Somebody in this thread suggested AutopanoPro from France, so I bought it. I like it so far but haven't figured out (haven't had the time to try) the vignetting correction yet. (Baldy crosses his fingers.)
[FONT=Arial, Helvetica, sans-serif]"The single most important component of a camera is the twelve inches behind it." -A.Adams[/FONT]
0
BaldyRegistered Users, Super ModeratorsPosts: 2,853moderator
edited September 18, 2008
The problem is, I can't tell who the show host and guest was on the second story TV. All I can tell is one of them is wearing a blue suit and tie with a white shirt.
Maybe if I shoot this two rows with an 800mm instead of 600, I can correct this deficiency.
The problem is, I can't tell who the show host and guest was on the second story TV. All I can tell is one of them is wearing a blue suit and tie with a white shirt.
Maybe if I shoot this two rows with an 800mm instead of 600, I can correct this deficiency.
Cross-Pollination Maybe?
I wonder if Canon would consider this project an interesting way to demonstrate the capabilities of the new 5D MII?
X www.thepicturetaker.ca
0
BaldyRegistered Users, Super ModeratorsPosts: 2,853moderator
edited September 19, 2008
Good question about Canon. The 5D MKII does seem like the camera for this. I've loved shooting with the D3 but 21 megapixels... Also we have a Canon 600 f/4 prime, which seems to be doing the trick.
That was shot at ISO 200, f/4.5, 1.3 seconds, mirror up, IS off, autofocus. The only thing I could wish for is a faster shutter to freeze motion better.
BaldyRegistered Users, Super ModeratorsPosts: 2,853moderator
edited September 19, 2008
So I have an idea for shooting a dark (on the first part of the Bay Bridge) to early pre-dawn in the financial district, to first light on the Golden Gate with this shot. Tell me if you think it would work:
I know the time diff between darkness and first rays.
I know how many frames it takes to cover the pano.
So I time it. I shoot the first two frames, one high and one low, at say 5:30. There are about 90 frames across. I shoot another every 45 seconds roughly to get me to the Golden Gate at the right time.
Sounds like a good experiment, I would increase your overlap to give you more coverage to blend. Say 120 frames across and every 30 sec?
I agree. In the few panoramics I've stitched together even moderate changes in cloud density show in the final pano, even tho I didn't even notice the difference during the shoot. More overlap will help to smooth the differential.
More overlap will also help with lens vignetting.
At the sizes you intend to print, I think you would notice with too little overlap.
With the changing light you'll have to adjust exposure, by changing the shutter speed I assume or lowering the ISO. This would give you faster shutter speeds (as you wanted!) for the latter half. Also, temperatures will vary slightly during this whole process, but will it be enough to matter? Truly epic project, I hope this works!
Also, take a couple photos of the setup you're using. When it's printed on the wall and people are drooling over it, it would be fun to show the "making-of" this historic print.
The other way would be to shoot the whole panorama 2 or 3 times as the light changes, run the alignment on all the pictures together so they are all in the same space, then run the merge on each set separately.
That will give you 3 panos that should line up perfectly that you can mask and blend to your hearts content
Being unfamiliar with the area a I need a bit of help with the orientation here Baldy. Are you shooting Northward, panning East to West? -- Right to Left in the photo -- thus putting the early morning light on the side of the bridge facing the camera and the night-time portion to the right in the photo?
The other way would be to shoot the whole panorama 2 or 3 times as the light changes, run the alignment on all the pictures together so they are all in the same space, then run the merge on each set separately.
That will give you 3 panos that should line up perfectly that you can mask and blend to your hearts content
Also a great idea, much more control over the finished product!
As for location, I'm going to guess the tip of Treasure Island shooting from left to right, south to west.
Complete novice, but figure I will say I am enjoying reading the thread. I was thinking about your thermals issue, and I wonder if a higher vantage point farther away from the water would be better. Since there is less humidity directly radiating from the water as you get higher up would that reduce the amount of thermal effects you see. Like I said, pure conjecture so I could be wrong.
Also I am not sure that they are all thermal issues, some could simply be wind swirling and getting the reflection from some of the particles in the air. For example with Pathfinder's shuttle launch experience, my personal hypothesis is that one gets the thermals off the water as well as the pressure wave pushing particles in the air around (and to some degree at the viewer) from the shuttle launch even at 9 miles away.
In terms of steadying the camera, have you thought about a tent to keep it out of the wind?
The other way would be to shoot the whole panorama 2 or 3 times as the light changes, run the alignment on all the pictures together so they are all in the same space, then run the merge on each set separately.
That will give you 3 panos that should line up perfectly that you can mask and blend to your hearts content
and yes, I too pity your computer.
So that's the way I shot the one you saw me post a day or two ago. I got 5 panos and posted the middle one. Two are lighter and two darker.
But, in order to get it in a resonable amount of time, I have to shoot one row with a 400mm lens, instead of two rows with a 600 or maybe even 800, which cuts the resolution.
We are working on blending the 5 exposures now but if you pity my computer trying to handle one layer, imagine five images layered... We're having to break it in segments.
And. If I shoot it once, 180 frames with the 600, I may have time to bracket the exposure 2 stops each way per frame. Then HDR the 180 frames together and do the adjustments before doing the stitch.
I dunno how to open an image with more than 30,000 pixels in a dimension in Photoshop. I've heard the 64-bit version of cs4 beta for Windows might be able to open it, so maybe there's hope. The machine would probably need 64 gigs of ram. Probably have to hijack one of our servers from the datacenter for awhile before it gets put in the datacenter.
I don't know if you can do this in autoPano Pro, but my technique is to give the bracketed images to the stitcher (ptgui pro) and get back a 32bit .tif or .psb (photoshop large format) which has the entire dynamic range in it, (though you can also have ptgui spit out each layer separately into the photoshop file or to tif's) it's easier to work with a high dynamic range image then multiple layers, but if you're shooting under different lighting conditions you'll have to use them as separate layers.
You can tone the image using the tone mapping in photoshop, it pops up the dialogue when you convert it from 32bit to 16 bit. I've also had good results with the exposure fusion module in ptGui. Or the real hard graft is to dodge and burn by hand painting layers with multiply & screen blending modes.
It doesn't take that much of a machine to do this, I've stitched 33k wide pano's on my macbook pro (2.2ghz, 3 gig ram) and worked on them in photoshop with 4 or 5 layers. Though these days I tend to use my spare work machine running 64bit windows, (dual quad core xeon, 6 gig ram) it'll stitch a big pano in 4 or 5 hours rather then 16. Ram helps but fast disks make a big difference, 100gig+ of caching is fairly common.
Baldy, do you offset your camera so that your pivot point is below the first lens? Seems to me that if the sensor is above the pivot point the first lens is in a different space thus a different perspective. I have only tried to create a pano-shot once and noticed the pictures didn't line up easily even with Canons "stitch assist".
0
BaldyRegistered Users, Super ModeratorsPosts: 2,853moderator
I was thinking about your thermals issue, and I wonder if a higher vantage point farther away from the water would be better.
Also I am not sure that they are all thermal issues, some could simply be wind swirling and getting the reflection from some of the particles in the air.
In terms of steadying the camera, have you thought about a tent to keep it out of the wind?
All good points. On this last attempt I found a higher vantage point on the island out of the wind that got me maybe 200 feet above the water, and got the sharpest results to date.
Comments
Have you told the owners about the project and showed them the tests? They might come on board and light special hours if they knew how it would look?
(Is it run by the San Francisco Recreation & Park Department?)
The phone number to the tower appears to be (415) 362-0808.
Moderator of the Cameras and Accessories forums
I await the outcome
Now that is a lot of pixels, and a lot of gigabytes:D
Moderator of the Technique Forum and Finishing School on Dgrin
w00p now we're talkin
Portfolio • Workshops • Facebook • Twitter
It's been fun watching your process evolve, can't wait to see the end result.
You can hardly see the clock tower in the low-res pano I posted, above. But 4 gigapixels gets you chicks. I mean ticks. Ticks on the clock.
Maybe if I shoot this two rows with an 800mm instead of 600, I can correct this deficiency.
What was your budget again?
http://hobday.net/fm/n1200-1700/sn200015/slides/_MG_8669.jpg
Moderator of the Cameras and Accessories forums
Speaking of big stitches, have you seen this?
http://www.newyorker.com/archive/2005/04/11/050411fa_fact?currentPage=1
I wonder if Canon would consider this project an interesting way to demonstrate the capabilities of the new 5D MII?
That was shot at ISO 200, f/4.5, 1.3 seconds, mirror up, IS off, autofocus. The only thing I could wish for is a faster shutter to freeze motion better.
I know the time diff between darkness and first rays.
I know how many frames it takes to cover the pano.
So I time it. I shoot the first two frames, one high and one low, at say 5:30. There are about 90 frames across. I shoot another every 45 seconds roughly to get me to the Golden Gate at the right time.
Then stitch and blend.
What am I missing?
I agree. In the few panoramics I've stitched together even moderate changes in cloud density show in the final pano, even tho I didn't even notice the difference during the shoot. More overlap will help to smooth the differential.
More overlap will also help with lens vignetting.
At the sizes you intend to print, I think you would notice with too little overlap.
I pity your computer.
Moderator of the Cameras and Accessories forums
Also, take a couple photos of the setup you're using. When it's printed on the wall and people are drooling over it, it would be fun to show the "making-of" this historic print.
-J
That will give you 3 panos that should line up perfectly that you can mask and blend to your hearts content
and yes, I too pity your computer.
As for location, I'm going to guess the tip of Treasure Island shooting from left to right, south to west.
-J
Also I am not sure that they are all thermal issues, some could simply be wind swirling and getting the reflection from some of the particles in the air. For example with Pathfinder's shuttle launch experience, my personal hypothesis is that one gets the thermals off the water as well as the pressure wave pushing particles in the air around (and to some degree at the viewer) from the shuttle launch even at 9 miles away.
In terms of steadying the camera, have you thought about a tent to keep it out of the wind?
Pictures | Website | Blog | Twitter | Contact
But, in order to get it in a resonable amount of time, I have to shoot one row with a 400mm lens, instead of two rows with a 600 or maybe even 800, which cuts the resolution.
We are working on blending the 5 exposures now but if you pity my computer trying to handle one layer, imagine five images layered... We're having to break it in segments.
And. If I shoot it once, 180 frames with the 600, I may have time to bracket the exposure 2 stops each way per frame. Then HDR the 180 frames together and do the adjustments before doing the stitch.
I dunno how to open an image with more than 30,000 pixels in a dimension in Photoshop. I've heard the 64-bit version of cs4 beta for Windows might be able to open it, so maybe there's hope. The machine would probably need 64 gigs of ram. Probably have to hijack one of our servers from the datacenter for awhile before it gets put in the datacenter.
You can tone the image using the tone mapping in photoshop, it pops up the dialogue when you convert it from 32bit to 16 bit. I've also had good results with the exposure fusion module in ptGui. Or the real hard graft is to dodge and burn by hand painting layers with multiply & screen blending modes.
It doesn't take that much of a machine to do this, I've stitched 33k wide pano's on my macbook pro (2.2ghz, 3 gig ram) and worked on them in photoshop with 4 or 5 layers. Though these days I tend to use my spare work machine running 64bit windows, (dual quad core xeon, 6 gig ram) it'll stitch a big pano in 4 or 5 hours rather then 16. Ram helps but fast disks make a big difference, 100gig+ of caching is fairly common.
I am out in San Francisco on business and when I flew in I thought about the pano and just kind of wondered.
Pictures | Website | Blog | Twitter | Contact
It's gonna be a great couple of days with the weather should come clouds.