can't get fulla (or Hugin) to work for vignetting (and chromatic aber cor)

[Alister Ling]

Hi, I'm having some issues creating a blended pano - sorry for the long post. I'm using 2019.2.0.6c.... Log output gives me "enblend: excessive image overlap detected; too high risk of defective seam line", which I kinda understand... I have 5 post sunset sky/foreground shots which overlap about 80% (from solstice to solstice, 2 cover the left side, one in the middle, and 2 cover the right side). I can't really crop anything off the side edges to have "only the middle". But then I have no output.

There are two pairs of bracketed images which may be confusing things. i.e. I have 2 stacks in there. I can't seem to find the right stitch tab check boxes that will give me useful output. The output is really bad if I click calculate photometrics (in part because all the images were not taken with the same exposure settings?) - but I thought this was what Hugin could handle. 

Anyway, out of frustration, I decided on doing remapped images, no exposure correction (and no-crop unchecked) and that worked , so I can use enfuseGUI to blend them. Worked decently....BUT, I have vignetting that I would like to mitigate. So this is where fulla comes in! It would even handle chromatic aberration! Except I get, on Windows,  "Precondition violation! did not find a matching codec for the given file extension" no matter if I use jpg, jpeg, JPG, tif, tiff. I've just downloaded PTlens from sourceforge as the web page says to do, but can't do anything with it give the current error. Seems to me that even with a newer lens not in that older database, I could get the a,b,c,d from the new PT lens and feed the params into the fulla command line. 

I have not exhaustively looked for fulla tutorial examples on the net, but after 30 minutes have yet to find anything other than the manpage listing options, and no mention of codecs. I get the sense I'm so close, but missing something stupidly small.... so your help would be most welcome!

Thanks,

Alister.

[T. Modes]

Am Dienstag, 23. Juni 2020 03:56:03 UTC+2 schrieb Alister Ling:

There are two pairs of bracketed images which may be confusing things. i.e. I have 2 stacks in there. I can't seem to find the right stitch tab check boxes that will give me useful output. The output is really bad if I click calculate photometrics (in part because all the images were not taken with the same exposure settings?) - but I thought this was what Hugin could handle. 

Normally, Hugin can handle this. But in this case - where the images are mainly dark - it has too less information to calculate vignetting factors. But you can use another project with an higher dynamic range, store the vignetting parameters there in an ini file or in the lens database and load these parameters then in the other project.

Anyway, out of frustration, I decided on doing remapped images, no exposure correction (and no-crop unchecked) and that worked , so I can use enfuseGUI to blend them. Worked decently....

This can also be achieved in Hugin by choose "Exposure fused from any arrangement" in the stitcher tab.

BUT, I have vignetting that I would like to mitigate. So this is where fulla comes in! It would even handle chromatic aberration! Except I get, on Windows,  "Precondition violation! did not find a matching codec for the given file extension" no matter if I use jpg, jpeg, JPG, tif, tiff.

It works here fine. Have you given the file extension for the input *and* output image? e.g.

fulla --output=output.tif --vignetting=... input.tif

Thomas

 

[Jens Scheidtmann]

T. Modes <Thomas...@gmx.de> schrieb am Di. 23. Juni 2020 um 19:46:

Am Dienstag, 23. Juni 2020 03:56:03 UTC+2 schrieb Alister Ling:

There are two pairs of bracketed images which may be confusing things. i.e. I have 2 stacks in there. I can't seem to find the right stitch tab check boxes that will give me useful output. The output is really bad if I click calculate photometrics (in part because all the images were not taken with the same exposure settings?) - but I thought this was what Hugin could handle. 

Normally, Hugin can handle this. But in this case - where the images are mainly dark - it has too less information to calculate vignetting factors. But you can use another project with an higher dynamic range, store the vignetting parameters there in an ini file or in the lens database and load these parameters then in the other project.

You could try the following:

Put a white T-Shirt In front of your lens, shine with a white lamp on the T-shirt and take a few pictures. This needs the same stop, but otherwise you can let the camera decide on iso and exposure. This is called a “flat” by astrophotographers. 

Either Hugin is able to use these for calculating  vignetting factors or:

Using Fitswork, or similar software, divide your Image by the flat (pixel wise).

More Information on using flats for vignetting correction can be found in DeepSkyStacker’s documentation. 

Hope this helps,

Jens

[Alister Ling]

Hi Thomas and Jens, thank you for your quick replies!

Good catch Thomas, I was missing the extension (I got sucked in by the default for multiple images)

I will encourage my friend to make the flat properly - he has the equipment and is currently busy so I am doing the legwork on his images for him - I don't have his camera equipment to do it. I was hoping a theoretical correction would be close enough and quick. Interestingly lens fun does not have the vignetting for the lens. 

Coming to chromatic aberration: lensfun gives 

<tca model="poly3" focal="10" br="0.0001349" vr="1.0002043" bb="-0.0000252" vb="1.0001449" />

I see the red, red, blue, blue values. But the fulla command line asks for 4 for each colour:

fulla  -r 0.0001368:0.0002725:-0.0006605:1.0007630 \
       -b 0.0011642:-0.0046154:0.0055706:0.9989218 \
        input.tif

Not sure what to do here. 

I appreciate the time you put in to help folks - very kind of you.

[Gunter Königsmann]

The numbers look a bit like you could approximate a n from lensfun by an n:0:0:1-n in fulla...

Kind regards,
Gunter.
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.

[Paul Womack]

On Tue, 23 Jun 2020 at 18:46, T. Modes <Thomas...@gmx.de> wrote:

Normally, Hugin can handle this. But in this case - where the images are mainly dark - it has too less information to calculate vignetting factors. But you can use another project with an higher dynamic range, store the vignetting parameters there in an ini file or in the lens database and load these parameters then in the other project.

Yes - it is well worth taking a calibration image - I recommend a simple 360 rotation of a distant landscape, with around 50% overlap, taken in daylight. This is very easy to capture and process.

The 360 means that hugin can calculate the actual view angle of the lens, not estimate it from focal length, and the overlap allows vignetting evaluation.

   BugBear

 

[Alister Ling]

Thanks all. While technically not 100% solved, I am well on the way to addressing it. 

In http://hugin.sourceforge.net/tutorials/tca/en.shtml  one enters the distortion params and it calculates the values to put in fulla. Fulla runs properly, but my colours are not quite right, which suggests that I need to do as advised and calibrate it with a full 360 then save the lens data.

The "can't get Hugin" part, which gave seam errors and no output suddenly started working fine when we changed the output projection from equirectangular to rectilinear. 

Regards, have a good a summer as possible!