News: Rumors: the new Canon EOS 7D dslr and new lenses

[Cem_Usakligil]

Right now, there are all kinds of rumors about the introduction of a new Canon EOS camera, the 7D. Along with it some new lenses, most remarkably a 100mm macro L IS lens.

Here and here you can find some info and pictures.

Cheers,

 

[Asher Kelman]

Right now, there are all kinds of rumors about the introduction of a new Canon EOS camera, the 7D. Along with it some new lenses, most remarkably a 100mm macro L IS lens.

Here and here you can find some info and pictures.

Cheers,

Another x1.6 camera with too many pixels! What the world wants:

1. A full frame camera with great pixels just 9 MP would be fine, rangefinder size and silent.

2. The above with changeable lenses

3. The above with upgradable sensors.

That's it!

Asher

 

[Doug Kerr]

Hi, Cem,

Right now, there are all kinds of rumors about the introduction of a new Canon EOS camera, the 7D. Along with it some new lenses, most remarkably a 100mm macro L IS lens.

I was quite startled by the indication in the Neutralday site that the frame size will be "1.6X".

 

[Cem_Usakligil]

Hi, Cem,



I was quite startled by the indication in the Neutralday site that the frame size will be "1.6X".

Hi Doug,

Nothing escapes your sharp eye, as usual. That would make it the first MF digital camera by Canon, lol. Where can I buy one please?

Cheers,

 

[Mike Shimwell]

Another x1.6 camera with too many pixels! What the world wants:

1. A full frame camera with great pixels just 9 MP would be fine, rangefinder size and silent.

2. The above with changeable lenses

3. The above with upgradable sensors.

That's it!

Asher

1. Rumours are flying about a Leica M9 with an 18 to 24Mp full frame sensor supplied by Kodak, Canon or someone. Zeiss are rumoured to have a digital Ikon in development - presumed with Sony sensor tech.

2. M mount lenses are interchangeable.

3. No happened yet, but if they can make a digital rf work as well in sensor terms as a 1Ds3 or 5D2 then you wouldn't really need to upgrade that quickly. If they cna setp on a generation then that will give you longer. One ting that Zeiss stated at the launch of the Ikon was that they wanted to avoid drawing customers into the rapid turnover digital model (though they put it interms of protecting customers investment) - hence they need a degree of maturity before considering a digital ZI. Leica clearly saw the M8 as a stopgap.

For my part a digital Ikon or M9 (will cost too much!!) and the 1Ds3 would be perfect complements. The film ZI would likely still see some black and white, but that would be plenty.

Mike

 

[Doug Kerr]

Hi, Cem,

Hi Doug,

Nothing escapes your sharp eye, as usual. That would make it the first MF digital camera by Canon, lol. Where can I buy one please?

Well, I was of course referring (against my better judgment) to that repugnant convention in which a "1.6x" sensor has dimensions 1/1.6 times those of the full-frame 35-mm frame (as for my EOS 40D)!

Best regards,

Doug

 

[Asher Kelman]

1. Rumours are flying about a Leica M9 with an 18 to 24Mp full frame sensor supplied by Kodak, Canon or someone. Zeiss are rumoured to have a digital Ikon in development - presumed with Sony sensor tech.

2. M mount lenses are interchangeable.

3. No happened yet, but if they can make a digital rf work as well in sensor terms as a 1Ds3 or 5D2 then you wouldn't really need to upgrade that quickly. If they cna setp on a generation then that will give you longer. One ting that Zeiss stated at the launch of the Ikon was that they wanted to avoid drawing customers into the rapid turnover digital model (though they put it interms of protecting customers investment) - hence they need a degree of maturity before considering a digital ZI. Leica clearly saw the M8 as a stopgap.

For my part a digital Ikon or M9 (will cost too much!!) and the 1Ds3 would be perfect complements. The film ZI would likely still see some black and white, but that would be plenty.

Mike

Hi Mike,

I like your ideas and look forward to Leica and Zeiss giving us new models.

The G10 is absolutely silent but the image quality degrades rapidly above ISO 200. Perhaps the S90 or Canon G11 will be both silent and usable at higher ISO. My interest is being able to shoot during serious music performance and for not inserting my presence in candid photography. Ability to forgo flash is important. So far, the G10 is perfectly silent but poor image quality at ISO 1600 so pretty useless for most professional shooting I have to cover. The 5DII is fine for image quality but the shutter can be heard on stage. Even with a blimp the thing is not silenced completely. So I don't want anything with a mirror!

Asher

Note: My question of low light silent shooting moved to make a new topic here.

 

[Mike Shimwell]

Hi Mike,

I'd hope Leitz an Zeiss are prepared to use the silicon chip as the shutter and not have to rely on a mechanical one!

Asher

Canon are half way there with the 5D2 and ccd has had this ability for some time. For you an rf with live view would seem to be perfect - the best of both worlds.

A current alternative might be a panasonic G1 or Olympus EP-1. I've handled both and the pana is more to my liking, though the oly has image stabilisation. Both can take M mount lenses via an adaptor - and a Leica 90/2 becomes a 180/2 after allowing for the smal sensor. The pana isn't silent, but much much quieter than a dslr and probably than an Ikon (though I've not tried them side by side.

Mike

 

[Asher Kelman]

On the 100 m m Macro lens, for sure it's great to have a new L lens available. I'd have thought it would have been more popular at 150mm and then it could also be used for portraits. Still, I guess it will be a fine lens with a lot of following.

Still, I wonder whether Canon reads any of the online fora with the issues on their wide angle offerings. Wouldn't it seem more logical for Canon to produce a 14-24 to match Nikon's utterly superb offering.

Asher

 

[Michael Fontana]

The existing 100 mm macro - while not beeing a L - is quite a good lens, one of the best from Canon, anyway.

Therefore I don't understand the need for a new one. but I might miss some points.

Wouldn't it seem more logical for Canon to produce a 14-24 to match Nikon's utterly superb offering.

The 16-35-II isn't that old, but not a real competitor for the N 14-24.
Therefore I wouldn't expect something new in that range, in short terms.

 

[Bart_van_der_Wolf]

The existing 100 mm macro - while not beeing a L - is quite a good lens, one of the best from Canon, anyway.

I agree, it's a good lens and it's very versatile because it can be used at larger distances as well.

Therefore I don't understand the need for a new one. but I might miss some points.

While the current 100mm Macro is good, there is room for improvement. That improvement might come at a price though. I also haven't seen any MTF info or bokeh results of the new one yet, but I expect them to be better than what the current model offers (see P.S.). So for the time being, that would create 2 offerings with a popular focal length, at 2 price points. Assuming that production capacity is limited, that could create a profitable choice, and an upgrade potential.

Wouldn't it seem more logical for Canon to produce a 14-24 to match Nikon's utterly superb offering.

The 16-35-II isn't that old, but not a real competitor for the N 14-24.
Therefore I wouldn't expect something new in that range, in short terms.

I'm not so sure, but it would be nice. I think that the new TSE 17 and 24mm lenses offer a huge quality jump and the control that's needed for many architectural settings. The tilt capability is also valued by landscape photographers. The sample images of the 17mm that I see on the internet confirm my expectations, the FOV is too wide (when shifted) to prevent surreal projection/anamorphic distortion (but sometimes there is no alternative due to restricted shooting space). The 24mm turns out to be a very versatile lens, and it is (besides flat stitching) also superior for traditional stitching because it requires only a few tiles for huge FOV and high resolution with decent DOF, virually without geometric lens distortion and with very little CA, and it allows the use of filters (CPOL!).

The Nikon 14-24 G lens is very good from what I've seen. Because it doesn't offer T/S capability, it forces a (different) type of shooting that may mitigate the anamorphic distortion a bit, but there is no escape from the laws of physics. It also requires to be stopped down for DOF, but that's not an issue for architectural use (24x36mm sensors offer more sensitivity than a view camera anyway, e.g. to stop motion or camera shake).

Bart

P.S. The MTF curves are now published (old and new, both on 3rd tab), and they tell the following:
Wide open performance is improved from edge to edge, but the extreme corners are worse.
Stopped down performance is marginally more contrasty in the center than its predecessor, but the corners are worse. I'm not sure how these corners will actually look, e.g. in a reproduction.
The 9 rounded aperture blades will probably help bokeh a lot.

 

[Michael Fontana]

I agree Bart on all

especially the TSE 24 having a ideal focal lenghts for architecture for many shots..
while the 28 beeing not wide enough, the 24 is in the good range between the wides and the ultrawide, still providing a naturally looking FOV/look and not the wide angle effect of the 21 or 18. Sure, sometimes you can work arround that, but not alwith.

However, I don't see that much of the use of the TSE as classical stitch lens:
with 3 images only, you' re at 120 degs of HFOV, that's to much for most uses in architecture, corresponding to a 10 mm-lens. For landscapes that might work, though...

For flatstitching the 24 looks to be good to me - beeing a quick alternativ to real stitching or a single shot.

 

[Bart_van_der_Wolf]

However, I don't see that much of the use of the TSE as classical stitch lens:
with 3 images only, you' re at 120 degs of HFOV, that's to much for most uses in architecture, corresponding to a 10 mm-lens. For landscapes that might work, though...

Consider the following scenario.
Single row stitching by camera rotation through the entrance pupil, camera levelled in portrait orientation, lens with vertical shift to keep verticals parallel. That will give a huge vertical coverage without keystoning, also in the center, for a cylindrical or rectilinear projection of arbitrarily wide FOV (as much as the subject allows before anamorphic perspective distortion becomes objectionable). A slight fixed tilt may be used for all images, with little impact on stitching accuracy but lots of DOF.

This scenario would on a 1Ds3 or a 5D2 result e.g. in:
3:2 ratio, 3 horizontal tiles with 40% overlap, 8237x5616 pixels, 16mm equivalent FOV (H=95 degrees, V=74+ degrees). When required, the HFOV can be increased to 109 degrees (13mm equivalent) with 10% overlap to a 10483x5616 pixels file size (3.7:2 aspect ratio).
A 3-tile scenario doesn't take too much time to shoot, and bracketing still produces a manageable amount of files. The single row approach makes stitching also very predictable (only 2 seams, and a good horizon), with a simple/stable pano setup, and the final file size offers a lot of magnification potential. Another benefit of stitching is also the automatic levelling as part of the stitching process, just select a few verticals if available, and everything is squared automatically. One also has a choice of output projections to some what reduce anamorphic perspective.

Of couse, for moving subject matter one may need to use a 14mm for single shot wide angle views, and postprocessing will take less time. For that the Nikon 14-24mm G lens is ideal, and lens distortion is low enough to require very little correction (if any).

Bart

 

[Asher Kelman]

Canon's new 100 MM Macro 2.8L: A portrait lens?

I agree, it's a good lens and it's very versatile because it can be used at larger distances as well.

While the current 100mm Macro is good, there is room for improvement. That improvement might come at a price though. I also haven't seen any MTF info or bokeh results of the new one yet, but I expect them to be better than what the current model offers (see P.S.). So for the time being, that would create 2 offerings with a popular focal length, at 2 price points. Assuming that production capacity is limited, that could create a profitable choice, and an upgrade potential.


P.S. The MTF curves are now published (old and new, both on 3rd tab), and they tell the following:
Wide open performance is improved from edge to edge, but the extreme corners are worse.
Stopped down performance is marginally more contrasty in the center than its predecessor, but the corners are worse. I'm not sure how these corners will actually look, e.g. in a reproduction.
The 9 rounded aperture blades will probably help bokeh a lot.

Bart,

I found that the 50mm Macro is a great portrait lens but slow to focus. This new 100mm lens has better bokeh and wide open performance. corners off, no problem for me! So maybe this lens will be great for portraits too! But at what distance?

Asher

 

[Asher Kelman]

A classic description of the Canon 24mm T/S lens for flat 3 image stitching!

Consider the following scenario.
Single row stitching by camera rotation through the entrance pupil, camera levelled in portrait orientation, lens with vertical shift to keep verticals parallel. That will give a huge vertical coverage without keystoning, also in the center, for a cylindrical or rectilinear projection of arbitrarily wide FOV (as much as the subject allows before anamorphic perspective distortion becomes objectionable). A slight fixed tilt may be used for all images, with little impact on stitching accuracy but lots of DOF.

This scenario would on a 1Ds3 or a 5D2 result e.g. in:
3:2 ratio, 3 horizontal tiles with 40% overlap, 8237x5616 pixels, 16mm equivalent FOV (H=95 degrees, V=74+ degrees). When required, the HFOV can be increased to 109 degrees (13mm equivalent) with 10% overlap to a 10483x5616 pixels file size (3.7:2 aspect ratio).
A 3-tile scenario doesn't take too much time to shoot, and bracketing still produces a manageable amount of files. The single row approach makes stitching also very predictable (only 2 seams, and a good horizon), with a simple/stable pano setup, and the final file size offers a lot of magnification potential. Another benefit of stitching is also the automatic levelling as part of the stitching process, just select a few verticals if available, and everything is squared automatically. One also has a choice of output projections to some what reduce anamorphic perspective.

Of couse, for moving subject matter one may need to use a 14mm for single shot wide angle views, and postprocessing will take less time. For that the Nikon 14-24mm G lens is ideal, and lens distortion is low enough to require very little correction (if any).

Bart

This I'll bookmark. The most concise helpful comment for me on this new lens!

asher

 

[Bart_van_der_Wolf]

Bart,

I found that the 50mm Macro is a great portrait lens but slow to focus. This new 100mm lens has better bokeh and wide open performance. corners off, no problem for me! So maybe this lens will be great for portraits too! But at what distance?

Hi Asher,

We'll have to wait for some actual (usable) results to come in, but I expect this lens to be very good for portraits as well. The 'old' 100mm macro was already very suitable for portraits and is no slough when focussing (although a bit slower than non macro short throw lenses), I expect the new one to be even better (wider aperture performance and bokeh).

The IS for these focal lengths is also a bonus at portrait distances, it will help with a steady viewfinder composition, and with available light. I'm tempted ...

Cheers,
Bart

 

[Michael Fontana]

Bart

The new 100 mm:
initally I wanted to write about the new 100 mm macro might be adjusted for the high-MP- crop cams. But it's a completly new design; you see it if you look at the drawing below the mtf's. And I agree, for reproductions - I use it for that purpose - the old one looks better in the corner.

True, bokeh should improve with the 9 aperture blades. In that regard, the old 100 isn't a winner, the Sonnar 85 is miles ahead when looking for a nice bokeh. Creamy like a swiss alp butter ;-)

 

[Michael Fontana]

Bart again: 24 TSE & stitching:

Just to make clear, with post 13, I was talking about the architecture photography use, therefore in rectilinear projection of the TSE-24, while stiched.

Well, you did the math a bit differently, I calculated a overlap of 25% which gave that - in my eyes to wide pano. But true, you can play with the overlap as well and make it less wide, while keeping the rel. big height.

I wrote the 2nd parth of post 13 because I felt sometimes, that even my 3-vertical images-stich with the Distagon 28 gave me a HFOV to wide - 105 degs (with 25 % overlap) hence, sometimes I prefer the distagon 35, which is marvelous for stitching, (3 vertic resulting in a 90 deg-stitch) but not that much used in single shots.

I'm sure we're talking about certain personal preferencies as well - but we can't deny them.

 

[Asher Kelman]

The daughter topic on "Silent low light tele photography" is transferred to its own location, here.

 

[Mike Shimwell]

Bart again: 24 TSE & stitching:

I wrote the 2nd parth of post 13 because I felt sometimes, that even my 3-vertical images-stich with the Distagon 28 gave me a HFOV to wide - 105 degs (with 25 % overlap) hence, sometimes I prefer the distagon 35, which is marvelous for stitching, (3 vertic resulting in a 90 deg-stitch) but not that much used in single shots.

I'm sure we're talking about certain personal preferencies as well - but we can't deny them.

Completely off topic, but I think the 35 Distagon is a marvellous lens for one shot everyday use on the 1Ds3. Some of my favourite pictures are grabs with it. Colour, in and oof rendering are all beautiful It's a bit harder to focus on a 5D though.

Mike

 

[Michael Fontana]

Michael

I agree about the little underrated marvel; in my holidays, I just had the Distagon 35 and a Sonnar 85.. a light but very nice combination. I've a auto confirm-adapter on it, thus no problems when focusing.

In my jobs, I need it just wider, which is done when using it for stitching.

 

[Bart_van_der_Wolf]

Example of TS-E 24mm II, 3 rotated image stitch

This I'll bookmark. The most concise helpful comment for me on this new lens!

Hi Asher,

This is what 3 rotation stitched tiles, portrait orientation, 10mm vertical shift, results in (the tiles were each made of 5 bracketed exposures and Photomatix Exposure Fusion):

Copyright © 2009, Bart van der Wolf

The full width is something like 135 degrees FOV on this rectilinear stitch, which is too much for most subjects. It also means that the L/R edges are stretched beyond reasonable resolution at full size. The Red rectangle marks approx. 110 degrees horizontal FOV and 70 degrees vertical FOV.

Autopano Giga gets thoroughly confused by the shifted images, but PTAssembler can manage them just fine, as would Hugin I assume.

Bart

 

[Doug Kerr]

Hi, Bart,

Hi Asher,

This is what 3 rotation stitched tiles, portrait orientation, 10mm vertical shift, results in (the tiles were each made of 5 bracketed exposures and Photomatix Exposure Fusion):

Lovely result. Great use of the technique/technology. You are indeed a master.

Best regards,

Doug

 

[Bart_van_der_Wolf]

Hi, Bart,


Lovely result. Great use of the technique/technology. You are indeed a master.

Hi Doug,

Thanks for the kind words. A nice subject of course also helps ;-)

Cheers,
Bart

 

[Asher Kelman]

Hi Asher,

This is what 3 rotation stitched tiles, portrait orientation, 10mm vertical shift, results in (the tiles were each made of 5 bracketed exposures and Photomatix Exposure Fusion):

Copyright © 2009, Bart van der Wolf

The full width is something like 135 degrees FOV on this rectilinear stitch, which is too much for most subjects. It also means that the L/R edges are stretched beyond reasonable resolution at full size. The Red rectangle marks approx. 110 degrees horizontal FOV and 70 degrees vertical FOV.

Autopano Giga gets thoroughly confused by the shifted images, but PTAssembler can manage them just fine, as would Hugin I assume.

Bart

Beautiful result. But tell me, Bart, how do you cross inform each batch of images in Photomatix that they are going to end up together. IOW, what if the range of light was different on one side. Where is this adjusted?

Asher

 

[Bart_van_der_Wolf]

Beautiful result. But tell me, Bart, how do you cross inform each batch of images in Photomatix that they are going to end up together. IOW, what if the range of light was different on one side. Where is this adjusted?

Hi Asher,

In this particular case I did the following.
I selected the 5 exposures for the first tile for input to the Exposure Fusion algorithm of Photomatix. I selected the Raw files, to try how the result would be from a PM conversion, so it was all very quick. I did the same for the 2nd tile and 3rd tile. Each exposure fusion result was saved as a TIFF. I then fired up PTAssembler, set the parameters for projection, and the initial estimated yaw (it helps the optimization engine with a better starting point).

Since I now only had 3 images to stitch, the controlpoints were generated automatically, with subsequent addition of a few verticals and horizontals. A number of iterrations of control point tuning later, the Smartblended result was postprocessed in Photoshop (mainly an S-curve).

The light differences in a natural light environment were partly handled by the camera (I usually shoot in Aperture priority mode), partly by Photomatix's blending algorithm, and partly by Smartblend (an overlap of 50% between tiles allows to blend the tile-to-tile differences).

Cheers,
Bart

 

[Michael Fontana]

.......
Autopano Giga gets thoroughly confused by the shifted images, but PTAssembler can manage them just fine, as would Hugin I assume.Bart

Interesting, how comes?

As this is a single row pano only, my guess was that the NNP was constant, even with the shiftet lens. There' arent any vertical tilts involved. Or does the NPP moves horizontally, when a lens is shiftet? I' ve a problem to believe that.

The church looks good... you aren't fusing anymore?

 

[Bart_van_der_Wolf]

Interesting, how comes?

As this is a single row pano only, my guess was that the NNP was constant, even with the shiftet lens. There' arent any vertical tilts involved. Or does the NPP moves horizontally, when a lens is shiftet? I' ve a problem to believe that.

I'm not exactly sure (who knows what 'Auto' does?), but in PTAssembler I have to include the vertical shift (e) parameter in the CP optimization, which makes sense for shifted input images. The amount of (e) shift is automatically found, and corresponds to the percentage of lens shift versus the sensor dimensions. I suspect that APG doesn't understand shifted input images, or doesn't try to optimize enough for decentered lenses.

The church looks good... you aren't fusing anymore?

I use different methods, and sometimes I use them all to see which one does best. In this particular set I made a mistake when setting up my gear, so I chose a quick conversion+fusion approach (I shot a different, larger FOV, regular unshifted pano with the correct settings after this one). I sometimes use an approach via an HDR+tonemapping for the individual tiles, but often like the TufusePro route best for natural looking results. Photomatix's Exposure Fusion is not bad at all, and offers a nice integrated workflow together with Raw conversion and alignment which works fast. For the best results I will do individual Raw conversions with Capture One, because that offers most control over the input images for the subsequent trials of different tonemapping methods.

Cheers,
Bart

 

[Michael Fontana]

I'm not exactly sure (who knows what 'Auto' does?), but in PTAssembler I have to include the vertical shift (e) parameter in the CP optimization, which makes sense for shifted input images. The amount of (e) shift is automatically found, and corresponds to the percentage of lens shift versus the sensor dimensions. I suspect that APG doesn't understand shifted input images, or doesn't try to optimize enough for decentered lenses.

okay, then it works with PTGui (& Hugin, you're right) as well, as it has the e param, too.

....... For the best results I will do individual Raw conversions with Capture One, because that offers most control over the input images for the subsequent trials of different tonemapping methods. Cheers,
Bart

That's my approach too; I found better results when stitching all the bracket shots as tiffs to multiple panos and mixing them later.