Being no expert, but fully remembering what it was like in the beginning, here
are added comments and changes to the previously suggested workflow that
perhaps are useful.
> 1) Start up Hugin, go to Expert Mode
(Hugin in Expert Mode shows options you often really need. These options might
be mentioned in tutorials without explaining that you would not see the option
in other than the Expert Mode. Having a mouse pointer other than a trackpad is
helpful.)
> 2) In the Photo tab, Add images. Lens type should show Rectilinear 18mm,
> Focal Length Multiplier 1.5
> 3) Under Feature Matching, select Hugin's CPfind, then Create Control Points
(Most Hugin functions operate on the images selected in the photos list or
operate on all the photos if none are selected. This is useful when selections
are intended but will stymie your intent when some images are accidentally
selected.)
3.2) Under Edit, pick Fine-tune all Points. This operation normalizes how well
the control points match to roughly a scale between 0 and 1 where the smaller
numbers are the worst matching points. The numbers serve as a matching rating.
In the next workflow step you will look at the ratings to judge how well the
matching went and you will delete the really bad control point matches or the
control point matches that should not be used. For example you don’t want to
match items that move, like clouds. Also, control point normalization values
below 0 or equal to 1 or larger than 1 indicate something very odd about the
control point.
3.4) Under View, pick Control point table. Click on the Correlation column
header to sort the list from low number to high number. Selecting any control
point reveals what control point pair in context with the photo image pair.
Delete low scoring control points at this time by judging their score against
the “Correlation Threshold” in Hugin”s Control Points Editor preferences.
Hugin’s default value for that is 0.8. You will need to make your own judgement
dependent on how well the images match.
> 4) Under Geometric, Run Optimize on Positions (incremental, starting from
> anchor)
(After this point the control point correlation normalization values between 0
to 1 seen after step 3.2 go away. The correlation numbers in the control point
table will now show the poor matches as having higher numbers than the better
matches. You will again look at the control points to toss out the worst
points.)
4.2) Again, Under View, pick Control point table. Click on the Correlation
column header to sort the list the other way from high number to low number. In
the images you posted you will quickly see at least two relatively higher value
zingers. Selecting them reveals their image context. You will see right away
these points are worthless and should be tossed. I would not try to fine tune
the points as described in step 11. Getting rid of these bad points early
removes their influence to subsequent alignment operations.
> 5) Repeat above but using, Positions, View, Barrel
> 6) Repeat above using, Everything without translation
> 7) In the Stitcher tab, Calculate Field of View and Calculate optimum size
> 8) Back to Photo tab, run optimise Everything without translation
> 9) Got to the Menu, View, Control Point Table
> 10) Sort by distance, descending. You'll see one biog outlier. Click on it.
> You should be taken to the Control Points Tab
(Step 10 repeats the added workflow step 4.2 and is useful for a final check.
As previously mentioned trying to fix useless bad matching points is not
advised but sometimes you might have to so. The control point list sorted by
correlation shows at a glance the match correlation distribution. Often you see
where the correlation jumps to higher values (higher is poorer). Those higher
correlation value control points are candidates for deletion at this time.)
> 11) Zoom to 100%. Move the CP with the large error so that they match a
> feature on both sides. Press Fine Tune and check that the CP is better
> aligned.
> 12) Back to Photos Tab, optimise Everything without translation, the maximum
> error should now be smaller. Repeat steps 10)-12) with any other CP with high
> distance until satisfied. You can close the Control Point Table.
(The following steps 13 through 16 I could not follow. In my experience Hugin’s
default settings for Photometric optimization work very well. I do not have
images needing anything other than the "Low dynamic range” optimization
calculation. Pay attention to which image is the image exposure anchor. This is
the image having “C” shown in the Anchor column in the photos list and is by
default set to the first image added to the list. Right clicking an image
allows one to make an anchor designation change.)
>> 13) On Photos tab, Photometric, Select Custom Parameters, but do not
>> optimise yet
>> 14) On Exposure tab, untick EMoR for the lens 1
>> 15) Tick the Er and Eb boxes for the 2nd and 3rd images
>> 16) Press Optimise Now
> 17) Press Preview Panorama, press Centre and Straighten. The panorama won't
> be matched in exposure.
("Preview Panorama” is original Hugin interface for this function. The “Fast
Preview panorama” interface was introduced to use computer graphics
capabilities that are now standard. In it are other functions you would
normally need that are not present in the old interface.)
> 18) Go back to Photos Tab
> 19) Right click the 2nd image name, select Lens, New Lens. Repeat for 3rd
> image. Now all image will have been assigned a different lens number (0,1,2)
> 20) Back on the Exposure Tab, press Optimise Now.
> 21) Back in the Preview Window, the exposure will now match a lot better
21) Using the “Fast Preview panorama” interface, select the Crop tab. Then
select Autocrop.
> 22) In the Stitcher tab, Calculate Field of View and Optimum Size, then press
> Stitch Now, give a file name.
--
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ
---
You received this message because you are subscribed to the Google Groups
"hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to hugin-ptx+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/hugin-ptx/1DCE261E-6FC6-4F8E-8A78-633154A23013%40gmail.com.
For more options, visit https://groups.google.com/d/optout.
