What lens or how would this photo be taken?

Wide angle lens. Probably around 15mm.

edit
after looking at it on the pc vs my phone, lower than 15mm. Probably closer to like 10mm.
 
That's possible too. I didn't see any EXIF data on the image itself, so I'm not sure what the focal length was. It's for sure pretty wide. Makes me miss the 10 - 22mm canon lens I was borrowing from a friend a couple years ago. :(
 
It could also be a fish-eye, with an adjusted distortion slider. The Nikkor 10.5mm for example, is capable of producing a 180 degrees images, which this is what this appears to be.
 
I think it is a stitched image panorama made with an Equirectangular projection in the software.
 
I would say it was a 15mm fisheye. The face of the (servers?) are severely bent, and the tile floor makes a "U" shape. I'm guessing that actually the tile floor is completely straight, as are the machines and NOT an "L" shaped (or round) series of aisles. The only way then to get that kind of distortion and angle (180 degrees) would be from a fisheye.

If it is a panorama (which I doubt) it is the laziest I've ever seen, as a lot of the issues this image exhibits could have easily been corrected. IE the massive floor tile and ceiling tile distortion. The metal racks above as well. The severe appearance of bending servers rather than straight... etc.

============================

EDIT: After looking at other images of the ALMA supercomputer, I'm doubling down on my fisheye hypothesis.

An angle down the hall:
http://www.eso.org/public/archives/images/screen/eso1253c.jpg
What it looks like completely from the side, if it wasn't distorted:
http://www.aanda.org/index.php?opti...=/articles/aa/full/2007/05/aa4519-05/img1.gif
This is also another fisheye shot, but it shows what the servers look like from the corner:
http://almaobservatory.org/gallery2...DownloadItem&g2_itemId=4139&g2_serialNumber=2

(All non-hotlinked intentionally)


You'll note that like I said, the servers are actually straight, as are the aisles between the servers. This distortion looks completely like fisheye distortion to me.
 
Last edited:
If it were a fisheye, the verticals would be curved and severely distorted too, but because they are straight, and the fact that its 180 in the horizontal, it has to be stitched

Fisheyes are 180ish in all directions

The stitching is lazy, but for that, I'd say that it's simply automated by photoshop
 
If it were a fisheye, the verticals would be curved and severely distorted too, but because they are straight, and the fact that its 180 in the horizontal, it has to be stitched

Fisheyes are 180ish in all directions

The stitching is lazy, but for that, I'd say that it's simply automated by photoshop
So would the lens used be something like a 17-40mm ultra wide angle, possibly on a cropped sensor, and the camera rotated 90 degrees to have more vertical coverage?

EDIT: I'm thinking it would probably be a full frame actually.
 
If it were a fisheye, the verticals would be curved and severely distorted too, but because they are straight, and the fact that its 180 in the horizontal, it has to be stitched

Fisheyes are 180ish in all directions

The stitching is lazy, but for that, I'd say that it's simply automated by photoshop

You bring up a good point about the verticals, however the other point of 180 degrees in all directions doesn't quite work, as full frames (as an example) are 3:2, so they have the tops and bottoms cut off, as the sensor itself isn't square. I'm also not certain to the degree in which fisheyes can be fixed in post to rectilinear (although I have seen images demonstrating correction). Would you say it's possible to have the verticals corrected in software while still have the bent look?

I guess the problem I see is that if they didn't use a fisheye then they were just lazy, not only with the post work, but also the setup and the taking of the shots themselves. I suppose you could be right though. It wouldn't be the first time I was wrong.
 
I can't say more other than, it's not a fisheye, and it was stitched. How many shots were taken and what lens used, I don't know. The vertical coverage doesn't seem much, so probably a normal 18-55?

For all we know, this was a P&S, or even a P&S with a panorama feature. There's nothing to indicate that it was a dSLR

Edit: ^^ you are right about the ratio, however, true fisheyes render as a circle on a rectangle frame. As far as the correction, they can be fixed to rectilinear and I've done that a lot, but not in one direction afaik

See sample
http://upload.wikimedia.org/wikipedia/commons/6/67/Glasgow_Science_Centre_fisheye_from_tower.jpg
 
Last edited:
If it's stitched, then it's a wide angle stitch.
 
The image looks like more than 180 degrees to me. I have a 10.5mm fish-eye, and the distortion of the image really doesn't seem to match what you would normally see from that lens. A fish-eye has kindof a bubble-effect centered in the frame. Notice how the wall on the left edge flares out instead of tapering down away from the center.

I can't say more other than, it's not a fisheye, and it was stitched. How many shots were taken and what lens used, I don't know. The vertical coverage doesn't seem much, so probably a normal 18-55?

For all we know, this was a P&S, or even a P&S with a panorama feature. There's nothing to indicate that it was a dSLR

Edit: ^^ you are right about the ratio, however, true fisheyes render as a circle on a rectangle frame. As far as the correction, they can be fixed to rectilinear and I've done that a lot, but not in one direction afaik

See sample
http://upload.wikimedia.org/wikipedia/commons/6/67/Glasgow_Science_Centre_fisheye_from_tower.jpg

Yeah, this ^^^. Hard to say what camera/lens, but I'm ~95% sure it's stitched. I've seen some auto-panos people have made with the new iPhone that have similar distortion to this. The shot could probably have been done with an iPhone given some good lighting and maybe a tripod/mount to keep it steady. It's not too much of a stretch.
 
I think it is a stitched image panorama made with an Equirectangular projection in the software.

I came to the above conclusion due to the straight verticals, and the straight tiles which fall off to the left and right, and even more distinctly the left most tile almost splits and goes to the left and right.

See this pool of images at flickr: http://www.flickr.com/groups/equirectangular/pool/

There are many similar examples.

Technique (not sure if this is the best example): http://www.flickr.com/groups/equirectangular/discuss/72057594138421542/
 
Last edited:
I came to the above conclusion due to the straight verticals, and the straight tiles which fall off to the left and right, and even more distinctly the left most tile almost splits and goes to the left and right.

See this pool of images at flickr: http://www.flickr.com/groups/equirectangular/pool/

There are many similar examples.

Technique (not sure if this is the best example): http://www.flickr.com/groups/equirectangular/discuss/72057594138421542/

I agree.
 
I have to laugh because I used to do stitching shots like this before I got my own fisheye lens... The fact is that until now I didn't realize what the advantage of the photo stitching process can do that the fisheye can't --and I have examples of it on my own site!! Thanks everyone.
 
Back
Top