John Houghton's forum messages

back to John Houghton's profile

Search the forums:
messages 1-15 of 2025
first prev Prev 1 2 3 4 5 next next last
Forum/thread  Message 
Forum:
Q & A

Thread:
New lens ?
Re: New lens ?
Posted: 8 hours ago
go to thread
Sample images for download are available here:

www.360precision.com/360/rd/fisheye

John
alert moderator
Forum:
Q & A

Thread:
Newbie needs help !!!
Re: Newbie needs help !!!
Posted: 20 hours ago
updated: 20 hours ago
go to thread
David, To understand what arrangements of images are possible to cover the entire 360x180 degree view, you can use the Panorama Editor window in PTGui. All you need to do is add several of your camera images (containing anything) and then lie to PTGui about what focal length lens was used. You might say that they are 20mm rectilinear with a 1.6 crop factor, for example.

In the Panorama Editor window, you select the equirectangular 360x180 view, which shows a flat representation of the full spherical stitching surface. The images will be initially stacked up like a pack of cards in the central position (pitch=yaw=0). You can then drag individual images about to see how they can be fitted together, like this:



The two top left hand buttons enable you to select individual images for dragging. The third button selects the entire composite panorama image. The warped images assume weird shapes according to where they are positioned. In this example, there are just 4 images - part of the 12+12+12+1 arrangement. Predicting how images will overlap is difficult to calculate, but very simple this way. You can read out the yaw, pitch and roll angles on the Image Parameters tab (in advanced mode). It's a great help in understanding what's going on.

John
alert moderator
Forum:
Q & A

Thread:
Panos in restricted space
Re: Panos in restricted space
Posted: 23 hours ago
updated: 23 hours ago
go to thread
Roy, Out of interest, I tried a considerably offset nadir to fill the large nadir hole left by a horizontal row of six 10mm fisheye images taken at pitch=0. The normal optimization of the nadir into position (with all the other images fixed) gave a maximum cp distance of 474. Turning on the viewpoint option then gave a good optimization with a maximum cp distance of 3.8.

I patched the nadir in with PTEditor. It needed a very tiny adjustment with the transform/distort tool at two corners to bring it into perfect alignment all around. It should be easier if the horizontals are tilted down by 10 degrees to make the hole smaller.

John
alert moderator
Forum:
Q & A

Thread:
Panos in restricted space
Re: Panos in restricted space
Posted: 17 May 2008 at 14:07 GMT
updated: 17 May 2008 at 14:13 GMT
go to thread

Skridlov said:

can one expect to get a decent stitch using a hand-held nadir alone? It doesn't seem so to me.

Provided the shot covers the nadir hole adequately, then it should be possible to get a good stitch. I haven't tried the viewpoint feature with a really large hole. A nadir shot taken with the camera on the tripod provides many more features accurately positioned for control points with the handheld nadir.

John
alert moderator
Forum:
Q & A

Thread:
Panos in restricted space
Re: Panos in restricted space
Posted: 16 May 2008 at 14:06 GMT
go to thread

Skridlov said:

I have had no success whatever with the method of using 2 X 180 deg -90 shots off the tripod

Roy, I don't think anyone suggested taking 2x180 -90 shots off the tripod. It only makes sense to do this on the tripod. It just gives you a good quality view of the nadir region with the minimum area to patch out (the tripod). Because the shots are taken on the tripod, they will stitch in alongside the other shots because they all share exactly the same viewpoint. The nadir images need no preferential treatment, other than the addition of an alpha channel mask to hide the tripod and pano head hardware (and possibly parts of your own body).

If it's actually possible to "lock" the 7 shots so that the nadir can be tweaked, then I have no idea how to do this.

When optimizing, the optimizer is only able to change parameters that are checked. Not checking them effectively locks them. So if you uncheck the y,p,r boxes of all images except the nadir, say, then the optimizer will only be able to move the nadir image about. The remaining images will stay put. Likewise, if you uncheck the lens parameters, they will not be changed during optimization. You have complete control over what is going on. (This is in the advanced mode, of course).

The viewpoint correction method I haven't tried because I imagine it to be difficult - because it seems to me to be little more than a variant of hand-held. Ie the relatively uncontrollable position of the shot results in a viewpoint which is "out" in all three axes. Isn't this so?

Well yes - but correcting for the offset position of the camera is precisely what the viewpoint feature is for! This is a straightforward image transformation in the case of a flat floor. Only put control points on the flat area. Optimize the nadir y,p,r only, initially. This will give a poor optimization result, but the nadir will be shifted into approximately the right position. Then optimize the nadir additionally with the viewpoint option checked. That introduces the extra transformation necessary to fine tune the the alignment and get a good optimization report.

If you supply all the images (including an offset nadir) I will gladly create a project file to show how it's done.

John
alert moderator
Forum:
Q & A

Thread:
Blendind problems
Re: Blendind problems
Posted: 15 May 2008 at 11:43 GMT
updated: 15 May 2008 at 11:45 GMT
go to thread
When you say you "used masks to to remove the Panosaurus head", do you mean you added alpha channel masks to the input images? That ought to avoid smudgy problems. You could try a different blender. If that doesn't work, you could generate a full equirectangular image containing the nadir image only. Extract the rectilinear view with Superrune to use as a patch for the nadir instead of cloning.

John
alert moderator
Forum:
Q & A

Thread:
HDR vs EV results in PTGui / Lens hood on Nikkor 10.5
Re: HDR vs EV results in PTGui / Lens hood on Nikkor 10.5
Posted: 15 May 2008 at 7:26 GMT
go to thread

Hamish Tear said:

how do I use MLU on a Nikon D300 set at continuous shooting for bracketing?
The mode dial only allows one or the other. Is this the same on Canons?

Hamish, On my Canon 40D, I can switch to live view + continuous shooting + bracketing. In live view, the mirror stays up. The D300 has live view, but I don't know if the same technique is possible.

John
alert moderator
Forum:
Q & A

Thread:
Canon EOS 40D & Fisheye 8mm f/3.5 EX DG Users
Re: Canon EOS 40D & Fisheye 8mm f/3.5 EX DG Users
Posted: 14 May 2008 at 12:34 GMT
go to thread

mediavets said:

There are both zenitn and nadir holes when I stitched this set with Autopano Pro. Which is what I would have expected using Canon cropped sensor and Sigma 8mm.

Andrew, That's not what should be expected. With the camera slanted to bring the diagonal more-or-less vertical, the vertical angle of view is not cropped, so you get a full 180 degrees. Hence no holes. PTGui gets it right.

Mike, you should pay more attention to control point placement and delete the points on the clouds and waving foliage. A much improved optimization is then obtained.

John
alert moderator
Forum:
Tips & Tricks

Thread:
PTGui Viewpoint optimization
Re: PTGui Viewpoint optimization
Posted: 14 May 2008 at 7:43 GMT
go to thread
Hamish, I've seen PTGui (and the Panorama Tools optimizer) produce all sorts of bizarre results. Most commonly this happens when starting from scratch, with the images all assigned the same y,p,r values by default. Generally, you have a pretty good idea of what the values should be (and also the lens parameters) from previous projects. To me it seems perverse to deliberately withhold all this useful information from PTGui and then fret and fume when the optimizer fails to deliver the goods. I always either apply a template to initialize the project with sane values, or quickly enter rough yaw and pitch values on the image parameters tab, making use of Fill Yaw and updating values by column selections. When this is done, I find bizarre results are very rare.

John
alert moderator
Forum:
Q & A

Thread:
HDR vs EV results in PTGui / Lens hood on Nikkor 10.5
Re: HDR vs EV results in PTGui / Lens hood on Nikkor 10.5
Posted: 13 May 2008 at 6:43 GMT
go to thread
The shaved 10.5mm Nikkor on a fullframe camera does not quite give a full circular image. It has the sides cut off, rather like the Sigma 8mm on an APS-C sensor. The result is that the hfov in portrait orientation is 150 degrees approx. The Sigma 8mm gives a true full circular image.

John
alert moderator
Forum:
Q & A

Thread:
Help with Equipment
Re: Help with Equipment
Posted: 12 May 2008 at 20:13 GMT
go to thread
Prices from Hong Kong sellers on Ebay can offer considerable savings over UK prices, and some guarantee to reimburse any taxes that might be levied on the import. I have used UR Galaxy four times without any problems and haven't yet been asked to pay taxes on any of them. Delivery is around 1 week. Typical prices (inc shipping):

Sigma 10mm £340
Tokina 10-17mm £280
Sigma 8mm £330

John
alert moderator
Forum:
Q & A

Thread:
Converting QTVR to jpeg
Re: Converting QTVR to jpeg
Posted: 12 May 2008 at 10:15 GMT
go to thread

Magyar said:

... or can I convert 6 cubes (Tiff jpegs) to "normal" panorama?

This can be done with Pano2QTVR (Windows), and you can also convert them with any of the Panorama Tools based stitchers, such as PTGui and Hugin.

John
alert moderator
Forum:
Tips & Tricks

Thread:
panorama instructions
Re: panorama instructions
Posted: 12 May 2008 at 7:22 GMT
go to thread

michael przewrocki said:

most clever nodal-point-finding-solution

Michael, This is hardly a "clever" solution; merely the standard one that people have been using for years. It is not even an adequate description in the case of fisheye lenses, since it fails to take account of the shift of the entrance pupil (not nodal point, of course) for off axial rays.

John
alert moderator
Forum:
Q & A

Thread:
Agnos RingT query...
Re: Agnos RingT query...
Posted: 12 May 2008 at 6:42 GMT
go to thread
Stu, The three38 ( tinyurl.com/3fytox ) is not a rotator. It is a fixed adapter that can be used to connect to the top of a pole, for example, where you would rotate the pole to rotate the camera.

John
alert moderator
Forum:
Q & A

Thread:
Problems stitching nadir and zenith
Re: Problems stitching nadir and zenith
Posted: 11 May 2008 at 17:09 GMT
updated: 11 May 2008 at 17:10 GMT
go to thread
Dave, A rectilinear nadir view shows better what's going on, but it looks like you have a circular saw effect, with teeth that cut in an anticlockwise direction. This means you need to move the vertical arm in towards the pivot by a small amount - a mm or two. If you go to far, the sawtooth will change cutting direction. If you have eliminated parallax, and the sawtooth is minimal, then you should get a good stitch. If you don't, then your stitching technique might be found wanting.

John
alert moderator
messages 1-15 of 2025
first prev Prev 1 2 3 4 5 next next last