Big-Time, Big-Event Pano

Multimedia producers Bedel Saget, Gabriel Dance and Raymond McCrea Jones of the New York Times website created a fine panorama of the parade of athletes in the opening ceremonies of the Beijing Olympics. These guys have come a long way from where they started. Practice makes perfect. I'm hoping we see a number of panos from this team over the next couple of weeks. I suspect they have something very cool up their collective sleeve!

Introduction to Panoramas on digitalartwork.net

A great resource for multimedia journalists is digitalartwork.net, the site of the multi-talented Zach Wise. He is a multimedia producer at the Las Vegas Sun who I met at the 2007 NPPA Multimedia Immersion in Portland, OR. He's also a great teacher.
As part of his multimedia work, he shoots 3-shot panoramas with a Sigma 8mm and a Canon 5D. This is a great setup for photojournalists because it's fast, enabling the capture of dynamic scenes. There's sufficient overlap between frames to allow for subject movement in those areas.
Zach has put together a very cool demo that shows the overlap between frames. What happens in those areas between frames is crucial to the success of our panoramas.
The lasvegassun.com website features a link to Zach's panoramas (including one from last week's NASCAR race) and the staff's multimedia on its home page. I'm very envious.

PTGui for Beginners from John Houghton

John Houghton has written several excellent tutorials on his site. His PTGui for Beginners is as good a tutorial as I have found for using PTGui in its Basic mode. It's written for Windows users, but if you're using a Mac, the information about PTGui is nearly identical.

If you've been trying PTGui and are getting frustrated, grab a set of pano images (John provides the set of images used in the tutorial), set aside a couple of hours and go through this tutorial - slowly.

A couple things:
The trial version of PTGui will work just fine for this tutorial.
At first, I would recommend ignoring the part about Smartblend, as the blender built into PTGui will work just fine.
Under the header "Viewing the Stitched Panorama", Mac users can also use Cubic Converter, as well as Pano2VR to convert the equirectangular image into a QTVR file.

Panoramas from the trail

I have been on the road this week in South Carolina with the Democratic candidates. My primary mission was to shoot stills for the paper and online slideshows. We have been running a daily slideshow on the politics page on charlotte.com.
Of course, I also wanted to shoot panoramas!

The rig I used is simple: Sigma 8mm, Nikon D70s, hardwire shutter release, Nodal Ninja 3 head, tall monopod. I keep the strap on the camera and with the monopod retracted, it fits over my shoulder, just like having a third still body. One pano (Monday MLK rally) was handheld, holding the camera high above my head. You can see my shadow in the image.
This lens and camera combo provides enough overlap to allow shooting just four frames in each situation.

When in the buffer at campaign events, I'd kneel and extend the monopod fully. Drawing it as close as possible to me, I'd visually select a spot on the ceiling and try to keep the monopod aligned with that spot as I rotated it 90 degrees for each frame.
In other situations, I'd either stand behind the camera and move along with its rotation, or put the monpod on top of something firm (riser, rope-line stand, abandoned TV tripod). The problem with standing behind the camera and moving with it is that you need quite a bit of floor space, which is at a premium in these situations.

I learned several things.
First, if you're using this rig above your head, don't look up! Your face will be in the nadir of the pano.
Second, practice this at home! If the situation is moving, you have to be quick. Looking at my EXIF data, I was shooting a frame a second in some situations. Click-rotate-click-rotate, all the time trying to keep the monopod relatively vertical.
Third, shoot lots of images. It's very hard to look behind you while keeping the monopod vertical, so you have no idea what's going on back there.

Stitching was helped immensely by setting vertical control points in each frame. This allows PTGui to compensate for variations in the verticality of the monopod between images. In the Control Point window, have the same frame in both panes. Look for vertical lines (door frames, walls, etc., something architectural). Place a point on one end of the line the left pane, then a point on the other end of the line in the right pane. Try to set at least two in each frame, then let PTGui generate control points for the overlapping frames.

Stitching also was helped by adding control points manually in the nadir and zenith regions of each overlapping image pair.
There will be large errors in your initial optimization. I first optimize with no parameters (FoV, a, b, c) selected. Delete the worst of the control points (CPs). I try to visualize the control points as a fever chart. If there's a big drop, I'll delete the points above the drop.
Then select FoV and re-optimize. Delete bad CPs, add the b parameter and reoptimize. Repeat for c, then a. It's surprising how few CPs are needed in the end to optimize.

As you can see, a couple of the final panos have their fields of view restricted. Since I'm not shooting nadir and zenith on these images for the most part, there are unretrievable stitching errors in the polar regions, which often don't contain much content.

What would I do differently the next time?
I'd get a more substantial remote release. I had a cheap one for the D70s which broke, forcing me to switch to another camera (D200) with poor performance in high ISO (Saturday Obama finale).
The holder that flag bearers use would be a useful addition for my belt. I could put the monopod in that and use myself as an extender.
I'd get a full-frame camera to allow greater overlap between frames, plus I could possibly shoot only three frames which would make shooting action easier. Anyone want to loan me a D3?

I hope this is enough info to help you try getting off the sticks and shoot more dynamic subjects. Questions? Just ask.

NOTE: this post updated 9 August 2008 to change links to reflect change in the Charlotte Observer site domain name.

In the Bathtub with NYT

The latest from the New York Times team of Ray Jones and Gabriel Dance. This time, they take us to two levels in the big hole at Ground Zero to take a look at the construction going on in lower Manhattan. The ambient sound is a good touch. Zoom navigation is a bit different here - they've opted to use a right-click (or control-click) submenu to control zoom, rather than keyboard+click shortcuts. I've asked Ray about the tripod being left in the image. As soon as he replies, I'll update. And I will post a nadir patch tutorial as soon as I can.

UPDATE: This from Ray via email: "Well, the tripod thing is something we'll do better next time. I didn't shoot any handheld nadir shots, and that was my fault. And I didn't feel comfortable just trying to clone the tripod away. So we felt it was best to just leave it. I think we're improving each time and it will take us a few more times probably before we nail everything. "

This project was the lead tout on the nyt.com site for awhile over the weekend. That'll bring panoramas to the attention of more viewsers.

I find it useful in this kind of situation - deep shadowed areas and bright sky - to use a Photoshop action called HDR for Dummies written by famed panographer Jook Leung. (You've seen his Times Square at New Year's Eve panos).

Rather than bracket exposures (and risk lots of subject movement), I acquire two tiff files from each RAW image, one adjusted for shadows, one adjusted for highlights. The HDR action blends the two, allowing for a longer dynamic range. Kind of reminds me of split-filter printing on variable-contrast paper, back in the b/w darkroom days.

Your views on using this kind of blending in photojournalism? Discuss. Please feel free to add to the comments.

Happy New Year!

OK, let's get back to work!

To kick off the new year, here are a couple of panoramas from New Year's Eve in Rio de Janiero. I found these on the PanotoolsNG Yahoo group.

Willy Kamena

Ayrton Carmago

Willy Kamena of Bremen, Germany was visiting Ayrton Camargo of Rio de Janiero, and the two went to the beach for the New Year's Eve fireworks. You can see both Ayrton and Willy at work in these two images.

Note they are both using fill flash (absolutely necessary in this situation), Ayrton's fill balance is just about perfect with the light provided by the fireworks. A reader on the Panotools NG group asked him about his flash rig, here's what he had to say: "I use the Gary Fong diffuser, always pointing to the sky. Doesnt matter where I'm (at) if there's a ceiling or not! And I put the flash NOT on TTL. Always use on A (automatic)."

Ayrton is using a pole rig to get his camera up higher. Willy is using a monopod, but at eye level.

Also note the use of simple ambient sounds embedded in the full-screen html. I think this adds to the "being there" effect.

Why are these panos cool? These photographers are using the techniques we need to shoot photojournalistic panoramas. They're traveling fast and light, using monopods and poles rather than tripods. They're using fill flash to reveal content in areas where there is nothing but darkness. And they're working under difficult lighting conditions with moving subjects.

Plug-ins for Flash Panorama Player

For years, panoramas have been presented on the web using QuickTime VR or Java-based players.
Flash Panorama Player ("FPP") has come on the scene recently and is rapidly becoming the presentation tool of choice for VR photographers. Created by Denis Chumakov, FPP is a viewer engine built on Adobe Flash Player.
FPP is a great tool. So is Flash. They're great together, and even better if you are a programmer. I am not.
At its most basic, FPP can be used to publish panos just by copying files. Here's an example of what you get when you simply copy files. It's a basic presentation, but it works well. Note the inertial damping of the motion, and the window resizing. These are just two of the features of FPP.
Fortunately, Patrick Cheatham and Zephyr Renner ARE programmers. They've created Flashpanos.com, a site dedicated to making available inexpensive FPP plugins and (this is the best part) rewriting the documentation included with FPP. They describe what they're doing this way: "Some of this is taken from existing documentation, some is reworded, some is from experience. The idea isn't to replicate existing FPP documentation, but to rework and expand it."
As an example, they've posted FPP documentation for fullscreen panoramas here. More documentation is in the works.
If you have a colleague who is a Flash programmer, turn them on to this now. Let them help you build an integrated Flash presentation for your panoramas.