[i83N]

Then you are loosing the FF advantage the reason for buying it in the first place...
Besides, you can crank up ISO on APS similarly as well with correspondingly low IQ loss. Comparing formats at different ISO is comparing apples and oranges.

I wasn't talking about size, I was talking about resolution.

[jsherman999]

The weight and size comes from it kind of being a 28-50/2.8 FF lens with a speed booster bolted on the end. Had they "converted" a 24-70/2.8 lens, they probably had ended up with something like the size and weight of Pentax DA* 60-250/4.

Yes, or larger.

[kenyee]

1- sigma is clearly abandoning pentax mount for new release...3 out of 4 are not in pentax mount and the most useful fo us.

This actually has me worried. First Tamron abandons Pentax (understandable since they probably don't want to reverse engineer Pentax's SDM tech) and now it looks like Sigma might release fewer lens for Pentax. I wonder what their Pentax sales look like?

[goubejp]

Moore's law applies to sensors too you know, so higher ISO without IQ loss is not a big problem anymore.

May be you are not aware of the quantic nature of the light; this mean that for every pixel, only on limited number of photons hit each pixel of the sensor; the higher the iso, or the higher pixel count, the less photons that are collected by each cell. An ideal sensor convert each photon into 1 electron, and the current sensors are already very close from this ideal sensor. So the limit is here, when a variation of +- 1 photon on the pixel has a significant impact on the output voltage. IE assume that for a given pixel density and iso, only 128 photons hit each pixel. One more photon increase the level of 1 %, this lead that the signal vs noise ration cannot be better than 1/128 -> 20log10(128) = 42dB or 7EV; significant drop of IQ will arise when iso increase : double the iso, there are only 64 photons left, S/N = 36 dB or 6 EV and so on...
The better improvement will not result by moore law but by a sensor able to separate for each pixel the red, blue and green photons and measure the result. Currently with the bayer array the filter stops 2/3rd of the collected photons. Hope I have been clear enough

[gazonk]

Then you are loosing the FF advantage the reason for buying it in the first place...

But the point in this case is the other way round: By buying a lens like this one, you're losing the APS-C advantage, because you get something as heavy as those 24-70/2.8 zooms and you gain nothing but lose zoom range.

[Macario]

But you can simply bump ISO if you need that shutter speed - you have at least one stop advantage in order to do so.

What we see here is a manufacturer trying very hard to match what's already natively available for FF, and it's fantastic that Sigma can come this close - but would you want to buy an 18-35 f/1.8 lens for an aps-c camera? It's going to be very large, probably very expensive, probably not great IQ wide-open at the long end, and have a pretty limited range.

Are those characteristics going to play well in the aps-c space?

At some point, I think, if this lens appeals to you - you have to look at FF, because then all the lenses have this type of advantage!



.

Well, it probably makes more sense to put this lens on APS-C DSLR, than putting all those large lenses on NEX bodies.

But, how ever you look at it, the f-stop still stands for the amount of light that will fall on a sensor (or film). DOF is just a part of it all. So Pal is right, it is not an equivalent for an FF 2.8 other than DOF perception.

And how obsessed people these days are with DOF, like there is no photography without razor thin DOF. If you really want razor thing DOF, then shoot large format wide open.

And let's not speculate on IQ before we see shots made with this lense. Maybe at the long end it will be very goods (as yesterday it was impossible to have a f1.8 zoom lens, but here we are and it is possible). The only thing that has me a bit worried, how did they handle distortion. That has always been my concern a bit with Sigma lenses.

[Rondec]

You can still shoot with the same shutter speed because you can use higher ISO without IQ loss on the larger sensor.

The thing that I don't see is that this seems to be true with regard to noise, but not so much with regard to dynamic range. Dynamic range seems to be based on sensor technology and not so much on format, unless I am reading DXO Marks graphs all wrong.

[i83N]

May be you are not aware of the quantic nature of the light; this mean that for every pixel, only on limited number of photons hit each pixel of the sensor; the higher the iso, or the higher pixel count, the less photons that are collected by each cell. An ideal sensor convert each photon into 1 electron, and the current sensors are already very close from this ideal sensor. So the limit is here, when a variation of +- 1 photon on the pixel has a significant impact on the output voltage. IE assume that for a given pixel density and iso, only 128 photons hit each pixel. One more photon increase the level of 1 %, this lead that the signal vs noise ration cannot be better than 1/128 -> 20log10(128) = 42dB or 7EV; significant drop of IQ will arise when iso increase : double the iso, there are only 64 photons left, S/N = 36 dB or 6 EV and so on...
The better improvement will not result by moore law but by a sensor able to separate for each pixel the red, blue and green photons and measure the result. Currently with the bayer array the filter stops 2/3rd of the collected photons. Hope I have been clear enough

Have you seen how much space on sensors is left unused because pixels are round and big. I assure you there are plenty of photons that aren't used. Light is also an electromagnetic wave...

[gazonk]

The thing that I don't see is that this seems to be true with regard to noise, but not so much with regard to dynamic range. Dynamic range seems to be based on sensor technology and not so much on format, unless I am reading DXO Marks graphs all wrong.

If you compare similar sensors, it's very close to one step DR difference except at the lowest ISO.

See for instance Nikon D7000 vs. Nikon D800:

DxOMark - Compare cameras side by side

For instance, ISO 1600 DR for D7000 is almost exactly equal to ISO 3200 DR for D800. The D800 "cheats" a little more on ISO than the D7000, so in reality it's a little less than one step, but not much.

[Cynog Ap Brychan]

This actually has me worried. First Tamron abandons Pentax (understandable since they probably don't want to reverse engineer Pentax's SDM tech) and now it looks like Sigma might release fewer lens for Pentax. I wonder what their Pentax sales look like?

I don't know about Tamron abandoning Pentax. That (approx.) 12 mm to (approx.) 28 mm on the Pentax road map for 2013 (or later) looks remarkably like the recently released Tamron AT-X 12-28 Pro DX lens.

[L33tGreg]

Does this announcement make anyone else think of the DA20-35ltd ? It should be out sometime this year....

[RonHendriks1966]

Well this is a very hefty lens. I don't get it, why not 18-35mm/f2.0, that already would make the lens lighter and still unique. Who is it for? Well basicly for toprated aps-c camera's! So for canon 7D II and nikon D400 I guess. I don't see any rebel-shooters buying this. As for the K-mount, I don't see many people buy this unless they didn't complain about the price for the 5D IIs.

[Jonathan Mac]

Wow, an f/1.8 constant zoom. Now that looks fantastic, though big (72mm thread!!) and narrow zoom range. No mention of a Pentax mount though

If it were available and decent wide open then I would very likely get one. It would have to be good wide open though, otherwise it would be pointless over my Tammy 17-50.

[Rondec]

If you compare similar sensors, it's very close to one step DR difference except at the lowest ISO.

See for instance Nikon D7000 vs. Nikon D800:

DxOMark - Compare cameras side by side

For instance, ISO 1600 DR for D7000 is almost exactly equal to ISO 3200 DR for D800. The D800 "cheats" a little more on ISO than the D7000, so in reality it's a little less than one step, but not much.

Exactly. So it is dependent on sensor tech (assuming the D800 uses the same tech that is in the D7000). So, if you mount a 28-70 on a 5D MKIII and compare it to a 18-35 f1.8 on a D7000, your dynamic range is going to be a lot worse on the 5D.

[gazonk]

Exactly. So it is dependent on sensor tech (assuming the D800 uses the same tech that is in the D7000). So, if you mount a 28-70 on a 5D MKIII and compare it to a 18-35 f1.8 on a D7000, your dynamic range is going to be a lot worse on the 5D.

Sure. But you can also say that if you mount the 18-35 on a Canon, your dynamic range is going to be a lot worse than if you mount it on a Nikon.