FourPiSr
Posts: 107
Location: Stockholm, Sweden,
Registered: 17 Dec 2006
|
Explaining the optimizer
Posted: 25 Mar 2007 at 19:53 GMT
|
|
A while back I asked in a thread if there was any interest if I were to try explain how the optimizer works. Well, it took a bit more than just a couple of days to complete the article. Maybe you have given up on me but I've been writing on it a bit on and off since then. Here is the result: www.4pisr.se/panoguide/How%20the%20optimizer%20wo...
Feedback is appreciated. You can give your feedback either through this forum or by e-mail to the e-mail address in the article.
Lennart
|
|
alert moderator
|
|
DorinDXN
Posts: 2853
Location: Timisoara, Romania
Registered: 14 Nov 2006
|
Re: Explaining the optimizer
Posted: 25 Mar 2007 at 20:25 GMT
|
|
Hi Lennart, very useful explanation
I must admit that I'm bad in trig so please take this question as presumbly wrong but maybe you could add some information to correct me, you said there
---8<-----------------------------
In this example the images are 1024 pixels wide by 1536 pixels high. The horizontal field of view is 46.3 degrees. Each pixel then represents 46.3/1024 = 0.04521484375 degrees.
---8<-----------------------------
that should be some kind of average, I guess that the pixells from middle has different portion of field of view than those from sides.
Apollogise if I understand wrong
thanks for the work there
Dorin
|
|
alert moderator
|
|
eagle
Posts: 80
Location: Kona, Hawaii,
Registered: 16 Jan 2006
|
|
FourPiSr
Posts: 107
Location: Stockholm, Sweden,
Registered: 17 Dec 2006
|
Re: Explaining the optimizer
Posted: 26 Mar 2007 at 19:51 GMT
|
|
Dorin,
You're right the field of view per pixel varies. To start with the image is flat and the sphere is curved so that needs some correction which will be more the wider the lens is. Barrel distorsion causes the value to change even more especially for a circular fisheye.
I've simplified more things such as the calculation actually only is valid for the horizon. As you get closer to zenith and nadir the calculation changes. With an 18 mm lens you need three rows. If you have almost no overlap at the horizon and almost no overlap between rows, then the images in the top row overlap at zenith. Then the zenith point is on the middle (horizonatlly) and near the top of all the images in the top row. This has to be taken into account when calculating yaw and point distance.
Lennart
|
|
alert moderator
|
|
