Panorama shooting equipment

[mark]

I've been wanting to do this for quite some time with my DSLR. I've been reading and watching a lot and I'm confused. Since I am wanting to shoot at the maximum 6x17ish perspective I am trying to figure out the piece(s) of kit I need to purchase.

I have a Nikon D7100 that I am plenty happy with and LOVE my Sigma art lenses. Don't want a new camera, so I will be using that. Here is my issue

There is a lot of discussion about rotating around the nodal point of the lens and not rotating around the base of the camera. This makes sense except for the fact that it is still a rotation and thus, in my mind will still create a curved image.

Then a stumble on only a few mentions, but no discussion of a slider to move the camera horizontally. This make a lot more sense to me, especially from my large format camera usage. No curved image. PC lenses are way out of my current wallet capacity. Since no PC lens it seems to me I would just need a slider to shift the camera over, and not something that moves the camera around the nodal point.

Am I wrong in that thinking?

Yes, cropping is an option, but one that cuts my pixel count down and I want all of my available pixels in the image.

Anyone out there able to give me some advice?

 

[Dan Fromm]

Sliding? Try shifting your camera a few inches sideways. I've tried it, it doesn't give the effect you want.

Rotation? Try rotation and stitching. All you need to try is a tripod with a pan head. Hint, you want aball leveler so that you can make the pan axis vertical. If the pan axis isn't vertical, panning will make the horizon rise (or fall). Don't ask, try.

I've got good results with film and my Nikon AP-2.

 

[grain elevator]

As Dan says, sliding won't help unless you can slide many meters. Your alternative to rotating is using a shift lens, that you can shift around to essentially use all of its image circle as if you were using a larger sensor camera. For a panorama that isn't very wide, you can have your stitching software calculate a rectilinear projection.
If you rotate for a wide panorama, you do get (with stitching software) a projection like an iphone panorama or a rotating lens panormic camera or a pinhole camera with curved film make - horizontal lines that aren't center image will curve, vertical lines will be straight, diagonals will become more or less s-shaped. On the upside, heads near the edges are still round-ish, rather than stretched out like with a rectilear wideangle.

 

[mark]

As Dan says, sliding won't help unless you can slide many meters. Your alternative to rotating is using a shift lens, that you can shift around to essentially use all of its image circle as if you were using a larger sensor camera. For a panorama that isn't very wide, you can have your stitching software calculate a rectilinear projection.
If you rotate for a wide panorama, you do get (with stitching software) a projection like an iphone panorama or a rotating lens panormic camera or a pinhole camera with curved film make - horizontal lines that aren't center image will curve, vertical lines will be straight, diagonals will become more or less s-shaped. On the upside, heads near the edges are still round-ish, rather than stretched out like with a rectilear wideangle.

Is there software that will straighten those curves?

 

[grat]

Large format shifting works because you're shifting the angle between the lens and the film plane. It's like a large, asymmetrical "X", where the nodal point is the middle of the X. Because the "bottom" of the X is small, and the top is large, shifting the optical path has a multiplier effect.

The point of rotating around the nodal point when doing panos is to avoid parallax errors. If you hold your finger in front of your left eye, and open/close each eye, the finger will seem to move. If you can place your finger on the line between your eyes, at the right distance, then when you alternate eyes, your finger should stay in place. Rotating around the nodal point reduces those apparent errors, although if you have objects in your foreground and the background, you can still have issues.

Yes, you will produce a curved image-- a "spherical projection" usually-- that looks a bit like the inside of a sphere. But most pano software can correct that to a flat projection, and if you have no parallax errors, produce a clean, rectilinear image.

 

[grain elevator]

Is there software that will straighten those curves?

That means changing the projection to a rectilinear one. You can, but for very wide panoramas, you quickly run out of image area in the middle and of resolution toward the sides. Theoretically these issues could be mitigated by taking more images upward and downward and with longer lenses, I'm sure someone has done it but it must be a pain. For very wide views, I would embrace the spherical projection. For moderate ones it's fine. But you don't gain so much resolution, compared to cropping your images.
None of the projections, including rectilinear, is more or less realistic than another and all introduce distortions. Distortions are necessary to fit a three-dimensional world in two dimensions, just like with maps.
You can try this all out without any special equipment, hand-held or with a normal tripod head, and with free software. You can ignore the nodal point issue for trials that don't include a near foreground, you might get minor issues but you can see the effects of different projections.

 

[Sirius Glass]

It sounds like you want a Circuit Camera.

 

[grat]

That means changing the projection to a rectilinear one. You can, but for very wide panoramas, you quickly run out of image area in the middle and of resolution toward the sides. Theoretically these issues could be mitigated by taking more images upward and downward and with longer lenses, I'm sure someone has done it but it must be a pain. For very wide views, I would embrace the spherical projection. For moderate ones it's fine. But you don't gain so much resolution, compared to cropping your images.

You must be using fairly primitive software. I haven't encountered this issue, although I rarely take more than about 8 frames across.

Hugin stitching software is a good starting point, although both Photoshop and Affinity have pretty decent autostitching software in them (I have NOT tried a large multirow pano with those two, however).

Interesting fact, one of, if not the first gigapixel images on the internet was done with this technique, and a 6 megapixel digital camera.

 

[bernard_L]

That means changing the projection to a rectilinear one. You can, but for very wide panoramas, you quickly run out of image area in the middle and of resolution toward the sides. Theoretically these issues could be mitigated by taking more images upward and downward and with longer lenses, I'm sure someone has done it but it must be a pain. For very wide views, I would embrace the spherical projection. For moderate ones it's fine. But you don't gain so much resolution, compared to cropping your images.
None of the projections, including rectilinear, is more or less realistic than another and all introduce distortions. Distortions are necessary to fit a three-dimensional world in two dimensions, just like with maps.

+1

You must be using fairly primitive software.

This has nothing to do with primitive software. Just hard rules of geometry, as grain elevator stated.

• Rectilinear. 2-D shapes (circles, squares...) on a planar surface (wall...) shot head-on will be preserved. 3-D spheres (basket ball...) will appear elongated in projection in the outer parts of the frame. Ditto with heads: you can do 'environmental portraits' with a wide angle lens, but keep the subject's head near the center! All straight lines will remain straight.
  Accomplishes the same as a view camera.
  Hint: one can use the alignment part of a panorama software, applied to a single image, to render an architectural photo as if it had been shot properly with a view camera (spirit level, front shift).
  Field of view must remain strictly below 180°.
• Other: cylindrical, spherical... will lessen the issues with 3-D objects. 360° panorama possible. But (most) straight lines will be rendered as curved, except verticals in a cylindrical projection (with a vertical cylinder alignment).

 

[grat]

I understand the geometric issue, but the way around that is to have more data. If you take a single image in spherical projection and convert to rectilinear, yes, you will have the issue described.

If, on the other hand, you take a series of images, align them on the 'inside' of a spherical projection, and then create a rectilinear projection from that, the effect that you and grain elevator describe simply doesn't happen.

I've made dozens of stitched panoramas over the years, and the internet as a whole has made hundreds of thousands or more, and this isn't a issue. Did you look at the Max Lyons page I linked? Did you look at the 1 row full-size crop of his Bryce Canyon image? Where's the loss of resolution? Where's the elongation?

This is like telling people bumblebees aren't capable of flight, while standing next to a swarm of them.

 

[pbromaghin]

Then there is this:
http://www.appcott.co.uk/VINTAGE6X6/PANORAMICS/FOUR-ISOLETTE-APOTARS/PANORAMIC_4CAM-APOTAR.htm

 

[Dan Fromm]

Then there is this:
http://www.appcott.co.uk/VINTAGE6X6/PANORAMICS/FOUR-ISOLETTE-APOTARS/PANORAMIC_4CAM-APOTAR.htm

Your link doesn't work. This http://www.appcott.co.uk/VINTAGE6X6/PANORAMICS/FOUR-ISOLETTE-APOTARS/PANORAMIC_4CAM-APOTAR.htm does.

Edit: Curses! I pasted the wrong link. Here's the right one: https://web.archive.org/web/2016033...UR-ISOLETTE-APOTARS/PANORAMIC_4CAM-APOTAR.htm

 

[bernard_L]

If, on the other hand, you take a series of images, align them on the 'inside' of a spherical projection, and then create a rectilinear projection from that, the effect that you and grain elevator describe simply doesn't happen.

Sorry, that is just not true. Based on past experience, I will not waste time to "explain" especially if you made up your mind.

Where's the loss of resolution?

I wrote nothing about loss of resolution in https://tawbaware.com/maxlyons/gigapixel.htm, that image is technically superb, as far as I can tell.

Where's the elongation?

The "technical details" below the image https://tawbaware.com/maxlyons/gigapixel.htm inform the readers that the horizontal FOV is 63°. That is moderate as panoramas go, about the H-FOV of a 28mm lens on a FF sensor (24x36mm). Furthermore, there are no objects such as spherical balloons or human heads in sight, whose distortion near the field edges would look un-natural.
The elongation effect of 3-D objects in rectilinear projection) can be seen easily by, e.g. taking a snapshot, with FL 24mm or shorter (full frame equivalent) including faces near the edges of the frame. My only such lens mounts on a FM2n, so no quick proof.

This is like telling people bumblebees aren't capable of flight, while standing next to a swarm of them.

Yes, that would be kind of stupid. Err, am I supposed be the one "trying to tell...". Discussion is starting to derail. Signing off.

 

[grat]

Sorry, that is just not true. Based on past experience, I will not waste time to "explain" especially if you made up your mind.

No, I've made panos. I haven't run into this issue.

Have you produced panoramic images from stitching multiple photos?

I wrote nothing about loss of resolution in https://tawbaware.com/maxlyons/gigapixel.htm, that image is technically superb, as far as I can tell.

You agreed with grain elevator's statement of " for very wide panoramas, you quickly run out of image area in the middle and of resolution toward the sides.", which is what I took exception with.

The "technical details" below the image https://tawbaware.com/maxlyons/gigapixel.htm inform the readers that the horizontal FOV is 63°. That is moderate as panoramas go, about the H-FOV of a 28mm lens on a FF sensor (24x36mm). Furthermore, there are no objects such as spherical balloons or human heads in sight, whose distortion near the field edges would look un-natural.
The elongation effect of 3-D objects in rectilinear projection) can be seen easily by, e.g. taking a snapshot, with FL 24mm or shorter (full frame equivalent) including faces near the edges of the frame. My only such lens mounts on a FM2n, so no quick proof.

Oh, so now it only happens on "very wide" images-- how wide? 80°? 90°? 120°? The goal is not to produce a fisheye projection. The goal is a large, rectilinear projection based on multiple smaller images. And I said, for a single image, you're not wrong, but if you're stitching a pano, you are, by definition, not using a single image, therefore, you can do all the testing you want, even if the only lens you've got is a 200mm telephoto.

Yes, that would be kind of stupid. Err, am I supposed be the one "trying to tell...". Discussion is starting to derail. Signing off.

You're telling me that something I've been doing for nearly 20 years (Since the Bryce Canyon image was posted, in fact) doesn't work, and refusing to explain why, because "I've made my mind up"-- and now you're leaving the conversation.

 

[pbromaghin]

Your link doesn't work. This http://www.appcott.co.uk/VINTAGE6X6/PANORAMICS/FOUR-ISOLETTE-APOTARS/PANORAMIC_4CAM-APOTAR.htm does.

Actually, neither of them works for me. My firewall has changed since the last time I looked at it and I posted in hopes that it was still there.

 

[Dan Fromm]

Actually, neither of them works for me. My firewall has changed since the last time I looked at it and I posted in hopes that it was still there.

Thanks for the bad news. Seriously. I pasted the wrong link, have edited the one you want into post #12 above.

 

[pbromaghin]

Thanks for the bad news. Seriously. I pasted the wrong link, have edited the one you want into post #12 above.

Oh, that works great! Thank you. It's such a cool idea that I might have to try it someday.

 

[grain elevator]

No, I've made panos. I haven't run into this issue.

Have you produced panoramic images from stitching multiple photos?



You agreed with grain elevator's statement of " for very wide panoramas, you quickly run out of image area in the middle and of resolution toward the sides.", which is what I took exception with.



Oh, so now it only happens on "very wide" images-- how wide? 80°? 90°? 120°? The goal is not to produce a fisheye projection. The goal is a large, rectilinear projection based on multiple smaller images. And I said, for a single image, you're not wrong, but if you're stitching a pano, you are, by definition, not using a single image, therefore, you can do all the testing you want, even if the only lens you've got is a 200mm telephoto.



You're telling me that something I've been doing for nearly 20 years (Since the Bryce Canyon image was posted, in fact) doesn't work, and refusing to explain why, because "I've made my mind up"-- and now you're leaving the conversation.

Yes, indeed it happens only with "very wide" images, that's exactly what I wrote and what, as you say, bernard_L agreed with.
I thank that's the issue, the website with the stitched image of Bryce Canyon you linked to - in my understanding that's not panoramic. A panoramic picture I understand to be much wider than tall.
Anyhow, just try making one of those smartphone camera panos - do you think they curve horizontal lines just to look funny?
Or try something like three pictures with a 28mm lens next to each other (typical use case for me), of something with horizontal lines, like a tiled wall. Stitch in Hugin or whatever. You'll see what I mean.

 

[grat]

The issue is, this conversation followed a fairly common standard on the internet. OP says "I want to do X", and someone else says "You can't do Y, because of <problem>".

I understood the OP's question to be, "how can I take multiple photos and produce a wide image, such as 6x17, with digital?". They also wanted to know why they couldn't simply shift the camera left and right, which I attempted to explain as well, and that also reinforced my belief that they want to take a stitched panoramic image.

As you, and bernard_L, pointed out, a single ultra-wide image will have some distortion due to the nature of the beast-- but an ultrawide lens is not required and makes things more difficult, and that didn't appear to be what the OP was asking about.

Why would I use a 28mm lens? Unless it's of exceptional quality, there's going to be some distortion (which can be corrected in software pretty easily, assuming it's a well-known lens, and if you correct prior to stitching, problem is resolved). Even if it is a remarkably flat, sharp, high resolution lens,

The smartphone panos are curved because the camera wasn't kept level, or because the software was cheap, or the angle was simply too wide.

A 6x17 image would have an effective crop of 0.28 (give or take). So for a 50mm effective lens, your 39.6° FoV would be more like 141°. Three 28mm images (without overlap) would be ~196°, and yes, that would be a problem.

Far better to use something like an ultra-sharp 40 or 50mm lens, with multiple images, to produce a flat image, and then stitch it with Hugin or similar. The end result will be a flat, non-distorted image without the type of distortion or resolution issues you describe.

Personally, I'd take about 5-7 portrait images to maximize my resolution, and aim for something akin to 100 to 120 degree FoV. With a pan head marked in degrees, rotating the camera around the nodal point, would be trivial, and the end result would work just fine.

 

[grain elevator]

The issue is, this conversation followed a fairly common standard on the internet. OP says "I want to do X", and someone else says "You can't do Y, because of <problem>".

I understood the OP's question to be, "how can I take multiple photos and produce a wide image, such as 6x17, with digital?". They also wanted to know why they couldn't simply shift the camera left and right, which I attempted to explain as well, and that also reinforced my belief that they want to take a stitched panoramic image.

As you, and bernard_L, pointed out, a single ultra-wide image will have some distortion due to the nature of the beast-- but an ultrawide lens is not required and makes things more difficult, and that didn't appear to be what the OP was asking about.

Why would I use a 28mm lens? Unless it's of exceptional quality, there's going to be some distortion (which can be corrected in software pretty easily, assuming it's a well-known lens, and if you correct prior to stitching, problem is resolved). Even if it is a remarkably flat, sharp, high resolution lens,

The smartphone panos are curved because the camera wasn't kept level, or because the software was cheap, or the angle was simply too wide.

A 6x17 image would have an effective crop of 0.28 (give or take). So for a 50mm effective lens, your 39.6° FoV would be more like 141°. Three 28mm images (without overlap) would be ~196°, and yes, that would be a problem.

Far better to use something like an ultra-sharp 40 or 50mm lens, with multiple images, to produce a flat image, and then stitch it with Hugin or similar. The end result will be a flat, non-distorted image without the type of distortion or resolution issues you describe.

Personally, I'd take about 5-7 portrait images to maximize my resolution, and aim for something akin to 100 to 120 degree FoV. With a pan head marked in degrees, rotating the camera around the nodal point, would be trivial, and the end result would work just fine.

Dude, the op said: I want to make a panorama by "sliding" my camera, as in not turning it. And indeed someone said "you can't do that".
As for the geometry issues, it's like someone from South Florida is telling someone from Canada "winter tires are a scam. I've never needed them, nobody needs them". You operate within what I called "moderately wide panoramas" and/or not panoramas at all, but stitched images, and within that framework, you're right.

 

[grat]

What he actually said was (emphasis mine):

Panorama shooting equipment​

I've been wanting to do this for quite some time with my DSLR. I've been reading and watching a lot and I'm confused. Since I am wanting to shoot at the maximum 6x17ish perspective I am trying to figure out the piece(s) of kit I need to purchase.

I have a Nikon D7100 that I am plenty happy with and LOVE my Sigma art lenses. Don't want a new camera, so I will be using that. Here is my issue

There is a lot of discussion about rotating around the nodal point of the lens and not rotating around the base of the camera. This makes sense except for the fact that it is still a rotation and thus, in my mind will still create a curved image.

Then a stumble on only a few mentions, but no discussion of a slider to move the camera horizontally. This make a lot more sense to me, especially from my large format camera usage. No curved image.

I attempted to explain why rotating around the nodal point would produce a superior result, even though technically, it does produce a spherical image, which can be corrected to rectilinear. Because sliding the camera is less effective (it can work, but requires a LOT of sliding to get any substantial image shift), due to the lens and film plane being fixed relative to each other, unlike shift-tilt arrangements where the film plane and the lens can be somewhat uncoupled.

And then you and bernard_L jumped in and starting going on about super-wide panoramic photos produced with very wide angle lenses, loss of resolution and detail, which, as you (eventually) point out, is almost totally irrelevant to the question.

I have attempted to answer the OP's question. I have attempted to clarify comments by you and bernard_L, and I have been insulted, patronized and dismissed for my efforts.

It's no wonder so many posters on this forum have left in the past few years.

 

[5150Bronco]

I started working on nodal point and this works. It does not create distortion.

I have done pretty large pano's, not giga ones and getting the nodal slide is a good idea.

IG: Jr64photography

You can check out the BW pano of the lake.

 

[sfphoto]

I am wanting to shoot at the maximum 6x17ish perspective I am trying to figure out the piece(s) of kit I need to purchase.

I have a Nikon D7100 that I am plenty happy with and LOVE my Sigma art lenses. Don't want a new camera, so I will be using that.

Hello Mark! Here is my take-

1) Nodal point is essential for interior / close up rather than distant landscapes

2) Am thinking to equal the angle of view of my Linhof 617 w/90mm you would need three image files from a 50mm lens on your D7100 (3 to allow for overlap, but 2 might work)

3) Hugin (free) can stitch 3 images easily and if making a landscape it should look natural w/o nodal point use.

A tripod will help for landscapes but is not essential like it is for interiors. Manual focus & exposure and pre-set white balance (not AWB) are best.

----

First image - nodal point set, multi image using a rotation device on a tripod, then de-fished w/ Hugin



You can also use a photo editing program and overlap then use the select tool feathered to remove the overlap as done below. Two frames, 35mm EQ, handheld. Having the camera level (bubble attached if no digital option) prevents the 'End of the Earth effect'.



BTW I have to rotation adapters for sale listed here:

Sold - 67mm Polar filter

$23 67mm Polar filter

www.photrio.com

 

[xkaes]

A panoramic picture I understand to be much wider than tall.

A panorama can be horizontal or vertical. Anyone who has hung around a sequoia or grain elevator knows that.

 

[kinglang]

It sounds like you want a Circuit Camera.