Jump to content
You must now use your email address to sign in [click for more info] ×

drscheme

Members
  • Posts

    5
  • Joined

  • Last visited

Posts posted by drscheme

  1. 23 hours ago, R C-R said:

    I noticed in your macros that the Inpainting step was named "Fill" rather than "Inpaint" or "Delete." I assume that was because you chose Edit > Fill rather than Edit > Inpaint when you recorded the macros. Is that correct?

     

    Yes. Does this make any difference?

     

    23 hours ago, R C-R said:

    Anyway, when I tested your unedited macros on a few rotated & cropped photos, the "Inpainting 1px grow" version produced partially transparent results in the inpainted areas while the "Inpainting 250px grow" version did not. But the 250px version also copied a lot of the photo into the inpainting area, producing things like floating partial heads of people in the photo above or to their sides & other similar artifacts. Using "Dave's Alpha Inpaint" macro instead (which includes the "Select opaque" & "Invert selection" steps) produced much better results -- not entirely seamless but free of major artifacts like floating heads & any partially transparent areas.

     

    Yes, I can confirm what you described. With one photo the 1px version produced much transparency. Interestingly that did not happen when I tested this macro with another picture. But you are right that the 250px version does odd things. Always. ^^

     

    23 hours ago, R C-R said:

    I have a theory about that: Using the Select > Alpha Range "Select Fully Transparent" menu option not surprisingly excludes any partially transparent pixels. So, for rotated/cropped photos, where the edge pixels are antialiased, those antialiased pixels are excluded from the selection. (This is obvious if you zoom in until you can see individual pixel selection boundaries.) Minimally, this will create a seam where no inpainting occurs around the edges. It also can create partially transparent pixels in the inpainting area like I described above for the 1px grow macro. I am not sure why that happens but my guess is it may have something to do with sampling along partial pixel boundaries ... maybe.

     

    Regardless, this would (I think) explain why in my panorama merge example I got the same results when I used the "Select Fully Transparent" shortcut as with the two step "Select opaque" & "Invert selection" version -- the panorama merge function produced 'clean' edges with no antialiased pixels, so there was no difference either way in which pixels were included.

     

    While I was testing I noticed that the "Select Partially Transparent" option actually includes both fully & partially transparent pixels. So I tried using that as a shortcut instead of the fully transparent option. I have not as yet tested it with many rotated/cropped samples but so far it works with all of them, producing results identical to using Dave's macro.

     

    I like your theory. Sounds reasonable. When I have more time to play around with it I'll share results.

  2. 14 hours ago, John Rostron said:

    I tried to follow this procedure (not the macro) on an image produced from the panorama merge. I found that I had to expand the selection of the transparent areas before it would work. It then worked well. It also said 'Delete' whilst it was executing the inpainting.

     

    Do I understand correctly? You first took the shortcut by NOT selecting the opaque areas and inverting the selection but instead you only selected the transparent areas (which also in my eyes should be equivalent)? I tried exact the same thing and this didn't work for me either. I played around with expanding the selection to include more non-transparent area of the image as I thought the inpainting algorithm need more input data. With 250px added selection I get quite odd results: the inpainted area is quite smudgy:

     

    Bildschirmfoto 2017-08-30 um 08.55.12.png

     

    However, it works better/fine when adding 1px to the selection.

     

    Bildschirmfoto 2017-08-30 um 08.55.30.png

     

     

    I added two versions of my macro for reference: docs_edge_inpainting.afmacros

  3. Thanks for your answers. I tried what you had suggested and this worked. However, I do not understand why this works. The image is already a raster image (JPEG) instead of a vector graphic... Odd.

     

    One addendum: When you straighten the image in Lightroom and then edit it in Photo to get rid of the edges, the inpainting fill works like a charm. Here the rasterization step is not necessary.

  4. I have a problem with Affinity Photo. I straightened the horizon line of an image. This of cause resulted in some white edges.I have watched a Tutorial of a guy on YouTube who used the Free Hand Selection Tool to mark (roughly) the area around the white spot, then he selected Edit -> Fill -> Impainting -> Apply and the white area was gone. I tried the same, but no effect. Can anybody tell me what is going wrong?

    Bildschirmfoto 2017-08-27 um 12.34.04.png

    Bildschirmfoto 2017-08-27 um 12.34.20.png

    Bildschirmfoto 2017-08-27 um 12.34.51.png

    Bildschirmfoto 2017-08-27 um 12.35.06.png

×
×
  • Create New...

Important Information

Terms of Use | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.