MFT Particularly Good for Panoramic and Mosaic Stitches

[fike]

I have been really pleased to discover that my pano work is much easier with MFT than with the full-sized APS-C DSLR that I was using.

Let me first start off by saying that I am in the camp that says that it is a good idea to use a calibrated spherical tripod head when stitching subjects with complex foreground. I am a woodland photographer, so that means I very frequently use my calibrated tripod head. 

With larger lenses and larger-framed bodies the pano head calibration step is important.  It needs to be done for each lens, and for zoom lenses, at several intermediate focal lengths.  You also need a large enough bar that allows the camera to move far enough aftward to rotate around the entrance pupil location.  With a big lens like a Canon 24-70, that location can move quite a bit.

On the other hand, I have found with the smaller bodies and lenses of MFT, you can eyeball it to the middle of the lens and get, for all intents and purposes, perfect alignment (at least as perfect as I ever got).  Furthermore, you can use a much smaller bar because you don't need to move as far aft.

While bigger is better in many domains, lots of stuff gets easier when the camera system gets smaller.

[BernardLanguillier]

I have been really pleased to discover that my pano work is much easier with MFT than with the full-sized APS-C DSLR that I was using.

Let me first start off by saying that I am in the camp that says that it is a good idea to use a calibrated spherical tripod head when stitching subjects with complex foreground. I am a woodland photographer, so that means I very frequently use my calibrated tripod head. 

With larger lenses and larger-framed bodies the pano head calibration step is important.  It needs to be done for each lens, and for zoom lenses, at several intermediate focal lengths.  You also need a large enough bar that allows the camera to move far enough aftward to rotate around the entrance pupil location.  With a big lens like a Canon 24-70, that location can move quite a bit.

On the other hand, I have found with the smaller bodies and lenses of MFT, you can eyeball it to the middle of the lens and get, for all intents and purposes, perfect alignment (at least as perfect as I ever got).  Furthermore, you can use a much smaller bar because you don't need to move as far aft.

While bigger is better in many domains, lots of stuff gets easier when the camera system gets smaller.

Yes, but:
- 43 sensors are not yet at the right level in terms of DR,
- Depending on your needs, you may need more frames since sensors resolutions are typically lower,
- That is more true if you are looking for panos with more DoF - typically landscape.

Cheers,
Bernard

[theguywitha645d]

m4/3 does a great job. Plenty of DR and resolution for the job.

[Borealis]

yes, m4/3 does a great job and has hmmmmm... some DR  and good resolution.... depending on your technique/software etc.. DR may be the most important issue since often times one area of the image may be much brighter than others... I've got FF and M4/3 systems.... and if at all possible I use FF or a Nikon D800 for that job ;-)
William

[BernardLanguillier]

m4/3 does a great job. Plenty of DR and resolution for the job.

DR needs for panos tend to be greater since you often cover a greater range of illuminations in scenes.

But I guess you'll soon be able to retire your 645D if 4/3 cameras meet your DR needs. Panos will be able to achieve the same level of quality at a fraction of the cost, weight and bulk.

Cheers,
Bernard

[OldRoy]

Well, I'm not in the same league as some of the august contributors here but it's news to me that the E M-5 (aka OMD, for the present), which I own and use, suffers from notably inferior DR (assuming that DR means what DxO assign to the term.)
Nor have I seen many people complaining about this property of the camera, quite the opposite. I also own a D700, hence the comparison.

OMD: 12.3 EVs
D700: 12.2 EVs
D4: 13.1 EVs

However I imagine nobody on LL these days uses anything but the D800: 14.4 EVs... groan.

As an aside, guessing the NPP of M4/3 lenses on a spherical pano head may work ok for cylindrical or rectilinear panos (you can often get away with HH after all) but for full 360x180 sphericals you'de be advised to set it up accurately! Not that anyone hereabouts is likely to be shooting VR panos, I imagine.
Roy

[fike]

I wasn't intending to debate the known attributes of MFT as compared to FF.  There are lots of discussions about that.  Yes, FF has better dynamic range.  This has been beaten to death.  Opinions vary about the criticality of the difference. I won't try to say they are equal, but I will say that the tradeoff is one I willingly accept. This is a new forum for compact system cameras.  I am presuming others discussing these cameras have made the same willing tradeoff. 

As for resolution that is a non-issue.  If you are shooting panoramic and want even higher resolution, shoot more frames.  I see no disadvantage to shooting a 4x2 instead of a 3x2 or 2x2.  If I am going to stitch, it really isn't an issue to me to shoot a few more frames.

Yes, usually with panos you are trying to maximize depth of field.  If you really want that creamy bokeh everywhere then shoot FF.  Again, I hadn't intended to debate the FF versus MFT thing all over again.

And finally one comment on dynamic range.  I have had very good luck with varying the exposure slightly across the span of a panoramic.  Sometimes I will bracket and then pick the exposures I want to use on the PC afterward.  Other times I will just shoot one exposure for the foreground and a slightly different one for the sky and background.  With enough overlap and smart processing in ACR, the viewer never needs to know that the foreground was -1EV and the background was +0.5 EV.  This technique is one great advantage of panoramic stitching.

[fike]

...

As an aside, guessing the NPP of M4/3 lenses on a spherical pano head may work ok for cylindrical or rectilinear panos (you can often get away with HH after all) but for full 360x180 sphericals you'de be advised to set it up accurately! Not that anyone hereabouts is likely to be shooting VR panos, I imagine.
Roy

This is the point of my posting. In the past, I assiduously calibrated my APS-C Canon 7D with each different lens.  If I didn't calibrate, I would find misalignment in branches and other foreground elements.  In my recent experience, when estimating the entrance pupil with a small lens like the 12mm f/2.0 the error is likely to be at worst case 1cm.  This amount of difference is generally inconsequential, so I haven't seen the same misalignments I would have in a larger lens and body combination.

[BJL]

Yes, but:
- 43 sensors are not yet at the right level in terms of DR,

To Bernard:
Could you describe what you mean by "the right level in terms of DR", perhaps by listing current cameras that do meet your standards?

To all:
I have some other questions about stitching. How feasible is it to deal with scenes of high subject brightness range [SBR] by
a) "exposure stacking" (a.k.a. HDR) with several frames at each "patch" of the scene
b) varying exposure level between different patches of the scene, and then adjusting levels before stitching?


P. S. I am a total stitching beginner: the only stitching I have done is handheld, using the panorama mode of the E-M5 and with the stitching then done semi-automatically by the Olympus Viewer software that comes with the camera. Sometimes the results are fine for web display; other times they are, well, "amusing" or "artistic".

[thierrylegros396]

Sony automatic Pano system works very well, because it uses only a small portion of each frame and is able to align them almost perfectly.

Sony HDR also works well, much better than Canon one which doesn't register the photos !

Tip with Sony Pano, use it vertically and scan horizontally to have more height in your final picture.

Have a Nice Day.

Thierry

[fike]

To Bernard:
Could you describe what you mean by "the right level in terms of DR", perhaps by listing current cameras that do meet your standards?

To all:
I have some other questions about stitching. How feasible is it to deal with scenes of high subject brightness range [SBR] by
a) "exposure stacking" (a.k.a. HDR) with several frames at each "patch" of the scene
b) varying exposure level between different patches of the scene, and then adjusting levels before stitching?
...

Absolutely, see my post two above yours.  They must have passed in the night.

[Bart_van_der_Wolf]

To all:
I have some other questions about stitching. How feasible is it to deal with scenes of high subject brightness range [SBR] by
a) "exposure stacking" (a.k.a. HDR) with several frames at each "patch" of the scene
b) varying exposure level between different patches of the scene, and then adjusting levels before stitching?

Hi,

I routinely do both, a) when the dynamic range of individual tiles is high, b) works best if you use a liberal amount of overlap (say 50%) and use a good blending engine to adjust brightness between tiles. I'm not so sure if simple stitching software can handle such scenarios well, but you can allways do the HDR/exposure blending first, and then hand the results over to the stitcher.

The 'Hugin' stitcher software allows to experiment for free.

Cheers,
Bart

[fike]

Here is a vertical pano stitch with differing exposures:
http://www.trailpixie.net/general/sneakoshot_alon.htm

[scooby70]

Yes, but:
- 43 sensors are not yet at the right level in terms of DR,

I think I'm right in saying that the best MFT sensor now has more DR than any Canon DSLR ever made.

[scooby70]

Here is a vertical pano stitch with differing exposures:

I'm new to pano, I did a 21 frame stitch from my G1 and was impressed but it never occurred to me to do a vertical stitch.

Thanks for the inspiration!

[Ken Bennett]

I think calibrating is still a good idea. I've found some significant differences between lenses and especially with the zoom I use (Panny 12-35.) Overall, though, I'm very happy with the GH2 and the 14mm lens for panos, using the smallest Nodal Ninja. A nice, compact, lightweight pano outfit -- heck, I can even use my smaller tripod.

[BernardLanguillier]

To Bernard:
Could you describe what you mean by "the right level in terms of DR", perhaps by listing current cameras that do meet your standards?

Don't get me wrong, I am not knocking MFT at all and yes, the OMD is probably in the same league as a 5DIII but that only shows how far behind Canon currently is.

I did stitch reasonnably succesfully with a D2x that certainly has less DR than an OMD.

If I can select what gear I can stitch with I'd rather use the D800 since it will be able to deal with a larger set of situations. I have been shooting more casually recently, using the D800 handheld with 2 primes (85mm f1.4 and Sigma 35mm f1.4 mostly) with no paralax issues also.

Anyway, yes, the OMD is definitely a good option for stitching.

Cheers,
Bernard

[Ed B]

Here is a vertical pano stitch with differing exposures:

I haven't tried stitching very much and I know it takes quite a bit of skill to do it right but I find the silhouetted trees at the top of the frame fighting the bottom of the frame that would "normally" be darker. The image to me comes off as fake. I'd really like to see it with a darker foreground.

Sorry for taking this OT.

[BJL]

@Bernard: thanks for the clarification.

@fike: yes I had seen your brief comment on my item (b), but wanting to get more discussion on such techniques. And thanks for the example.

@Ed B.: there seems to be a persistent problem when trying to produce a single image of a scene like that with high Subject Brightness Range: to me, almost all HDR image look a bit fake.  And that is mostly because they are! I think that the problem is that when we view real scenes like that, our eyes adjust to the different lighting levels as we scan different parts of the scene, and in particular our irises close down when looking at a brightly sunlit scene more than they will when looking at even pure white on a computer screen or a print. So prints and even on-screen images do not generate the same degree of brightness compensation by our visual system. (Aside: I find even more fake and annoying almost every attempt to reduce SBR in a nature scene by filling the shadows with flash!)

This is, I think, mainly an artistic challenge for the image processing: finding a compression of luminosity levels for various parts of the final image to simultaneously avoid being too pale in the well-lit areas and too dark in the shadow areas, while fooling our eyes and brains into accepting it as realistic. Actually, my visual system accepts fike's example: maybe I interpret the foreground as lit by light breaking through between the trees from an unclouded part of the sky.

[thierrylegros396]

Really well done , Fike !!

HDR without the drawbacks often found in lots of websites

Thierry