From a Papywizard point of view, you can shoot in any direction, with the head in vertical position. Just point start/end corners, and go

From a APP point of view, I don't think this can lead to problems. It is like shooting a spherical pano, with only a little part of it. I guess.

Besides (optional) camera orientation data being absent or wrong in EXIF, this should work in both cases.

You will have to place the Center Point by yourself (recording its actual location using a buble level to be placed on the lens front part and using the corresponding shot as a reference could help) and you will have to adjust the panorama orientation around this point (roll)  by yourself.

Hopefully, the ceiling is adorned, many CPs will be availlable, no orphan will remain after the pano is detected...

;o) I remember having used adhesive tape (or you name it) to place newspapers on a room ceiling to test my first fisheye lens!

mediavets wrote:

I can't say I really understand why I got this result - my mind is incapable on envisaging quite what's going on. Nor do I understand why APP calculates the focal length to be 65.98mm.

The panorama FOV is rather low (around 50° - see Phi min/max and Theta min/max) so that you can use planar projection. This should result in straight lines on the ceiling being straight in the resulting pano.

If straight lines are still curved, try to restitch the pano after unselecting the lens distortion correction. The relative placement of source images (large overlap - image centers nearly at the same location) probably results in Autopano not being able to find the right focal length (the trap Autopano can't avoid could be: the higher the focal length the smaller the resulting pano + the smaller the resulting pano the better the RMS. First versions of Pano Tools were known to reduce the pano FOV to zero when no other way was found to get a decent RMS ;o) )

That PapyWizard decided to shot too many images in the zenith direction is the reason spherical panos need a particular kind of template (many map sheets are needed to describe the countries next to the equator but a single one is enough for teach of the Poles.)

mediavets wrote:

If I set the 0/0 conventionally with the head in vertical orientation then specifiy the pano using Start and End as opposite corners of a rectangular ceiling I just get a single row of shots with total yaw coverage of 180 degrees.

Strange... I'll investigate.

Here's a crude sketch of how I think you could perhaps mount a Merlin head horizontally on a tripod using a bracket - but it could be hard to get the centre of mass over the tripod screwmounting point and also to have the whole setup be stable and resistant to vibration.

Any ideas?

Last edited by mediavets (2009-01-29 16:31:56)

fma38 wrote:

Ok, I see the problem: here we need to make complete turn of the yaw axis for each pitch position... So, you have to point the starting yaw/pitch position. Then for the end position, just use pitch=90 (zenith), and make a complete turn of yaw axis (a little more for overlap). It should work.

I'll try it.

fma38 wrote:

But it is not the best solution, as you will get too much images near the zenith. In act, your first idea is the correct one: put the head horizontal...

Yes, but I think mounting a Merlin/camera/lens setup horizontally on a tripod is not so easy.

So I think the easiest and most practical approach to shooting panos of ceilings is probably to use a variant of a full spherical custom preset omitting the lower rows - what you might call an 'a-bit-less-than-hemi-spherical' custom preset?

Now I don't know why I ever thought to try and do it as a Mosaic mode shoot. I plead senility.

Last edited by mediavets (2009-01-29 17:01:39)

fma38 wrote:

Ok, I see the problem: here we need to make complete turn of the yaw axis for each pitch position... So, you have to point the starting yaw/pitch position. Then for the end position, just use pitch=90 (zenith), and make a complete turn of yaw axis (a little more for overlap). It should work.

But it is not the best solution, as you will get too much images near the zenith.

It does work.

Here's my result. Two images didn't get taken - I think the overhead lights may have overwhelmed my IR shutter release - could be problem outdoors in bright sunlight - it should have been 3 rows of 16 around, 48 images not 46.

As you say there are too many shots near the zenith - OTOH I was suprised that APP managed to link many of those shots near the zenith as thety were both relatively featureless and hideously underexposed. if these shots had been better spaced - for example as in a custom spherical preset - then I don't think APP would have created the links and I would have had orphaned images.

mediavets wrote:

I am just trying to get my head around why this would produce a better image that using a  just-about-hemispherical custom preset. In both cases the images shot would cover the scene.

Yes, it is absolutly the same. But you will need a lot of positions to compute in your preset if you use a 300mm lens That's why I said mosaic mode is better (easier, in fact).

You just need a piece of metal, (a 'L') to put the head in horizontal position: you don't need to adjust the NPP on the tripod axis; just make sure the gravity center is not out of the tripod footprint

fma38 wrote:

You just need a piece of metal, (a 'L') to put the head in horizontal position: you don't need to adjust the NPP on the tripod axis; just make sure the gravity center is not out of the tripod footprint

You make it sound so simple but I think it might take the system quite a while to stablise between shots if one just used a bent mild steel bar as a bracket?

The system would need to be pretty stable if shooting with a long lens in relatively poor lighting conditions as often found with ornate celings in historic buildings.

bigwade wrote:

mediavets wrote:

Not to mention the fact that I don't think any 300mm lens/camera combo could achieve an adequate positive pitch shooting position to shoot a hemi-spherical pano of the ceiling with the Merlin in conventional vertical orinenation.

You would be surprised about the NPP of a 300mm.
http://tinyurl.com/cmfgob

Hmmm.......so what do you find is the fore-aft setting (Merlin axis-camara body screw mount) for NPP with a 300mm lens on your Canon XXD body?