Is the MKII perfect for stitching images

[Tim Gray]

Some comments:
Technique is more important than the camera - in fact I'd go so far as to say the camera is irrelevant (unless you're thinking of something like the Hassy xPan - lens' though, can be a factor).  Tripod and Camera have to be level.  If you have close foreground content, parallax can be a factor, and you need to know the nodal point of the lens and be able to pivot around that point, not the point where the camera attaches to the tripod - a panorama vr plate helps (ie not just one that turns, but one that lets you adjust the pivot point of the camera).  For some lenses, the nodal point changes with the zoom, so it's tough to easily replicate  the nodal point for the intermediate focal lengths.   The farther away the subject (including foreground) is, the less a problem parallax becomes.  You can do the stitching yourself, but PS CS now has a nice utility.

Finally, tilt/shif lenses make the parallax problem go away, you can use the shift to get basically a 2 frame pano without any hassle - but all are fixed focal length (Canon has 3).

Do a google search on panorama and virtual reality and you'll find lots of resources.

Also some good recent threads on DPReview.

[JJP]

I have done some stitching before, so I fully understand the parallax problem.  That is why I just purchased a horizontal and vertical vr head.
The reason why I posted:  if your taking some images containing clouds (or anything that has slow movement) then imo speed would be important ie....keeping the time as short as possible to shoot the series of images which you plan to stitch together so that everything aligns properly during post processing.
jj

[JJP]

And what specs makes a camera ideal for that application?
jules

[George Barr]

I do my stitching with a 10D, having used a Sony 707 before that. I did find that the 10D with less barrel and pin cushion distortion in it's lenses makes for easier but not better stitching. I use PTMac software from kekus.com, ptasemmbler is a similar programme for pc's. I shoot a single row of images horizontally, with the camera in vertical position. I use an arca ball head and I made a tripod adapter for the arca head which has it's own arca base at one end. It can slide back and forth in the ball head, the camera goes in the fitting at the end of the 8 inch wooden rail. I set up my camera on my deck with a variety of subjects (flower tubs) at varying distances but near each other in the view finder. I swung the camera back and forth until the near and far objects did not move relative to each other in the viewfinder as I swung the camera horizontally. I then marked the position on my wooden slide. I found that the 17-40 and the 28-105 had the same nodal point regardless of what focal length I set the lens at but the 70-200 had a huge shift in nodal point which was exactly opposite of what I would have predicted - the nodal point for 70 mm. was further to the front of the lens than when at 200 mm by more than two inches.

The slide is simply a piece of oak with angled routed grooves (or cut with table saw) to match the shape of arca base plates. It was easy to make (you might have to buy or borrow the right router bit shape). My camera store sells arca bases (what the plate fits into) though in fact I ordered a lever release from really right stuff to go with my L bracket from them - a very slick  combination.

Anyway, you asked what camera and the first answer indicating it doesn't really matter is pretty true - but having to stitch a single row is a lot faster than doing rows and columns and three 10D vertical images with about 35% overlap produces a VERY nice 13X19 of about 12 megapixels and at least as much resolution as the 1Ds.

I have run into one problem though and that is shooting images vertically with the camera in horizontal position, then my little slide doesn't fix the rotation round the nodal point problem. I plan to build an L bracket for my tripod which will hold my current slide, then I can rotate round the nodal point in either direction - I wonder if it will make the arca ball head superfluous. Hmn... Mind you 90+ percent of my images are horizontal and I recently shot some ice which was directly in front of the tripod - using just the slide, I simply looked up from the axis of the ball head and slid the slide back and forth till the nodal point of the lens passed through the line extending up from the axis of the ball head. - worked a treat - able to get image matching down to average of < 2 pixels in 20 megapixel image - ie. invisible seams.

Does that help, contact me if I can be of further info -

[Lin Evans]

In essence, and in respect to your question about the Mark II specifically, any dSLR shoots fast enough to minimize cloud movement as an issue.

The important thing is to have a feature which allows you to keep the same exposure so the stitching software or PhotoShop doesn't need to "push" or "pull" the images to a great degree to match exposure at stitch points.

Assuming you have reasonably good glass, with about a 30 percent overlap in your frames you should be able to get dynamite stitched images with normal care - especially of landscape type captures.

For example, Max Lyons had a 1 gigapixel plus resolution image printed on a LightJet and displayed at PMA this year. The 3 gigabyte sized file was composed of 196 separate frames captured with Max's six megapixel Canon D60 and stitched with Pano Tools. The print was 8 feet by 12 feet and truly has to be seen in person to appreciate. It shows much better resolution than available with an 8x10 film capture.

Obviously, few of us will ever even attempt such a feat, but it does demonstrate the possible. Here's a link for more information. And yes, the Mark II should make an excellent tool for stitched images.

http://www.tawbaware.com/maxlyons/gigapixel.htm