Fisheye correction prototype

[Jeffrey Warren]

Hi, all - I'd been meaning to for some time, but I finally got a quick prototype up for in-browser fisheye correction:

https://jywarren.github.io/fisheyegl/example/

It doesn't currently do full-size correction, but I think that'd be pretty easy to add, similarly to how the webgl-distort library worked.

Source code: https://github.com/jywarren/fisheyegl

I'd love to know if anyone uploads a Mobius image with some straight/gridded lines, and identifies the correct settings to fix Mobius images for the different lenses folks have. This could make it much easier to use MapKnitter on them. Please write back if you manage to get it working with some Mobius images. 

If this seems to work, we could try to set it up to run on a batch of images as well. 

I went ahead and posted this as a research note too -- so if you'd like to upload example images you've tried, the comments there are a good place to do that!

https://publiclab.org/notes/warren/12-30-2016/web-based-fisheye-removal-prototype-for-correcting-lens-distortion

Thanks!

Jeff

[Scott Eustis]

Hey, Jeff

I was taking pictures of gridded notebook paper with the Mobius a while back, i settled on a "Main -42 Edges 0" to barrel distort images to orthogonal, just based on my eyeball.

https://publiclab.org/notes/eustatic/11-11-2014/a-method-for-distorting-mobius-action-cam-imagery-in-gimp-foss

There are two "Lens Distortion" filters in GIMP

Filters -> Distort -> Lens Distortion

"Main" and "Edge"

Here is the original image

Main -40

Main -41 edges +5

Main -42 edges 0

Main -50 edges 0

--
-- Post to this group at plots-organizers@googlegroups.com. To unsubscribe, email plots-organizers+unsubscribe@googlegroups.com. Options at https://groups.google.com/d/forum/plots-organizers
---
You received this message because you are subscribed to the Google Groups "plots-organizers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plots-organizers+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Scott Eustis
504 484 9599

 
Public Lab mailing lists are great for discussions, but to get attribution, open source your work, and make it easy for others to find and cite your contributions, register for the website, and describe your work in a research note.

[Liz Barry]

Oh, so the answer is "42"?

[Jeffrey Warren]

I see you couldn't hold that in, Liz :-)

So, i had this idea that the problem with the prototype was that it was assuming square images, and I re-ran both my test and Scott's after cropping them square -- and it worked really well!

See my comments below:

https://publiclab.org/notes/warren/12-30-2016/web-based-fisheye-removal-prototype-for-correcting-lens-distortion