

#NUKE INPAINT HOW TO#

Nuke’s IBK works by taking a greenscreen shot and comparing it with a “clean” pass of the same greenscreen.

It plays a key part in copying Nuke’s IBK (Image Based Keyer). However, in regards to keying, the Inpaint node is awesome!
#NUKE INPAINT PATCH#
The correct way to remove that twig from the wall would be to create a cleaned up patch image and track it in. you can see in this example, this isn’t always the best solution for removing things from an image. I drew a shape around the twigs on the wall.Īnd with the inpaint node, you can see how it pulls the pixels into the gap created by the roto, filling it. Here I created an alpha channel with the mask tool.
#NUKE INPAINT FREE#
If you want to addĪdditional inpainting algos, feel free to add a type and activateĬycore’s wire removal tool also does the same type of thing. Ones noted above), my node should work fairly well. Known surrounding pixels with a weighted average into the unknown P.S.: There are a lot more sophisticated solutions for inpainting, someĪre convolution based, like my simple approach (which convolves the Use of it, feel free to commit to tomato branch. If I should commit to trunk, please let me know. The only thing missing in blender for that task, is said inpainting The semi-transparent regions and subtract that instead. That’s no surprise, since until now, the GreenScreen Background wasn’tĪctually known (we don’t have any clean plates shot in Mango).īut: we can inpaint the surrounding greenscreen into the area behind Successfully can pull an alpha-matte (using a color channel node), butĬurrently fail to subtract the GreenScreen Background properly from the Object, we want to seperate, you’ll notice, that we more or less If you consider the usual compositing equationĬomposite = Background * (1-ALPHA) + Foreground * ALPHAįor the case, that Background is our GreenScreen and Foreground is the (compare key_raw.png to inpaint_key.png ) You can see first steps in this direction here, which already do some Solutions within Blender fail on hair details rather badly. My aim is implementing something like the IBK Keyer in Nuke It will bring it’s computing time from O(n^3) down to O(n^2). Step mode) to make it perform a lot better for dilate iterations greater That’s something, one should probably add also to the dilate node (in Since it first builds a manhatten distance map and after that performs The node isn’t tile based (for fundamental reasons), but very fast,

The later is important to actually improve keying in blender. In case, you don’t know, inpainting does this: Since we are directly before a release, I haven’t commited it, but I’d I wrote an inpaint node for the compositor. This seems like a big deal for the compositor. Sadly it seems that Blender is in lock down for 2.64 but maybe Mango could wedge it in, as there is more potential for another detail keyer! Potential for denoising here, depending on the algos employed :eyebrowlift.
