This topic is in the 2013 archive. You can read it and like its posts here. It is read only.
Archive 201324mm TS-E POV Question
p.3 #124mm TS-E POV Question
BTW, does anyone of you here have experience with long exposures (usually 30s-8min) in bright sunlight with the 17 TS-E (or 24 TS-E)? I'm interesting in buying a ND-filter system (preferably Haida with 150mm filter sizes), but hesitate because I fear that light reaches the sensor through the shift mechanism. Is this a real problem or do you have a good workaround for keeping unwanted light out? I would really love to use the 17 TS-E for long exposures, because I carry the 16-35 f/4 lens for this purpose only with me, which is a good lens, but in most cases not wide enough for me and less weight to carry around wouldn't be a bad thing...
Sep 15, 2015 at 04:06 PM
p.3 #224mm TS-E POV Question
Not wide enough at 16mm? You going to do many long exposures of 8min a frame with shift stitching? I see that as full of problems.
I do long exposures on the 24 never noticed a problem but I am more nirmally using my 16-35 for that too
I do long exposures on the 24 never noticed a problem but I am more nirmally using my 16-35 for that too
Sep 15, 2015 at 06:26 PM
p.3 #324mm TS-E POV Question
Thanks Rob for sharing your encouraging experience with the 24!
For what I primarily do, 16mm is actually not wide enough. I use the 17 TS-E mainly either for stitching (the problem with 3 different long exposures in the clouds for example can be come by in post processing):
http://www.paintingwithlight.de/FilesEx/uw_13.jpg
Or I shift fully upwards to get the verticals straight without stitching:
http://www.paintingwithlight.de/FilesEx/uw_29.jpg
A 16mm wouldn't be able to capture such images as easily as the TS-E.
For what I primarily do, 16mm is actually not wide enough. I use the 17 TS-E mainly either for stitching (the problem with 3 different long exposures in the clouds for example can be come by in post processing):
http://www.paintingwithlight.de/FilesEx/uw_13.jpg
Or I shift fully upwards to get the verticals straight without stitching:
http://www.paintingwithlight.de/FilesEx/uw_29.jpg
A 16mm wouldn't be able to capture such images as easily as the TS-E.
Sep 15, 2015 at 06:47 PM
p.3 #424mm TS-E POV Question
Peter Figen wrote:
And also it should be noted that once you tilt the 17 t/s-e even as little as a degree and a half or so, the corners go really soft. This is particularly noticeable in a landscape (landscape as in Landscape) where you tilt forward for depth of field near to far. The far corners (upper) are soft while the center of the upper frame is sharp. First saw this on a 1Ds3 where it was very pronounced. Haven't tested on 5DS because I'm fairly certain it's only going to be worse. The 24 t/s-e, on the other hand appears...Show more →
Yes, the corners go soft just like when using extreme shifting but closing the aperture to f/16 seems to remediate it somewhat as I pointed it out.
Something else to keep in mind. When shifting the TS-E 17mm in portrait orientation with 3 shots (-12,0,12), the resulting image is equivalent to 12mm FL with "double" the resolution of the full frame sensor.
In the case of the 5DS, the resulting image would be equivalent to 100MP. To increase apparent sharpness, aside from shooting at f/16 we should also down-sample the image back to 50MP.
And also it should be noted that once you tilt the 17 t/s-e even as little as a degree and a half or so, the corners go really soft. This is particularly noticeable in a landscape (landscape as in Landscape) where you tilt forward for depth of field near to far. The far corners (upper) are soft while the center of the upper frame is sharp. First saw this on a 1Ds3 where it was very pronounced. Haven't tested on 5DS because I'm fairly certain it's only going to be worse. The 24 t/s-e, on the other hand appears...Show more →
Yes, the corners go soft just like when using extreme shifting but closing the aperture to f/16 seems to remediate it somewhat as I pointed it out.
Something else to keep in mind. When shifting the TS-E 17mm in portrait orientation with 3 shots (-12,0,12), the resulting image is equivalent to 12mm FL with "double" the resolution of the full frame sensor.
In the case of the 5DS, the resulting image would be equivalent to 100MP. To increase apparent sharpness, aside from shooting at f/16 we should also down-sample the image back to 50MP.
Sep 15, 2015 at 08:10 PM
p.3 #524mm TS-E POV Question
Why would down-sampling optimize the image quality? Because the edges don't resolve that much and the sharpness would be more uniform across the image? Or because of the diffraction you get with f/16? I tested it only with f/11 and at least the center looses resolution by down-sampling.
Sep 15, 2015 at 08:32 PM
p.3 #624mm TS-E POV Question
"Yes, the corners go soft just like when using extreme shifting..."
I wouldn't say it's at all like the corner softness when shifting. When you're shifting a full 12mm, you're out very near the edge of the image circle, but when you're not shifting at all and have a very minor tilt or swing, the corners completely fall apart, while the top center stays sharp. You expect anything at the edge of the image circle to suffer somewhat, but not necessarily a minor tilt. Coming from an array of Super Angulons, that's not something I'm used to seeing, or at least, there was a certain level of surprise when I saw those results. If you read the reviews at B&H, there is at least one person who sent their lens back because of this, but since most people use it primarily shifted, including myself, it's not quite a non issue, but something to be aware of at all times. That the 24 does not exhibit this softness with tilting is both interesting to note and good to know.
I've shot several recording studios with the three shot stitch with both the 24 and the 17, but so far all on 5D3's. The results are quite very good and even large prints - 40x50 or more, are phenomenal.
I do not, however understand your stating that for optimal image quality, a stitched 5DS file should then be down rezzed in order for best quality. Half the time I'm stitching is for focal length, but it's also for effective file size as well, and if that's what I need, I would prefer to have lenses up to the task rather than compromise the entire image. Maybe we're due for a couple of Mark II tilt shift lenses, which will no doubt become painfully necessary when Canon raises the bar once again somewhere down the road.
I wouldn't say it's at all like the corner softness when shifting. When you're shifting a full 12mm, you're out very near the edge of the image circle, but when you're not shifting at all and have a very minor tilt or swing, the corners completely fall apart, while the top center stays sharp. You expect anything at the edge of the image circle to suffer somewhat, but not necessarily a minor tilt. Coming from an array of Super Angulons, that's not something I'm used to seeing, or at least, there was a certain level of surprise when I saw those results. If you read the reviews at B&H, there is at least one person who sent their lens back because of this, but since most people use it primarily shifted, including myself, it's not quite a non issue, but something to be aware of at all times. That the 24 does not exhibit this softness with tilting is both interesting to note and good to know.
I've shot several recording studios with the three shot stitch with both the 24 and the 17, but so far all on 5D3's. The results are quite very good and even large prints - 40x50 or more, are phenomenal.
I do not, however understand your stating that for optimal image quality, a stitched 5DS file should then be down rezzed in order for best quality. Half the time I'm stitching is for focal length, but it's also for effective file size as well, and if that's what I need, I would prefer to have lenses up to the task rather than compromise the entire image. Maybe we're due for a couple of Mark II tilt shift lenses, which will no doubt become painfully necessary when Canon raises the bar once again somewhere down the road.
Sep 15, 2015 at 09:22 PM
p.3 #724mm TS-E POV Question
When we tilt, we are jockeying around the focal plane. That's not a problem with the optics, that's just a product of the physics / trigonometry involved @ the distance to the corners at there is no such thing as a free lunch ... it is a strategic compromise to push/pull things ... but it does come with a cost.
Can you offset that cost at the expense of another cost (i.e. f/16 @ diffraction) ... sure, but the use of tilt will always be an exercise in a strategic tradeoffs regarding the focal plane. Sometimes we can apply it in such a way to render it an unnoticeable cost. Other times it isn't as readily hid. Other times we compensate or compromise elsewhere.
Can you offset that cost at the expense of another cost (i.e. f/16 @ diffraction) ... sure, but the use of tilt will always be an exercise in a strategic tradeoffs regarding the focal plane. Sometimes we can apply it in such a way to render it an unnoticeable cost. Other times it isn't as readily hid. Other times we compensate or compromise elsewhere.
Sep 16, 2015 at 08:43 AM
p.3 #824mm TS-E POV Question
Ulff wrote:
Why would down-sampling optimize the image quality? Because the edges don't resolve that much and the sharpness would be more uniform across the image? Or because of the diffraction you get with f/16? I tested it only with f/11 and at least the center looses resolution by down-sampling.
Let's just consider that the TS-E 17mm is also a 12mm lens when using the entire image circle and similar aspect ratio (not panorama)
We can effectively have the field of view of a 12mm lens when shifting horizontally and capturing 3 shots in portrait orientation. After the stitching, we almost double image size as well. With the Sony A7RII (42MP) the resulting stitched image is 76MP.
The problem as discussed on this thread is that extreme edges will suffer noticeably and became almost unusable unless closing the aperture to at least f/11 and optimally f/16 reducing micro contrast due to diffraction.
However when disregarding the extra resolution benefit from stitching and downsampling the image back to 42MP I see a visible improvement in apparent sharpness, especially on the edges. Downsampling that image to a lower resolution will obviously not add any single detail that wasn't there in the 76MP original. It's more like, it increases the “apparent” sharpness (per pixel) and/or reduces the apparent noise (per pixel). It's like what you wrote, the edges won't resolve as much and the sharpness evens out across the image. I did some tests with the 11-24L and the TS-E 17mm stitched to 12mm field of view downsampled back to 42MP and they are very similar at the extreme edges.
Regarding tilting for landscape photography. I went back to my images and that is not an issue at all since the softness Peter wrote about would show up on the sky area. That is perhaps why I never noticed it. I have f/8 tilted shots with great foreground detail even towards the extreme edges.
Fred Miranda wrote
To optimize image quality, aside from shooting at f/16 we should also down-sample the image back to 50MP.Let's just consider that the TS-E 17mm is also a 12mm lens when using the entire image circle and similar aspect ratio (not panorama)
We can effectively have the field of view of a 12mm lens when shifting horizontally and capturing 3 shots in portrait orientation. After the stitching, we almost double image size as well. With the Sony A7RII (42MP) the resulting stitched image is 76MP.
The problem as discussed on this thread is that extreme edges will suffer noticeably and became almost unusable unless closing the aperture to at least f/11 and optimally f/16 reducing micro contrast due to diffraction.
However when disregarding the extra resolution benefit from stitching and downsampling the image back to 42MP I see a visible improvement in apparent sharpness, especially on the edges. Downsampling that image to a lower resolution will obviously not add any single detail that wasn't there in the 76MP original. It's more like, it increases the “apparent” sharpness (per pixel) and/or reduces the apparent noise (per pixel). It's like what you wrote, the edges won't resolve as much and the sharpness evens out across the image. I did some tests with the 11-24L and the TS-E 17mm stitched to 12mm field of view downsampled back to 42MP and they are very similar at the extreme edges.
Regarding tilting for landscape photography. I went back to my images and that is not an issue at all since the softness Peter wrote about would show up on the sky area. That is perhaps why I never noticed it. I have f/8 tilted shots with great foreground detail even towards the extreme edges.
Sep 16, 2015 at 01:14 PM
p.3 #924mm TS-E POV Question
Yes, often in landscape you end up with sky where the soft edges occur, but I've seen that even further down the from the corners on the frame edge where significant detail wasn't there anymore. And it's doubly important in product and architectural photography even though you seldom use tilts or swings for the architectural images. This is all really about knowing your lenses and how they perform in any given situation so you don't come home to a nasty surprise.
Sep 16, 2015 at 02:37 PM
p.3 #1024mm TS-E POV Question
I use the calculator on this page to get equivalent focal lengths for shifted images:
http://www.cambridgeincolour.com/tutorials/tilt-shift-lenses1.htm
http://www.cambridgeincolour.com/tutorials/tilt-shift-lenses1.htm
Sep 16, 2015 at 02:49 PM
p.3 #1124mm TS-E POV Question
johnahill wrote:
I use the calculator on this page to get equivalent focal lengths for shifted images:
http://www.cambridgeincolour.com/tutorials/tilt-shift-lenses1.htm
I used to use this calculator as well but learned a while back that it is not accurate at all.
The numbers for the TS-E 17mm and TS-E 24mm should be like this:
https://www.fredmiranda.com/forum/topic/1186421/1#13193755
Fred
I use the calculator on this page to get equivalent focal lengths for shifted images:
http://www.cambridgeincolour.com/tutorials/tilt-shift-lenses1.htm
I used to use this calculator as well but learned a while back that it is not accurate at all.
The numbers for the TS-E 17mm and TS-E 24mm should be like this:
https://www.fredmiranda.com/forum/topic/1186421/1#13193755
Fred
Sep 16, 2015 at 03:34 PM
p.3 #1224mm TS-E POV Question
RobDickinson wrote:
Shifting that setup to the maximum left and right (this is in reality 9 shots bracketed from -12mm,0mm,+12mm and -2,0,+2), this image is 9400x3700. Equivalent to about 14mm wide but tons more resolution, downside is the inability to shoot it in one frame so moving subjects are mostly out. Because I didnt shift the camera as well I have parallax issues with some stuff near to the camera on the right, technically I think I need to shift the camera in the opposite direction to the lens shift (keeping the lens in effectively the same place for each frame)
I'll certainly...Show more →
Surely it isn't possible to shift to correct for parallax and simultaneously shift stitch?
Bill
Shifting that setup to the maximum left and right (this is in reality 9 shots bracketed from -12mm,0mm,+12mm and -2,0,+2), this image is 9400x3700. Equivalent to about 14mm wide but tons more resolution, downside is the inability to shoot it in one frame so moving subjects are mostly out. Because I didnt shift the camera as well I have parallax issues with some stuff near to the camera on the right, technically I think I need to shift the camera in the opposite direction to the lens shift (keeping the lens in effectively the same place for each frame)
I'll certainly...Show more →
Surely it isn't possible to shift to correct for parallax and simultaneously shift stitch?
Bill
Jan 09, 2016 at 04:19 PM
p.3 #1324mm TS-E POV Question
Yes. For example, if you shift the lens 10mm to the left and then shift the camera 10mm to the right (say the camera is on a sideways-siding rail), the sensor still sees the 10mm left shift but there's no parallax because the lens is in exactly the same place as before the left-shift was made. This is easy if the shift lens is on its own tripod mount (because you shift the camera, not the lens) but you can get the same result by shifting the lens one direction and then moving the camera the same amount in the other direction.
Bunter wrote:
I'll certainly have to learn the right trick of shifting the camera at the same time to avoid parallax issues. The low distortion makes stitching with 24mm OK, I still think normally your better off with a longer focal length (50mm on FF seems ideal)."Surely it isn't possible to shift to correct for parallax and simultaneously shift stitch?
Bill
Bunter wrote:
RobDickinson wrote
Shifting that setup to the maximum left and right (this is in reality 9 shots bracketed from -12mm,0mm,+12mm and -2,0,+2), this image is 9400x3700. Equivalent to about 14mm wide but tons more resolution, downside is the inability to shoot it in one frame so moving subjects are mostly out. Because I didnt shift the camera as well I have parallax issues with some stuff near to the camera on the right, technically I think I need to shift the camera in the opposite direction to the lens shift (keeping the lens in effectively the same place for each frame)I'll certainly have to learn the right trick of shifting the camera at the same time to avoid parallax issues. The low distortion makes stitching with 24mm OK, I still think normally your better off with a longer focal length (50mm on FF seems ideal)."
Bill
Jan 09, 2016 at 05:18 PM
p.3 #1424mm TS-E POV Question
Bunter wrote:
Surely it isn't possible to shift to correct for parallax and simultaneously shift stitch?
Bill
As said yes its easy, I use an L plate with marked 10mm spots either way now and when I need to I shift the lens and camera
Surely it isn't possible to shift to correct for parallax and simultaneously shift stitch?
Bill
As said yes its easy, I use an L plate with marked 10mm spots either way now and when I need to I shift the lens and camera
Jan 09, 2016 at 06:42 PM
p.3 #1524mm TS-E POV Question
Bunter wrote:
Surely it isn't possible to shift to correct for parallax and simultaneously shift stitch?
Bill
As said yes its easy, I use an L plate with marked 10mm spots either way now and when I need to I shift the lens and camera
Surely it isn't possible to shift to correct for parallax and simultaneously shift stitch?
Bill
As said yes its easy, I use an L plate with marked 10mm spots either way now and when I need to I shift the lens and camera
Jan 09, 2016 at 06:42 PM
p.3 #1624mm TS-E POV Question
Thanks both. So that I make sure I understand correctly, you're both essentially describing having 2 mechanism to produce shift (one on the TS lens and one done using an L-bracket or tripod sliding rail). And you can use either to do either job (parallax or stitch shifting).
But what you can't do is shift using the lens shift for parallax and use the lens shift for stitching? Have I got that correct.
Thanks again
Bill
But what you can't do is shift using the lens shift for parallax and use the lens shift for stitching? Have I got that correct.
Thanks again
Bill
Jan 10, 2016 at 11:13 AM
p.3 #1724mm TS-E POV Question
The key point is that parallax is introduced when you move the lens relative to the subject (e.g. the ground). If the shift lens itself is mounted on the tripod (like a Schneider-K TS lens), then shifting results in moving the camera; the lens does not move; and so there is no parallax. If the camera is attached to the tripod, then shifting the lens moves the lens with respect to the ground. Moving the camera in the opposite direction compensates for (i.e. removes) this lens movement with respect to the ground, and so there is no parallax. Moving the lens with respect to the camera body is how you make the shift-pano. Not moving the lens with respect to the ground is how you avoid parallax errors.
Jan 10, 2016 at 11:28 AM
p.3 #1824mm TS-E POV Question
Thanks. I think I've confused avoiding parallax errors with avoiding converging verticals, and they aren't the same thing at all! Sorry all!
Bill
Bill
Jan 10, 2016 at 04:19 PM
p.3 #1924mm TS-E POV Question
Peter Figen wrote:
And also it should be noted that once you tilt the 17 t/s-e even as little as a degree and a half or so, the corners go really soft.
Peter,
I only have the old 24 TS-E with it's smaller image circle. When using tilt I have found it pays to also use a little shift to "re-centre" the image circle, as I occasionally did when using wide angle lenses on sheet film. Obviously, a bit of composition juggling is required and, in the case of this lens, re-orientating the shift axis.
This may work the 17 TS-E?
Regards,
CraigZ
And also it should be noted that once you tilt the 17 t/s-e even as little as a degree and a half or so, the corners go really soft.
Peter,
I only have the old 24 TS-E with it's smaller image circle. When using tilt I have found it pays to also use a little shift to "re-centre" the image circle, as I occasionally did when using wide angle lenses on sheet film. Obviously, a bit of composition juggling is required and, in the case of this lens, re-orientating the shift axis.
This may work the 17 TS-E?
Regards,
CraigZ
Jan 13, 2016 at 10:14 AM
p.3 #2024mm TS-E POV Question
What is the optimum aperture for the 24mm TS-E when doing the same?
Jan 13, 2016 at 01:24 PM