Ad
Ad
Ad
Pages: « 1 [2] 3 »   Bottom of Page
Print
Author Topic: My primes not as sharp as zooms, huh?!?  (Read 11213 times)
Paul Kay
Full Member
***
Offline Offline

Posts: 131


WWW
« Reply #20 on: January 21, 2007, 07:30:50 AM »
ReplyReply

"It is well known that Canon wide angle primes are not very good."

Sorry to disagree but I think that this ought to say that is is "an internet promoted assertion that Canon wide-angle primes are not very good".

Some certainly are - I have a couple. Its true enough to say that some of their lenses are probably not that good - the 20/2.8 certainly has soft corners but others, such as my 24/1.4, are very usable indeed. My suspicion is that it is all too easy to find fault and broadcast it, and all too easy to forget that lenses are tools and are as good as they job they are being used for demands. In my experience it is rare for an image to be unacceptable due to the lens being used - far more are unacceptable due to the photographer making unwise choices which may or may not include the choice of lens. Having sold many thousands of images I have NEVER lost an image sale due to the image quality provided by the lens I was using - and this includes Canon wides.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8875


« Reply #21 on: January 21, 2007, 09:37:03 PM »
ReplyReply

Quote
Sorry to disagree but I think that this ought to say that is is "an internet promoted assertion that Canon wide-angle primes are not very good".

Paul,
There's no smoke without fire. I think I've demonstrated that the 24TS with camera vertical can produce remarkably detailed images using shift for stitching purposes, way beyond what could be achieved with a single shot.

However, now that I've seen the results at the extreme edges, with camera horizontal (on the 5D), there's no way I'm going to bother using this lens in this way. If I need a stitched image with camera horizontal, I'll use my 20D with the 24TS.

I notice that Kirk Gittings has been very quiet on the issue of which camera he's using with his TS-E 24mm.

What I'd like to see is a comparison between a P25 with Digitar 24mm ultra wide, and 2 stitched 24TS images with the 5D, camera vertical. The image size and aspect ratio should be almost identical in both cases, but I suspect the Digitar is a much sharper lens than the 24TS, and of course it's a much more expensive lens.

Another issue that concerns me is the benefits of a centre filter. I notice that Schneider Optics offer a centre filter for their 24mm, that provides 2 stops variation between centre and edges. It seems to me that this is what's required for the Canon 24TS.

However, I don't think one can assume that raising the exposure at the edges will solve the resolution fall-off. It seems to me that some lenses have a vignetting problem at the edges as well as resolution fall-off. Other lenses, like the Digitar 24mm, might suffer from peripheral light fall-off but not peripheral resolution fall-off.

Quote
In my experience it is rare for an image to be unacceptable due to the lens being used - far more are unacceptable due to the photographer making unwise choices which may or may not include the choice of lens. Having sold many thousands of images I have NEVER lost an image sale due to the image quality provided by the lens I was using - and this includes Canon wides.

Well, of course, this gets us back to the very old issue, that's been debated many times on this forum, of whether it's the equipment or the photographer that 'makes' the picture.

The answer is quite clear. It's both. I've never yet come across an instance of a photographer who has been able to take a photo without using any equipment (good or bad), but I have come across instances of photos being taken by equipment without a photographer. The Mars Rover is an example   .
« Last Edit: January 21, 2007, 09:38:49 PM by Ray » Logged
Ronny Nilsen
Sr. Member
****
Offline Offline

Posts: 340


WWW
« Reply #22 on: January 22, 2007, 02:41:56 AM »
ReplyReply

Quote
However, now that I've seen the results at the extreme edges, with camera horizontal (on the 5D), there's no way I'm going to bother using this lens in this way. If I need a stitched image with camera horizontal, I'll use my 20D with the 24TS.
[a href=\"index.php?act=findpost&pid=96916\"][{POST_SNAPBACK}][/a]

That's really not that suprising is it? On a 5D with full horisontal shift you are using an image circle dia. of 62.55 mm. (you get an 23.9 x 57.8 "sensor") The spec for the lens says that it have an image circle dia. 58.6mm. With the 20D you are only using an 46.95mm image circle dia. Doing a vertical shift with a 5D uses an images circel of 58.21 mm and should be ok acoording to the spec.

What is suprising is that the spec for the 90TS and 45TS states the same image circle (58.6mm) but my  lenses are sharp to the corner on my 5D letting me use an image circle dia. of 62.55 mm. So a 45TS on 5D migth be a better option than a 24TS on a 20D? Or just limit horisontal shit to +/- 9mm?
Logged

Ray
Sr. Member
****
Offline Offline

Posts: 8875


« Reply #23 on: January 22, 2007, 05:23:36 AM »
ReplyReply

Quote
That's really not that suprising is it? On a 5D with full horisontal shift you are using an image circle dia. of 62.55 mm. (you get an 23.9 x 57.8 "sensor") The spec for the lens says that it have an image circle dia. 58.6mm. With the 20D you are only using an 46.95mm image circle dia. Doing a vertical shift with a 5D uses an images circel of 58.21 mm and should be ok acoording to the spec.

What is suprising is that the spec for the 90TS and 45TS states the same image circle (58.6mm) but my  lenses are sharp to the corner on my 5D letting me use an image circle dia. of 62.55 mm. So a 45TS on 5D migth be a better option than a 24TS on a 20D? Or just limit horisontal shit to +/- 9mm?
[a href=\"index.php?act=findpost&pid=96948\"][{POST_SNAPBACK}][/a]

Well, yes, it is surprising. The primary use of the shift is for perspective control. I could be photographing a skyscraper or cathedral with camera in vertical position. If I were to use the full 11mm movement, the top of the cathedral would be unacceptably dark and degraded.

I've thought of getting the 45TS, but I ask myself if I would use it enough to justify the expense. These are also heavy lenses. On my recent trip to Nepal and Cambodia, I left behind my TS-E 90 because of weight considerations. I decided the wider lens would be more useful. Would I want to carry a 45TS as well as the much lighter 50/1.4?
Logged
Ronny Nilsen
Sr. Member
****
Offline Offline

Posts: 340


WWW
« Reply #24 on: January 22, 2007, 06:51:33 AM »
ReplyReply

Quote
Well, yes, it is surprising. The primary use of the shift is for perspective control. I could be photographing a skyscraper or cathedral with camera in vertical position. If I were to use the full 11mm movement, the top of the cathedral would be unacceptably dark and degraded.
[a href=\"index.php?act=findpost&pid=96958\"][{POST_SNAPBACK}][/a]

It would of course have been better if the lens had an image circle of 63mm, but it does not, it only have an image circle large enough to have the fuill +/- 11mm of vertical shift, and only +/- 9mm horizontal shift. That's in the spec of the lens and should not be a suprise.  

Just as you dont twist the focus ring to infinity without checking if that is the right focusing distance, you should also check the shift when doing horizontal shift, but you know that.  

Quote
I've thought of getting the 45TS, but I ask myself if I would use it enough to justify the expense. These are also heavy lenses. On my recent trip to Nepal and Cambodia, I left behind my TS-E 90 because of weight considerations. I decided the wider lens would be more useful. Would I want to carry a 45TS as well as the much lighter 50/1.4?
[a href=\"index.php?act=findpost&pid=96958\"][{POST_SNAPBACK}][/a]

I guess that depends on personal preferences and what you are shooting, but I would prefer the 45 over th 50.
Logged

Ray
Sr. Member
****
Offline Offline

Posts: 8875


« Reply #25 on: January 22, 2007, 11:13:12 AM »
ReplyReply

Quote
It would of course have been better if the lens had an image circle of 63mm, but it does not...[a href=\"index.php?act=findpost&pid=96964\"][{POST_SNAPBACK}][/a]

This is simply not true, Ronnynil. If the image circle was not 63mm, I could not possibly get any image at all in the corners   .

Of course, I understand that Canon might be saying indirectly, with their spec of 58.6mm, that this is in their opinion the maximum image circle for acceptable quality. In other words, 'don't use the full shift with our tilt and shift lens'.

I don't see any such warning on the packaging, however.
Logged
Ronny Nilsen
Sr. Member
****
Offline Offline

Posts: 340


WWW
« Reply #26 on: January 22, 2007, 01:40:18 PM »
ReplyReply

Quote
This is simply not true, Ronnynil. If the image circle was not 63mm, I could not possibly get any image at all in the corners   .
[a href=\"index.php?act=findpost&pid=97001\"][{POST_SNAPBACK}][/a]

I guess you are talking about the circle of illumination. Not the the image circle of a lens that are the circle of sharp definition WITHIN the circle of illumination? But english is not my native language so I may miss the finer points of your language here.  

Quote
Of course, I understand that Canon might be saying indirectly, with their spec of 58.6mm, that this is in their opinion the maximum image circle for acceptable quality. In other words, 'don't use the full shift with our tilt and shift lens'.

I don't see any such warning on the packaging, however.
[a href=\"index.php?act=findpost&pid=97001\"][{POST_SNAPBACK}][/a]

My manual for the TS lenses page 48 warns about this (Page title Precautions!):

Quote
2) When combining tilt and shift function with the TS-E 24mm f/3.5L lens, setting amount s indicated in red may cause shading.
3) When using the TS-E 24mm f/3.5L, shifting more than 8mm may cause shading on the long edges or in corners of the picture. Be especially carful when using a camera other than the EOS-1, EOS-1N.

So Canon states clearly in both the spec and the manual that you are using the lens outside the image area that they consider acceptable. They have even marked the shifting on the lens that can cause shading in red, not the gray used for the rest of the shifting range.

So I say again; your results should not be a suprise.  
« Last Edit: January 22, 2007, 01:43:19 PM by ronnynil » Logged

RedRebel
Guest
« Reply #27 on: January 22, 2007, 03:08:19 PM »
ReplyReply

I also have a 35mm/f2 and bought it while I had a 350D. Then I already noticed that it was about as sharp as my 17-85-IS when both used wide open and at f2.0 it's horrible. Then the 35mm/f2 also suffered from focussing issues, the lens was designed for film camera's and not for 1.6 crop bodies. The shop where I purchased it (Foto Konijnenberg) confirmed this.

When I purchased my 5D I noticed that the focussing issues with the 35/f2 were solved, but its not as sharp as my 24-105L.

Someone mentioned that "old design" primes are not as sharp/good as recent zoom lenses, and I think that's true. I think that you realy have to be carefull when selecting a very fast prime lens if you have very high demands in terms of lens speed and sharpnes. The 35/f2 also suffers from lots of CA and flare (contrast drop). I never use my copy of the 35/f2, it's a waste of money. The new 50mm/f1.2L is a great lens I think, but it's price is not justified for me.
« Last Edit: January 22, 2007, 03:09:06 PM by RedRebel » Logged
Kirk Gittings
Sr. Member
****
Offline Offline

Posts: 1547


WWW
« Reply #28 on: January 22, 2007, 03:16:32 PM »
ReplyReply

Quote
I notice that Kirk Gittings has been very quiet on the issue of which camera he's using with his TS-E 24mm.

Sorry I missed this. I am buried with shoots.

A 5D (which will likely become my backup camera when I see what is available in the spring).
« Last Edit: January 22, 2007, 03:35:06 PM by Kirk Gittings » Logged

Thanks,
Kirk

Kirk Gittings
Architecture and Landscape Photography
WWW.GITTINGSPHOTO.COM

LIGHT+SPACE+STRUCTURE (blog)
Ray
Sr. Member
****
Offline Offline

Posts: 8875


« Reply #29 on: January 22, 2007, 08:22:50 PM »
ReplyReply

Quote
I guess you are talking about the circle of illumination. Not the the image circle of a lens that are the circle of sharp definition WITHIN the circle of illumination? But english is not my native language so I may miss the finer points of your language here. 

Your English is fine. Most lenses suffer from both vignetting and resolution fall-off towards the edges even though it might not always be particularly noticeable. It seems that image quality just gets progressively worse away from the centre. There's no cut-off point like, 'now the image is good. Now it's not good'. The Photodo MTF charts tell the story. Wide angle lenses are generally much worse in this respect, but sometimes telephoto mirror lenses are particularly bad. The Sigma 14mm prime is particularly poor towards the edges.

This gradual deterioration of lens performance towards the corners is what makes cropped format APS-C cameras so appealing. You don't have to worry about such factors whatever the lens.

Quote
So Canon states clearly in both the spec and the manual that you are using the lens outside the image area that they consider acceptable. They have even marked the shifting on the lens that can cause shading in red, not the gray used for the rest of the shifting range.

Okay! You got me there   . I should have paid more attention to the manual and that little red line on the lens. I guess Canon are covering their backsides. Anyway, I bought the lens when I was using the D60 and 20D. There would seem to be little advantage (for my purposes) using it with the 5D if I have to crop the image to something approaching the FoV of the 20D.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8875


« Reply #30 on: January 22, 2007, 08:29:50 PM »
ReplyReply

Quote
Sorry I missed this. I am buried with shoots.

A 5D (which will likely become my backup camera when I see what is available in the spring).
[a href=\"index.php?act=findpost&pid=97051\"][{POST_SNAPBACK}][/a]

Kirk,
Sorry to labour the point. Are you paying attention to those red warning lines on the 24TS, or is your lens so good at the edges you can safely ingnore them?
Logged
Kirk Gittings
Sr. Member
****
Offline Offline

Posts: 1547


WWW
« Reply #31 on: January 22, 2007, 09:43:19 PM »
ReplyReply

Yes Ray, If there is important detail in the corner(s) in the shift direction. Understand we are almost always shooting tethered and it is easy to check this in the field.
Logged

Thanks,
Kirk

Kirk Gittings
Architecture and Landscape Photography
WWW.GITTINGSPHOTO.COM

LIGHT+SPACE+STRUCTURE (blog)
Paul Kay
Full Member
***
Offline Offline

Posts: 131


WWW
« Reply #32 on: January 23, 2007, 02:23:52 AM »
ReplyReply

"There's no smoke without fire. I think I've demonstrated that the 24TS with camera vertical can produce remarkably detailed images using shift for stitching purposes, way beyond what could be achieved with a single shot."

Absolutely! A good discussion which demonstrates the value of a forum like this. What I was trying to get at is that a sweeping statement regarding Canon prime wide-angles being poor is simply a potential internet myth in the making and a gross oversimplification of the technical issues involved. Take my 24/1.4 as an example - it works pretty well for me and I would be interested in knowing what other lenses exist to allow me to shoot wide open with this field of view on full frame? My ponit being that there is nothing to really compare it with and it delivers perfectly acceptable results.

Many people read information on the internet and if a statement is made often enough they may believe it whether it is truly accurate or otherwise. I think that it is important to qualify such statements and use a discussion forum as in this thread, to actually discuss issues and hopefully to learn accurately from such discussions.

As a specialist in underwater photography, I can say that Nikon still predominates probably due to the legacy of the Nikonos 5 and the perceived need for ttl flash control which remains an ingrained and perpetuated idea even today - although so much has changed that choices are far more wide-open now.

So sorry if I sounded off, but perhaps my point requires further discussion (elsewhere if not here).
Logged
Ronny Nilsen
Sr. Member
****
Offline Offline

Posts: 340


WWW
« Reply #33 on: January 23, 2007, 05:46:38 AM »
ReplyReply

Quote
Anyway, I bought the lens when I was using the D60 and 20D. There would seem to be little advantage (for my purposes) using it with the 5D if I have to crop the image to something approaching the FoV of the 20D.
[a href=\"index.php?act=findpost&pid=97094\"][{POST_SNAPBACK}][/a]

The advantage of the TS-24 on a 5D is that you can get almost the same resolution (12.8mpx) with one shot as you get with a stitched 20D image (16.3mpx) from the same lens. And the 5D will then also only use the best part of the image circle (more than the 20D).

But the  20D will give a 1:3 panorama if thats what you want.
Logged

Ray
Sr. Member
****
Offline Offline

Posts: 8875


« Reply #34 on: January 23, 2007, 08:35:18 AM »
ReplyReply

Quote
The advantage of the TS-24 on a 5D is that you can get almost the same resolution (12.8mpx) with one shot as you get with a stitched 20D image (16.3mpx) from the same lens. And the 5D will then also only use the best part of the image circle (more than the 20D).

But the  20D will give a 1:3 panorama if thats what you want.
[a href=\"index.php?act=findpost&pid=97142\"][{POST_SNAPBACK}][/a]

With 20D mounted vertically, shifting fully from one side to the other (completley ignoring the red lines   ), taking 3 shots then stitching, you get roughly the equivalent of a single shot with the 5D horizontal; almost the same aspect ratio and same FoV, but a much higher resolution, about 19 megapixels.
Logged
jani
Sr. Member
****
Offline Offline

Posts: 1604



WWW
« Reply #35 on: January 25, 2007, 06:46:57 AM »
ReplyReply

Quote
With 20D mounted vertically, shifting fully from one side to the other (completley ignoring the red lines   ), taking 3 shots then stitching, you get roughly the equivalent of a single shot with the 5D horizontal; almost the same aspect ratio and same FoV, but a much higher resolution, about 19 megapixels.
I'm not sure that I'd sign any paper that said ca. 19 megapixels is a "much higher resolution" than ca. 13, it's only just above a 20% angular resolution increase.

Even with a 400D, you'd get just about a third more angular resolution.

If the 5D gave you a 18x12 print in usable quality, the 20D triple-stitched might then give you a 21.5x14.5, and the 400D a 24x16.

I can see how it might be worth the trouble if you need that extra resolution, but why not just do a regular double-stitch with the 5D, also ignoring the red lines? Am I missing something?
Logged

Jan
Ronny Nilsen
Sr. Member
****
Offline Offline

Posts: 340


WWW
« Reply #36 on: January 25, 2007, 07:14:13 AM »
ReplyReply

Quote
I can see how it might be worth the trouble if you need that extra resolution, but why not just do a regular double-stitch with the 5D, also ignoring the red lines? Am I missing something?
[a href=\"index.php?act=findpost&pid=97461\"][{POST_SNAPBACK}][/a]

Becuse ignoring the red lines on TS24 will push the larger 5D sensor into the bad parts of the image circle.

But only using +/- 8mm shift on the 5D will give you a 18.4 mpx 1:2.1 image or a 21.2 mpx 1:1.1 image, which still is better than the 20D 16.3 mpx and 20.3 images. On the TS45 and TS90 you can do the full shift on a 5D with very god results.
Logged

Ray
Sr. Member
****
Offline Offline

Posts: 8875


« Reply #37 on: January 25, 2007, 09:17:15 AM »
ReplyReply

Quote
But only using +/- 8mm shift on the 5D will give you a 18.4 mpx 1:2.1 image or a 21.2 mpx 1:1.1 image, which still is better than the 20D 16.3 mpx and 20.3 images.[a href=\"index.php?act=findpost&pid=97465\"][{POST_SNAPBACK}][/a]

It's not better in the corners though. This is a lens with a reasonably flat MTF response (for a wide angle lens) but only within the normal image circle that applies to 35mm, ie. up to 21mm from the centre. Any shift at all (with FF 35mm)pushes the lens into noticeable degradation in the corners. With the 20D, you can get a 20mp image (perhaps 19 after cropping) which is acceptably sharp and free of vignetting, from corner to corner.
Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #38 on: January 25, 2007, 10:12:58 AM »
ReplyReply

Quote
I'm not sure that I'd sign any paper that said ca. 19 megapixels is a "much higher resolution" than ca. 13, it's only just above a 20% angular resolution increase.

Even with a 400D, you'd get just about a third more angular resolution.

...

Am I missing something?

Yes, the fact that angular resolution squared is how you calculate overall resolution. Doubling angular resolution quadruples image detail. As long as per-pixel image quality is comparable between the 13MP and 19MP images, the 19MP will look significantly better.
Logged

jani
Sr. Member
****
Offline Offline

Posts: 1604



WWW
« Reply #39 on: January 25, 2007, 12:50:39 PM »
ReplyReply

Quote
Yes, the fact that angular resolution squared is how you calculate overall resolution.
No, I did not miss that.

But this kind of "overall resolution" is not a term used in any other context.

With printers, for instance, we don't speak of 90,000 DPI squared as the "overall resolution", and since the target is -- hopefully -- print, speaking of an "overall resolution" on a different kind of scale is useless.

We don't even speak of monitor resolution as 10,000 PPI squared.

Quote
Doubling angular resolution quadruples image detail. As long as per-pixel image quality is comparable between the 13MP and 19MP images, the 19MP will look significantly better.
But only up to a print that's 20% larger in any direction.

I'm not sure that's "significantly better".


But I did miss the fact that "redlining" the 5D was worse than "redlining" the 20D, so thanks to Ronny and Ray for making that clear.
Logged

Jan
Pages: « 1 [2] 3 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad