No succes stitch spherical pano from Meike 6.5 mm circular fisheye.

[Henk Tijdink]

Dear readers, 

I've bought a Meike circular fisheyelens 6.5 mm f2 for a MFT camera. and want to use it for making full spherical panorama's.

You get than partly a circle  On the long side of the sensor it is shaved off. FOV there ca 120 degrees.

Making 4 pictures at 90 degrees turn and in the demo of PTGUI it stitches perfectly.

In Hugin however I can't get it all right. 

It finds control points between the images. I did the circular crop in the mask tab, but optimizing gives a very bad result.

In PTGUI is the lens in te database.

It gives the following values for the Meike lens. This for the Jpeg files from the camera

F  theoretic          6,5 mm                     After optimizing are the lens parameters

Practical              6,66 mm                   a= -0,093           and sensor shift.

Fov degrees         181,01                       b=  0,18              d= -0,18%

Fisheye factor    -0,824152                 c= -0,11               e= -0,17%

What do I wrong in Hugin that I can't get a good stitch, and in the demo of PTGUI it works flawless.

Is it possible to "translate" the lens parameters of this from the PTGUI database to hugin? 

I send the PTO file after optimizing . The values of control point distaces are then very large.

Kind regards, Henk

[Greg 'groggy' Lehey]

On Thursday, 1 October 2020 at 7:33:02 -0700, Henk Tijdink wrote:
> Dear readers,
> I've bought a Meike circular fisheyelens 6.5 mm f2 for a MFT camera. and
> want to use it for making full spherical panorama's.
> You get than partly a circle On the long side of the sensor it is shaved
> off. FOV there ca 120 degrees.
> Making 4 pictures at 90 degrees turn and in the demo of PTGUI it stitches
> perfectly.
> In Hugin however I can't get it all right.
> It finds control points between the images. I did the circular crop in the
> mask tab, but optimizing gives a very bad result.

I do this kind of panorama all the time using an Olympus 8 mm f/1.8
fisheye, also with 4 images, and it works fine. If you can make the
images available online (NOT on this list!), I'll take a look at them,
at least in part because I was rather interested in this lens.

Greg
--
Sent from my desktop computer.
Finger groo...@gmail.com for PGP public key.
See complete headers for address and phone numbers.
This message is digitally signed. If your Microsoft mail program
reports problems, please read http://lemis.com/broken-MUA

[Henk Tijdink]

I have send the pictures with Wetransfer. Further I found the Least parallaxpoint from the MEIKE lens with my modified Panosaurus and found that the least parallaxpoint or pivot point is at the red line on the barrel when you make 3 pictures for 360 degees pano.

Then camera in horizontal position. Then the camera in vertical position and taking 4 pictures for a full spherical pano then the  pivot point is 4 mm behind the red line. 

From lens axis and turning 6o degrees both sides then the red line is the pivot point.

From lens axis and turning 45 degrees both sides then the pivot poin is 4 mm behind the red line.

From lens axis and turning 70 degrees both sides then the pivot poin is 2 mm before the red line  

From lens axis and turning 80 degrees both sides then the pivot poin is 4 mm before the red line  

From lens axis and turning 90 degrees both sides then the pivot poin is 6 mm before the red line  

The Meike is a circular fisheye lens with a FOV of 190 degrees, so it should be possible to make a full spherical pano with 2 pictures when the circle fits the sensor, but that is not with a FT  camera.

Because with this lens now I will make single row full spherical pano's. So my setup has changed to a L-grip a rail and a tripod. A picture of the new setup attached.

Op vrijdag 2 oktober 2020 om 08:12:17 UTC+2 schreef Groogle:

[T. Modes]

Hi Henk,

Am Donnerstag, 1. Oktober 2020 16:33:02 UTC+2 schrieb Henk Tijdink:

In PTGUI is the lens in te database.

It gives the following values for the Meike lens. This for the Jpeg files from the camera

F  theoretic          6,5 mm                     After optimizing are the lens parameters

Practical              6,66 mm                   a= -0,093           and sensor shift.

Fov degrees         181,01                       b=  0,18              d= -0,18%

Fisheye factor    -0,824152                 c= -0,11               e= -0,17%

 

Is it possible to "translate" the lens parameters of this from the PTGUI database to hugin? 

This is not so easy. PTGUI fisheye factor does not translate 1:1 to a projection of libpano. The fisheye factor of -0.824 would result in something between orthographic and equisolid projection (see https://wiki.panotools.org/Fisheye_Projection).

The next pitfall is the fov: PTGUI calculates the fov to the cropped circle, libpano/Hugin is using the horizontal dimension of the full image (see https://wiki.panotools.org/Panotools_internals#Cropping ).

I had a look at your pto file. If I remove the control points in the nadir area it optimizes nicely. Without the images it is difficult to give further advices, but I would check the control points in the nadir area (maybe some repetitive pattern has confused the control point finder) and then try again.

Thomas

[Greg 'groggy' Lehey]

On Friday, 2 October 2020 at 6:50:35 -0700, Henk Tijdink wrote:
> I have send the pictures with Wetransfer.

OK, I have good news and not-so-good news.

First the good news: the panos stitched out of the box with no
problems. First I had to tell Hugin about the lens: circular fisheye,
190° FoV. You can see the results at
http://www.lemis.com/grog/photos/Photos.php?dirdate=20201002.

The not-so-good news: the fast panorama preview makes a mess of the
display, and may be the reason that you thought it didn't work. But
the normal preview (the one nobody uses) shows the display correctly.
This is probably a bug that needs to be corrected.

[Henk Tijdink]

Hello Thomas

Removing the control points in the Nadir did the job.

But when I see the result in the preview screen you think it goes wrong. You see a lot of sawteeth in the preview. However the stitch is perfect. Further Photometric optimisation doesn't work too for the fisheye lens. I get the message that there is no overlapping area. Are that bugs or is this normal behavior with fisheye lenses with a cropped area? Attached are a screenprint of the preview and the resulting stitch
Op vrijdag 2 oktober 2020 16:49:36 UTC+2 schreef T. Modes:

[Greg 'groggy' Lehey]

On Monday, 5 October 2020 at 14:21:34 -0700, Henk Tijdink wrote:
> Hello Thomas
>
> Removing the control points in the Nadir did the job.

Did you read my reply? For me it worked out of the box, and produced
better results than you attach. See
http://www.lemis.com/grog/photos/Photos.php?dirdate=20201002

> But when I see the result in the preview screen you think it goes wrong.
> You see a lot of sawteeth in the preview.

Yes, as I said. It's an issue with the *fast* preview screen. The
normal preview works.