How will new ultra high ISO bodies impact f/2.8 telephoto zoom lenses?

[GoldenEagle]

I apologize in advance to the cross-posting gods, as this comment was buried at the end of a 93-comment thread that died last week. I believe it deserves its own thread, as it is an important aspect of the new bodies coming out...

What I haven't seen discussed re: 5DM3 is the impact of higher useable ISO for lowlight shooters on the entire Canon L-series telephoto lens line. Currently, using Lightroom 3 noise reduction (@~35% slider), I can use 5000 ISO on my 5DM2 all day long, with very little noise impact on IQ. If the 5DM3 can get me a 2-stop ISO improvement, that equates to around 20,000 useable ISO. What that means, from a lens acquisition/ownership perspective:

A) I could work with a 70-200mm f/4L ($600/$1100 w/IS) instead of a 70-200mm f/2.8L ($2200/$2400) and still have sufficient shutter speed (1/200 or better) to freeze most subject movement, low light, no flash.

B) Similar comparison at 300mm for f/2.8L vs f/4.0L: $7300/5.2lb vs $1300/2.6lb. Savings=$6,000, plus no monopod required!

C) Even better, at 400mm focal length, with a 20,000 ISO, couldn't I get a 400mm f/5.6L for $1,200, instead of a 400mm f/2.8L for $12,000? I have neither budget nor the desire to drag an 8- to 11-lb monster around all night. With the 400mm f/5.6L, it's only 2.8 lbs, less than the 70-200 f/2.8L!

What I'm seeing potentially is the ultra-high ISO available in a 5DM3 (plus 1DX, potentially 7D2) making the incredibly expensive and heavy f/2.8L telephotos only being used by the elite/rich sports/outdoor shooters, allowing a lot of less-financially well-off shooters to produce still-great imagery using far less expensive and lighter f/4.0L-f/5.6L telephoto lenses.

Spending $1-2K on a 5D body upgrade from M2 to M3, just for the ISO alone, could/would pay for itself and then some, immediately in lens savings, wouldn't it? What am I missing?

Give me ISO! GE

 

[neuroanatomist]

A) I could work with a 70-200mm f/4L ($600/$1100 w/IS) instead of a 70-200mm f/2.8L ($2200/$2400) and still have sufficient shutter speed (1/200 or better) to freeze most subject movement, low light, no flash.

B) Similar comparison at 300mm for f/2.8L vs f/4.0L: $7300/5.2lb vs $1300/2.6lb. Savings=$6,000, plus no monopod required!

C) Even better, at 400mm focal length, with a 20,000 ISO, couldn't I get a 400mm f/5.6L for $1,200, instead of a 400mm f/2.8L for $12,000? I have neither budget nor the desire to drag an 8- to 11-lb monster around all night. With the 400mm f/5.6L, it's only 2.8 lbs, less than the 70-200 f/2.8L!

...What am I missing?

Aperture determines more than shutter speed. Most people don't buy an 85mm f/1.2L to use it in very low light (which turns out to be a good thing, because unknown to most people, when you shoot at f/1.2 on a dSLR your camera is actually bumping up the ISO to compensate for the fact that the sensor is less sensitive to light at very high incident angles, and not telling you...meaning over 1/2 a stop more noise in some cases). No, most people buy an 85L for the large amount of OOF blur you can achieve with a fast lens. For example, I can use a 70-200/2.8 quite effectively for portraits, whereas f/4 would not be nearly as effective for those shots.

Also, the image quality from a 300/2.8 or 400/2.8 is much better than that of the corresponding 'consumer' prime lenses. Even there, the high ISO is a benefit, because a sports shooter, for example, could choose to shoot at f/5.6 instead of f/2.8, yielding more DoF if needed.

But from a shutter speed perspective, you're quite correct - the increasingly better high-ISO performance of today's cameras definitely means you can often use an f/4 or f/5.6 lens in situations that used to require a fast prime or a flash.

 

[briansquibb]

C) Even better, at 400mm focal length, with a 20,000 ISO, couldn't I get a 400mm f/5.6L for $1,200, instead of a 400mm f/2.8L for $12,000? I have neither budget nor the desire to drag an 8- to 11-lb monster around all night. With the 400mm f/5.6L, it's only 2.8 lbs, less than the 70-200 f/2.8L!

I carry around the 400 f/2.8 and the 70-200 f/2.8II

This is what you might miss

 

[sheedoe]

Aperture determines more than shutter speed. Most people don't buy an 85mm f/1.2L to use it in very low light (which turns out to be a good thing, because unknown to most people, when you shoot at f/1.2 on a dSLR your camera is actually bumping up the ISO to compensate for the fact that the sensor is less sensitive to light at very high incident angles, and not telling you...meaning over 1/2 a stop more noise in some cases).

Hmmm interesting. I never knew the camera did that. Now, when you say that the camera bumps up the ISO without telling you, does it only apply in settings where the camera meters for exposure, or does it do it for manual exposures as well? In other words, if I choose an ISO of 200 manually, would the camera actually shoot with ISO ~300?

 

[neuroanatomist]

Now, when you say that the camera bumps up the ISO without telling you, does it only apply in settings where the camera meters for exposure, or does it do it for manual exposures as well? In other words, if I choose an ISO of 200 manually, would the camera actually shoot with ISO ~300?

Yes, it will do that on manual as well, and would report the setting you chose, e.g. ISO 200, in the EXIF.

For another example, consider Highlight Tone Priority. When you turn on HTP, you will notice that you can no longer shoot at ISO 100. That's because when you shoot at ISO 200, the camera is really shooting at ISO 100, then bumping the exposure up a stop with selective preservation of the highlights (i.e. a tone curve is applied) before writing the file (RAW and/or JPG), and recording the ISO as 200 in the metadata.

 

[kubelik]

What I'm seeing potentially is the ultra-high ISO available in a 5DM3 (plus 1DX, potentially 7D2) making the incredibly expensive and heavy f/2.8L telephotos only being used by the elite/rich sports/outdoor shooters, allowing a lot of less-financially well-off shooters to produce still-great imagery using far less expensive and lighter f/4.0L-f/5.6L telephoto lenses.

I should think not. as neuro and brian pointed out, the purpose of f/2.8 or wider apertures isn't necessarily to help you shoot in lower light; the point is to achieve shallow DOF. it's why consumer primes like the nifty fifty exist.

also, as neuro points out, having a wider native aperture to a lens allows for more light to hit the autofocus sensor, which means increased accuracy in lower light regardless of your ISO setting.

yet another point is lens sharpness. most lenses aren't at their sharpest at their max aperture setting. many f/4 lenses compare favorably in sharpness to f/2.8 lenses when compared at max aperture, but stop a f/2.8 lens down to f/4 to see apples-to-apples and it's likely to be sharper than the f/4.

lastly, I believe that until there's a significant breakthrough in the materials engineering of the silicon used in sensor production (which is likely to require an entirely different composition or type of silicon), you're unlikely to see continued, massive upward growth of high-ISO capability. the 1DX is proclaimed to have about a stop better performance than the 1DsIII, but they had to shave away 3 megapixels and it's only at the JPG level. whether RAW improvement after 5 years even comes close to 2/3-stop is ... unlikely.

 

[sheedoe]

For another example, consider Highlight Tone Priority. When you turn on HTP, you will notice that you can no longer shoot at ISO 100. That's because when you shoot at ISO 200, the camera is really shooting at ISO 100, then bumping the exposure up a stop with selective preservation of the highlights (i.e. a tone curve is applied) before writing the file (RAW and/or JPG), and recording the ISO as 200 in the metadata.

Another fact I didn't know. Great infos, thanks!

 

[JonJT]

As has been said previously, a lot of those fast 2.8 zooms offer better image quality than their slower cousins. In addition, the wider aperture offers more than just a full stop lower ISO to the shooter, you have the DOF effects and faster focusing as well.

I, personally, don't even think the ISO capabilities of newer DSLR bodies will be able to make up for the flexibility of having a 2.8 zoom. When shooting in low light with one of these things, that extra stop of aperture can make the difference between getting a shot with lots of noise, or not, and/or getting a shot with a sharp subject, or not.

 

[neuroanatomist]

I think our expectations will just increase in parallel. 20 years ago, I expected that I needed to change film rolls when going from bright, sunny outdoors to dim, tungsten-lit indoors. Today, I expect to be able to shoot by moonlight and have the images look like they were shot at high noon. In 5 years, maybe I'll expect to shoot a football game lit only by a spectator at each end zone holding a cigarette lighter. But I'll still need that f/2.8 lens to do it... :

 

[unfocused]

This may be one reason why Canon and other manufacturers seem to be placing more and more emphasis on high ISO instead of fast lenses.

Having grown up on prime lenses (and film) I'm always a little amused that today we think of f2.8 as a "fast" lens. But, it seems manufacturers have determined that they will compromise on lens speed and put more emphasis on ISO. I'm sure it comes down to cost/benefit analysis.

As the long lenses you mention demonstrate, the marginal cost of adding an extra stop of speed is far greater than the marginal cost of adding added ISO sensitivity to a sensor.

But, I do wonder if it's always the best choice. The new G1-X is a good example. If you read the interviews and literature, it's clear that Canon is arguing that the bigger sensor and improved ISO compensates for the slower lens. But, isn't the opposite true as well. Does the faster lens of the Fuji X-10 for example compensate for the smaller sensor? I suppose the only way to know will be when both models are available and we can see side by side comparisons.

Speaking of sensors, there is also another factor at play here that has an even greater impact and that is the crop factor.

For your 5D, you can get a 300 mm f2.8 for $7,300. But, with a 7D you can get an effective 320mm from the 70-200mm f2.8 zoom for $2,100. The 300mm f4 is still a cheaper option for the 5D, but again, on the 7D that's effectively almost a 500 mm f4 (actually 480mm), which is a $6,800 lens.

Two points really.

1) Right now, it comes down to what your needs and wants are. Some will prefer having the higher ISO options of a full frame should they need it. Others prefer the longer reach of the APS-C sensor. And, then there are those that need the speed of the fastest available primes, plus either the reach of the crop sensor or the high ISO sensitivity of the newest full frame sensors. And, as others have pointed out, there are also aesthetic considerations in both the speed of the lens and the size of the sensor.

2) This is one more example of the convergence that I think camera manufacturers are struggling with. As technology improves, the differences between camera bodies and lenses are getting smaller and more nuanced. As you point out, it gets harder and harder to justify spending thousands of dollars for relatively small improvements in technology.

Good news for "average" enthusiasts and even professionals. We are the winners in this because it gives us more choices.

Not so sure about the true specialists. Canon threw the 1.3 crop shooters under the bus. Probably because they determined they just weren't a large enough market and improvements in APS-C and Full Frame sensors were squeezing that market at both ends. I wonder if the same might happen at some point with the big white lenses. Can Canon and Nikon keep making $10,000-plus lenses and keep finding a market for them?

 

[Maui5150]

Of course they can. Just because a crop sensor gives more reach, does not mean you get a better picture.

For me reach is nice, but reach means nothing if I have to hold the shutter open longer to get a decent picture : i.e. ISO + speedy lens can make a HUGE difference in terms of sharpness and image quality.

 

[briansquibb]

For your 5D, you can get a 300 mm f2.8 for $7,300. But, with a 7D you can get an effective 320mm from the 70-200mm f2.8 zoom for $2,100. The 300mm f4 is still a cheaper option for the 5D, but again, on the 7D that's effectively almost a 500 mm f4 (actually 480mm), which is a $6,800 lens.

I consider the 70-300L as the ff equivalent of the 70-200 f/2.8L if you start to compare the DOF - and the 70-300L is cheaper by far

This was taken on the 1Ds3 + 70-300L @f/5.6 - this is just to show the shallow dof

 

[GoldenEagle]

lastly, I believe that until there's a significant breakthrough in the materials engineering of the silicon used in sensor production (which is likely to require an entirely different composition or type of silicon), you're unlikely to see continued, massive upward growth of high-ISO capability. the 1DX is proclaimed to have about a stop better performance than the 1DsIII, but they had to shave away 3 megapixels and it's only at the JPG level. whether RAW improvement after 5 years even comes close to 2/3-stop is ... unlikely.

Great discussion, many good points made!

Re: Megapixel count as a factor in high-ISO/useable noise images, I would encourage low-light shooters to try out the sRAW1 setting. In my 5DM2, I get a 3800x2500 pixel RAW file, which is just fine for up to 20x30 poster-size prints. How many people need to print larger than that? What I gain is far less noise at 5000 ISO, (plus my CF cards & disk storage go twice as far!

So I'm a fan of smaller files and less noise captured to begin with. Then let me use post-processing to reduce even that noise even further.

 

[neuroanatomist]

Re: Megapixel count as a factor in high-ISO/useable noise images, I would encourage low-light shooters to try out the sRAW1 setting. In my 5DM2, I get a 3800x2500 pixel RAW file, which is just fine for up to 20x30 poster-size prints. What I gain is far less noise at 5000 ISO, (plus my CF cards & disk storage go twice as far!

So I'm a fan of smaller files and less noise captured to begin with. Then let me use post-processing to reduce even that noise even further.

You aren't capturing less noise. If you capture a full size RAW file, then downsample it to 3861x2574 pixels, you'll achieve the same level of (apparent) noise reduction.

 

[ferdi]

Speaking of sensors, there is also another factor at play here that has an even greater impact and that is the crop factor.

For your 5D, you can get a 300 mm f2.8 for $7,300. But, with a 7D you can get an effective 320mm from the 70-200mm f2.8 zoom for $2,100. The 300mm f4 is still a cheaper option for the 5D, but again, on the 7D that's effectively almost a 500 mm f4 (actually 480mm), which is a $6,800 lens.

Don't forget that on the 7D a f/2.8 lens has DOF f/4.5 effectively, and the 300mm f/4 becomes a 480mm with DOF f/6.4.
This is not necessarily bad, I actually love this feature on 1.6x crop bodies.

Back on subject, GoldenEagle has a point. I shoot indoor group sports at ISO 6400 and f/4 just to get a few more eyes in focus. If the AF would be up to it I would even try f/5.6. But I still prefer f/2.8 or faster lenses, not just for shallower DOF, but also because the viewfinder would be a lot darker with an f/5.6 lens.

 

[Viggo]

2,8 lenses are usually more accurate to AF over the entire frame, whilst 5,6 lenses misses out on cross- and dual cross type af-sensors.

And a point, for me at least, is that there's no problem in getting conditions too dark for a 2,8 lens even with 4 times the iso of the 1d X. And a 5,6 lens in those dim conditions doesn't work, af or shutterspeed-wise

 

[pharp]

Should think It would have more affect on the need for fast wide angle primes! Really any need for a 24mm 1.4 anymore?

 

[tt]

Is focus speed another factor?
The subjective speed of a 1DX focusing an f4 lens seems quicker. Yes, it's got the extra help, but then won't these ultra high ISO bodies also be bringing in some level of improved focusing speed?
Eg - Will the mythical 5D Mark III / 7D Mark II likely have a faster focus speed to their predecessors, using the same lenses?

 

[Viggo]

Should think It would have more affect on the need for fast wide angle primes! Really any need for a 24mm 1.4 anymore?

Have you ever used a 24 f1,4 L II on a fullframe body?? The look of the picture you get at that wide angle and that shallow depth can't be done by anything else.

 

[JonJT]

Should think It would have more affect on the need for fast wide angle primes! Really any need for a 24mm 1.4 anymore?

Have you ever used a 24 f1,4 L II on a fullframe body?? The look of the picture you get at that wide angle and that shallow depth can't be done by anything else.

Indeed. And, even if you discount the ISO bump needed at the widest apertures, the 24mm 1.4 still has an advantage over a 2.8 zoom for low light, no flash shooting. Sometimes, it will mean the difference between getting the shot and not getting it.