Skridlov
Posts: 158
Location: United Kingdom
Registered: 12 Nov 2007
|
Nikon 10.5 mm c/a question
Posted: 27 Jan 2008 at 20:20 GMT
updated: 27 Jan 2008 at 20:28 GMT
|
|
Hi everyone
After months of agonising about the expense, I have just bought the above. Never having owned anything similar before I'd like to know how to judge whether the ca is within acceptable limits for this lens. One reviewer I read said that out of three examples, he had one which exhibited completely unacceptable ca. Looking at large magnifications of the lens periphery is pretty alarming! Like an old time tube TV camera severely out of registration! I have tweaked the red/cyan in CS2 to some effect but I'm curious about what the average example of the lens should look like.
Also, could someone point me to a location where the general issue of Lens correction parameters is outlined (succinctly if possible). & Has anyone established any vignetting values for the above lens at the various apertures?
A big thanks to John Houghton for the Beginner's Guide to 360 VR stitching!
Skridlov aka Roy
I forgot to mention that the files I looked at in this connection were shot at about F4 and were (Nikon D200) Normal JPEGs. Is it the case that the ca can be corrected better in RAW?
|
|
alert moderator
|
|
John Houghton
Posts: 2317
Location: Hitchin, United Kingdom
Registered: 17 Jan 2005
|
Re: Nikon 10.5 mm c/a question
Posted: 27 Jan 2008 at 20:58 GMT
updated: 27 Jan 2008 at 21:01 GMT
|
|
Skridlov said: Is it the case that the ca can be corrected better in RAW?
Roy, Yes, correcting CA in RAW is far more successful than in jpeg. With ACR, I have used adjustments like -38 and +14 on the top and bottom sliders, respectively (at around f/8). Maybe that will tell you whether your copy of the lens differs markedly from mine, but do use a RAW file for comparison purposes.
John
|
|
alert moderator
|
|
enbilaman
Posts: 73
Location:
Registered: 3 Mar 2006
|
Re: Nikon 10.5 mm c/a question
Posted: 28 Jan 2008 at 9:36 GMT
|
|
While it is still possible to correct a bit of CA from JPG in the newer ACR version, I agree with John and would recommend to shoot RAW.
I have found that the type of light has a strong impact on the correction setting though. The two sliders positions have to be set accordingly.
From my measurements on a special target I have defined the following for my own lens/camera combo:
Sun light (-35, +33); Tungsten lamps (-35 +15); Fluorescent (-54 +10).
Note: My camera is a Full frame Canon EOS 5D.
End of the succinct answer(as required by Roy)
Michel
|
|
alert moderator
|
|
enbilaman
Posts: 73
Location:
Registered: 3 Mar 2006
|
Re: Nikon 10.5 mm c/a question
Posted: 28 Jan 2008 at 10:15 GMT
|
|
My previous answer above has to be amended for the panographers who use the (Shaved) Nikkor 10.5-mm (or BTW the Tokina 10-17 mm at small focal) on a full-frame DSLR camera.
From full-frame (i.e. rectangular image coverage), the fisheye lens shall then magically become a (cropped) circular fisheye lens. The full angle of view (about 185-190°) shall thus be recorded along the length of the sensor.
Mainly but not only with tungsten light, these two lenses exhibit a similar reversal of the CA radial mapping that impacts principally the outer part of the coverage : tinyurl.com/32jjdz
Let's recall that these lenses were designed exclusively for APS-C sized camera...
I own both of these lenses and I now know that it is in fact probably impossible to totally correct CA under certain shooting circumstances. Trade-off is then needed to balance the remaining lateral color on the edge of the circle with a more central area.
With some practice the output shall be very good unless one zooms to very high level and scrutinize specific regions of the sphere while viewing the panorama.
IMHO the result is fortunately superb anyhow and nobody should regret from having shaved a fisheye in spite of this minor annoyance.
Regards,
Michel
|
|
alert moderator
|
|
Skridlov
Posts: 158
Location: United Kingdom
Registered: 12 Nov 2007
|
Re: Nikon 10.5 mm c/a question
Posted: 28 Jan 2008 at 18:34 GMT
|
|
John, Michel, thanks.
I have just compared some simultaneous shots NEF/JPEG - albeit hastily snatched in a church underlit with a range of different light sources. It's clear even from this hasty test that the raw files are more successfully tweaked for ca in ACR than the jpegs.
Michel, re:
"Sun light (-35, +33); Tungsten lamps (-35 +15); Fluorescent (-54 +10)"
I take it that this refers to r/c and b/y consecutively for each case?
John - thanks to your guide (and my apologies for going off half-cocked on the feedback - I got very confused earlier with the out of date Wiki tutorial) I managed to stitch these shots together into a reasonable 360 - my first. I wouldn't show it to anyone but I am very gratified!
A quick question about this - is it acceptable to level the finished file using the t1, t2 cp's? Or should this be done at an earlier stage?
Roy
|
|
alert moderator
|
|
John Houghton
Posts: 2317
Location: Hitchin, United Kingdom
Registered: 17 Jan 2005
|
Re: Nikon 10.5 mm c/a question
Posted: 28 Jan 2008 at 19:39 GMT
|
|
Skridlov said: is it acceptable to level the finished file using the t1, t2 cp's? Or should this be done at an earlier stage?
Roy, The t1/t2 points can be added early in the process or just before stitching. The stitched file can also be levelled - in a separate project - but this involves an additional stage of remapping and interpolation. This is not likely to have a serious effect on image quality, but for simplicity, I suggest you level just before stitching; it only takes two minutes or so at the most.
John
|
|
alert moderator
|
|
Skridlov
Posts: 158
Location: United Kingdom
Registered: 12 Nov 2007
|
Re: Nikon 10.5 mm c/a question
Posted: 28 Jan 2008 at 21:19 GMT
updated: 28 Jan 2008 at 21:22 GMT
|
|
Hi again John
I have just added the vertical line cp's & re-optimised (I chose the most obvious image upon which to set these but have not reset any other parameters anywhere, which I suspect may be incorrect). One thing I notice is that whenever I select the Spherical Equiractangular 360X180 projection as soon as I re-check this setting it has reverted to Equiractangular, it never seems to latch.The ouput file looks OK to my inexperienced eyes though.
Roy
|
|
alert moderator
|
|
John Houghton
Posts: 2317
Location: Hitchin, United Kingdom
Registered: 17 Jan 2005
|
Re: Nikon 10.5 mm c/a question
Posted: 28 Jan 2008 at 21:38 GMT
|
|
Roy, After adding the t1 points, it is not necessary to optimize, but if you do, you need to alter the parameters to be optimized on the Optimize tab (Advanced mode). You need to check y,p,r on all images, except you can leave y unchecked on any of the horizontal shots (to stop the whole panorama shifting sideways). But instead of optimizing, you can just go to the Panorama Editor window and click on Edit->Level Panorama. In that case, you need not make any changes to the optimize parameters (unless you need to re-run the optimizer for any reason at a later time).
Note that you can put t1 points on vertical features in more than one image.
I'm not sure what you mean by the projection setting reverting to equirectangular, when it was already equirectangular. PTGui's Project Assistant is inclined to helpfully(!) crop the output image vertically when there are holes at the zenith and nadir. You may need to reset the fov to 360x180 before stitching. I never run the Project Assistant myself as I like to have full control over what is happening, i.e. I make use of Advanced mode. You will gradually learn how to do this yourself in time.
John
|
|
alert moderator
|
|
back to threads list |
this thread is closed