensors larger area allows it to capture 2.56x more light, you can use an ISO 2.56x higher (just over a stop), without suffering from any more noise than the crop sensor.

You're trolling, surely.

ISO 10,000 on a typical APS-C sensor gives the same noise as ISO 25,600 on a typical FF. If you do choose to make use of the higher ISO's made available to you, the final exposure is the same, and the f1.8 crop system offers no low light advantage over an f2.8 full frame system.

There is some truth in this part of your statement but it has nothing whatsoever to do with the dimensions of the sensor.

 

[rs]

There is some truth in this part of your statement but it has nothing whatsoever to do with the dimensions of the sensor.

So you're saying pretty much all FF sensors are better than pretty much all crop sensors because of something other than their bigger size? I'm confused by your logic now

ps - it seems like you're slowing coming around to understanding my argument. Using that metabones analogy, you got the bit about how both the combined effect of aperture and image circle is equal to the total amount of light coming through the system. You now seem to get the other end of the argument - FF allows you to work at higher ISO's than crop. If you could just accept this missing part of the puzzle about it being the larger size of the sensor which allows you to work at these higher ISO's, I think you'll have it. Or are you going to start arguing about some other random part of the rationale?

 

[dirtcastle]

What about depth of field? Is the depth of field always in a 1:1 relationship with the f-stop? In other words, will this lens get equivalent depths of field of an f/2.8 lens on a full-frame camera?

 

[LOALTD]

I've decided to offer some help to Sigma with an FAQ set for potentially confused customers.

+2

So much mis-information on this, nice work!

 

[AdamJ]

There is some truth in this part of your statement but it has nothing whatsoever to do with the dimensions of the sensor.

So you're saying pretty much all FF sensors are better than pretty much all crop sensors because of something other than their bigger size? I'm confused by your logic now

Bingo!

If I use masking tape around the edges of my full size sensor to leave only an APS-C sized area, what does it do to the quality of the image on the exposed area? Nothing. What does it do to the exposure value? Nothing.

The reasons why FF sensors typically produce less noisy images in practice than APS-C sensors are that a) the (usually) bigger individual pixels produce a better signal-to-noise ratio, and b) the native image requires less enlargement when printing.

ps - it seems like you're slowing coming around to understanding my argument. Using that metabones analogy, you got the bit about how both the combined effect of aperture and image circle is equal to the total amount of light coming through the system. You now seem to get the other end of the argument - FF allows you to work at higher ISO's than crop. If you could just accept this missing part of the puzzle about it being the larger size of the sensor which allows you to work at these higher ISO's, I think you'll have it. Or are you going to start arguing about some other random part of the rationale?

No, I'm content to stick with facts rather than come round to your argument.

 

[rs]

There is some truth in this part of your statement but it has nothing whatsoever to do with the dimensions of the sensor.

So you're saying pretty much all FF sensors are better than pretty much all crop sensors because of something other than their bigger size? I'm confused by your logic now

Bingo!

If I use masking tape around the edges of my full size sensor to leave only an APS-C sized area, what does it do to the quality of the image on the exposed area? Nothing. What does it do to the exposure value? Nothing.

Print a whole FF image at 1m wide, the noise hides quite well. Now take a crop of that same image and print that crop at 1m wide, do you think the noise will be no less apparent?

You'd need to shoot the cropped image at a lower ISO to show an equivalent amount of noise in the final print.

 

[AdamJ]

There is some truth in this part of your statement but it has nothing whatsoever to do with the dimensions of the sensor.

So you're saying pretty much all FF sensors are better than pretty much all crop sensors because of something other than their bigger size? I'm confused by your logic now

Bingo!

If I use masking tape around the edges of my full size sensor to leave only an APS-C sized area, what does it do to the quality of the image on the exposed area? Nothing. What does it do to the exposure value? Nothing.

Print a whole FF image at 1m wide, the noise hides quite well. Now take a crop of that same image and print that crop at 1m wide, do you think the noise will be no less apparent?

You'd need to shoot the cropped image at a lower ISO to show an equivalent amount of noise in the final print.

OK, now I know you're trolling. You failed to quote the part of my post that answers this:

The reasons why FF sensors typically produce less noisy images in practice than APS-C sensors are that a) the (usually) bigger individual pixels produce a better signal-to-noise ratio, and b) the native image requires less enlargement when printing.

 

[jemping]

Not sure if anyone posted this already, but here is the website that has the sample images for this lens.

http://lcap.tistory.com/entry/Sigma-ART-18-35mm-f18-Preview

 

[rs]

OK, now I know you're trolling. You failed to quote the part of my post that answers this:

The reasons why FF sensors typically produce less noisy images in practice than APS-C sensors are that a) the (usually) bigger individual pixels produce a better signal-to-noise ratio, and b) the native image requires less enlargement when printing.

So you agree that a larger enlargement makes noise more visible? If so, it seems like we're agreeing on this, but arguing about semantics.

From my point of view, amplification and magnification amount to one and the same thing. Take two different sized sensors with the same number of MP (1D X and 7D for example), then the smaller sensor clearly has smaller pixels. Each pixel receives less light, so in order to give the same electrical signal to create a calibrated ISO 100, it has to amplify to a greater level the smaller signal created from the smaller number of photons collected at its smaller pixel. However, take two sensors of different size with pixels of the same density (eg 1Ds mk III and 30D, or D800 and D7000), and on a pixel level, each pixel is the same size, so it collects the same number of photons, creates the same strength electrical signal and requires the same amount of amplification. But it is magnified less from that larger sensor (each pixel, together with its noise is a less significant part of the whole image), so it's noise has a lower effect on the whole image. Not all FF and crop comparisons fall into one of these two convenient categories, so it's usually a combination of amplification and magnification differences between the two. But one thing is for sure - take the whole image, and the bigger sensor will be less noisy. And all things being equal (same generation technology etc), you'll find the noise ratio is directly in line with the area ratio.

Similarly speaking, a 1.6x crop of FF is no different from a 1.6x crop sensor. The exposure doesn't differ with different sized sensors. The noise does. Bigger area to capture light equals more light captured. That equals lower noise. Or you can turn it around by bumping up the ISO and create the same noise with greater sensitivity. That is what makes an f2.8 lens on FF equal to an f1.75 lens on 1.6x crop.