Jump to content

2 image panoramas - need simple, effective software


Recommended Posts

My objective is to attain clear, crisp images up to 16X24 with a dSLR, probably Canon (XTi) or Nikon

(D40X/D80).

I have a Canon A630 now that takes an image size of 13X18 @ 180 dpi (or so).

 

Most forums say that my desired size of output, for images that rival medium format,

It seems to my simplistic logic, that if I use the 'long' side of my image as the 'short' side of my eventual

stitched image, I can create roughly an 18X24 (accounting for slight overlap) image @ the same 180dpi.

Correct?

 

I'm not sure if the Canon Photostitch can stitch images across the long side.

I also have CS2 which has a stitching mode.

 

Also, I've read about the nodal point and my 630's tripod mount is slightly off-axis to the lens.

For urban street scenes and architecture shots, is this a killer for 2 image panoramas?

 

Am I correct in my basic assumption that using 2 stitched images will result in my desired final image?

If so, which panoramic software will get me there simply and effectively?

 

Thanks to all, Bob

Link to comment
Share on other sites

Perhaps you can rotate the images in Canon Photostitch in order to stitch them on the long edge.

 

Inexpensive programs like Panorama Factory and PTgui have a lot more flexibility and the ability to make very difficult stitches perfectly. Of the two, PTgui (which I recently acquired) is by far the better program.

 

There will always be detractors who say you can never make a good picture with a digital camera. Their opinions are worth exactly what they cost. Still, 180ppi is too coarse, and may show pixelation. Reduce the overall dimensions or resample so that you get 300ppi and you'll be fine. That said, the larger sensor of a DSLR and the availability of better glass would make a tremendous difference in quality.

Link to comment
Share on other sites

Parallax is only an issue with objects in the foreground. I have shot hand-held fifty-part multi-row panoramas and they turned out just fine...

 

If you do want to get a panoramic head I can recommend the Panosaurus for its reasonable price. For stitching CS3 is pretty good, but Autopano Pro is much better.

Link to comment
Share on other sites

I use Photoshop Cs3 (aka: PsCS3) Photomerge tool for most of my panoramic stitching work --I am shooting with a 1Ds MArk 3 generally in portrait orientation the way you want to use your lower resolution cameras --

 

Photomerge in PsCS3 is far more advanced and does a much better job than Photomerge in PsCS2 did. As the upgrade price for goign to PsCS3 fro ma registered version of PsCS2, psCS or Ps 7 is under $170.00 that is my software advice. There are other programs that can, once set up and programed properly, do a better job but the PsCS3 version Photomerge does a very fine job by itself. The panoramics I am making are 16" high and as much as 72" long and combine as many as 12 frames and take in an angles of view of up to 270 degrees along the long dimension.

 

Stitching it is definitely a way to get phenomenal detail resolution. I work with a stock agency that specializes in high resolution panoramic work and they like what I am doing too.

 

But as with everything else in this life you have to start with good materials (data in this case) going in so the hardware side is a little more complicated. For starters, you need to lock both focus and exposure so the camera needs to be in full manual mode. While some people may argue the point, what works best for me is to use single focal length lenses instead of zooms. I have also found that the stitching works best with lenses in the 28mm to 100mm range (on the 24x36mm format of the 1Ds mark 2 so with an APS-C camera don't go wider than an 18mm unless you really have to. A cable release is a good idea too.

 

You obviously need a solid tripod and head. What I've found that works best is to use a panning base on top of the head ( I use the really Right Stuff PCL-1), an RRS MPR-CL II nodal slide and an Arca-Swiss type "L" QR camera plate (once again I use the RRS product designed for my camera. With the head on the tripod ( I use an Arca_Swiss B1 Monoball but now think that a geared double tilt head like the Bogen/ Manfrotto 405 or 410 would make the scut-work of leveling the head go faster) would make for a better leveled the panning base on top makes for much smoother pans and fewer stitching errors. The purpose of the nodal slide is to get the nodal point for the lens you are using exactly over the rotation point. This also makes for much better stitching with fewer errors. Even though the RRS nodal slide and the RRS panning clamp have bullseye type levels built in I still triple check how level the camera is to both the vertical and horizontal planes with a separate high quality handheld level.

 

Make sure you overlap your frames when you shoot. I generally overlap by about 25-33%. Also for some reason that not even people at Adobe can explain, if I have glitches in the stitching the first time I process the iamges through PsCS3's Photostitch tool they almost (not always but 95% of the time) disappear after running the same images through a second time. Beyond three times I stop seeing any improvement.

 

To boil all of this down: expose well for the overall iamge; use a good lens; nodal point above the rotation point; overlap; and stitch twice in PsCS3 Photomerge.

 

One last technical note (learned the hard way) don't use a polarizing filter on the lens if you the result will be a wide angle view. While a 28mm lens (or the APS-C angle of view equivalent) lens may be fine with a polarizer for a single shot you have to keep in mind the final total angle of view.

 

One last aesthetic note: In your mind's eye frame the final iamge so that the primary organization --the way a viewer reads the photo , goes from left to right as your primary compositional goal and foreground to background as the secondary goal.

Link to comment
Share on other sites

"My objective is to attain clear, crisp images up to 16X24 with a dSLR ... Most forums say that my desired size of output, for images that rival medium format"

 

Before buying anything, go download the open source Hugin from http://hugin.sourceforge.net. It's free, and frankly works better than a handful of commercial programs I've sampled.

 

It's pretty painless to extract ~25MP of resolution from a frame of 6x7. After accounting for overlap, you'll need to stitch about 5 frames from a 10MP DSLR to get an equivalent image.

 

Also, it is not necessary to limit the stitch to a single row. Multi-row pano's work perfectly well and is not any harder to shoot. It is possible to do digital composites like this that exceed large format film in resolution. The limits are dictated more by patience than anything else.

 

A pano head is necessary to getting good stitchable frames in the general case. However, all you really need is a standard tripod to get started; limit the images to far field landscapes and parallax is not an issue. Actually, even a tripod isn't strictly necessary. Most times the only equipment I have handy is a compact Canon A550 digicam. I've a number of successful pano's that were done handheld, with the camera in continuous shot mode as I swept the horizon.

Link to comment
Share on other sites

Photoshop CS3 has a much improved photomerge. Far better than CS2.

 

Have you considered buying a tilt shift lens to go with your planned DSLR? This is a pretty painless way to take 3 images quickly that will stitch together painlessly. It is expensive, however.

 

Alternately just buy one of the new 12mp cameras. Should give you what you want in one shutter click.

 

(It should also be noted that I've had 6mp images blown up to 16x24 that have looked very nice. Depends on subject matter. It will often work fine. Sometimes not.)

Link to comment
Share on other sites

Two points worth adding:

 

If you use a good spherical stitcher like PTGui or Hugin, you do NOT have to keep the camera level and they will handle extreme wide angle lenses with ease - even 180 degree fisheye lenses.

 

Guidance on finding the required no-parallax point (not, incidentally, the nodal point but the entrance pupil) can be found here: http://www.johnhpanos.com/epcalib.htm .

Link to comment
Share on other sites

Another method is to merge the images manually using layers, and then clean them up with Eraser tool while leaving them on their layers. Just shared the following with a buddy recently:

 

"How do you do your panos? The stitching itself?

 

Reason I ask is because I'd been doing them with Photoshop's Photomerge. But sometimes results weren't what I'd hoped, and on big files I didn't have enough RAM for the computer to do its magic unless I let it resize the images smaller during the Photomerge process.

 

So I decided to try a method I'd read about in Lee Frost's book on pano photography, merging them manually using layers. It sounded clunky, but WTH...

 

You have Image 1 on a layer, then paste in Image 2. You set opacity on Layer 2 to maybe 30%, and then use the Move tool to get it close. Then zoom in tight, and use the arrow keys to move the pasted image *1 pixel at a time* till you have it just right. Then set opacity back to 100% and look for glitches. You can erase glitches in the new layer using the eraser tool, until you've got it perfect. You can also adjust contrast / brightness in the incoming layer till it's perfect.

 

I tried it last night, and works like a charm. Faster than having the Photomerge crash..."

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...