Ciaran 
To be honest, I just open the image in Photoshop, paste it on top of a bright background color so I can see the edge well, and use the eraser tool by hand, recreating or refining the masked area as I go.
The bright background - like cyan or even lime green - helps show any voids in the figure itself, too.
As you mentioned, if the outline is already good, I use the wand tool at maybe a sensitivity of between 8 and 32 or so to grab as much of the outline as possible and invert the selection to make it easier, and then I zoom way in and take the first swath of 20 to 32 pixels off from the figure using the eraser by hand.
Basically I recreate the outside of the last stroke before the background. And for stuff like the “V” in hair it’s pretty quick to detail it with a tiny eraser, like 2px or 3px, and then go back with a 6px or larger to get the rest of it.
With a lot of the images that I grabbed to help with this collab, doing it by hand seems more appropriate than using a mask, because they were traditional art or stylized digital art where the outline wasn’t contiguous - like where clothes join the figure, and there were lots of “gaps” in the outline that a mask would turn into deep gouges in the edge of the figure.
Sometimes smoothing a mask’s edge gets those, but for me it’s easier to just go over it by hand.
Then once I have an outline around the figure I zoom back out and grab the remainder of the background with the wand set to 128 or even 255 and then expand that selection 3px, and delete it.
Then I copy the remainder and paste it into a new file - this helps me see if there’s any dots that got missed and it resized the canvas to the new image.
If it’s a rough original, I’ll add smoothing to the eraser tool, which helps take out some of the jaggies.
On some images using an antialiased eraser works best, bit I try to keep the antialiasing off, and the smoothing off, unless the original calls for it.
I’ve got a huge Huron tablet, so that helps it go fast.
I know that masks can do the same thing, or even better, but I’ve been doing this since the 80’s to remove the backgrounds from scans of line art, or to take out the backgrounds from model shots - so much “remove the background but keep the detail in the hair” in my life - so it’s just faster for me to do it by hand most of the time.
I can do it with masking sometimes, but it’s super rare that masking it doesn’t take longer for me than doing it by hand.
I’ve always envied people who could do it with mask-fu though. That seems like an incredibly useful skill.
For me, in these images the edge seems more important than the body of the figures, because they’ll be laying on top of each other, and up you don’t know what colors it might end up laying on, so making sure the edge is as clean as possible is a real worry, because it effects not just the image you’re working on, but also the 2 or maybe even more images it will be on top of.
Senior Moderator
友情は魔法だ
I’d like to know how you was casting your bg removals so quickly, if you don’t mind?
it’s some Ps Action or just good quick mask jutsu?asking because I do bg removal by Ps Pen Tool, outline silhouette, then convert the path to selection, inverse it, modify if needed, del the bg, and then trim the remained transparent bg.this method gives very quality result, but kinda long, so I just wanna know how to make the process faster.
To be honest, I just open the image in Photoshop, paste it on top of a bright background color so I can see the edge well, and use the eraser tool by hand, recreating or refining the masked area as I go.
The bright background - like cyan or even lime green - helps show any voids in the figure itself, too.
As you mentioned, if the outline is already good, I use the wand tool at maybe a sensitivity of between 8 and 32 or so to grab as much of the outline as possible and invert the selection to make it easier, and then I zoom way in and take the first swath of 20 to 32 pixels off from the figure using the eraser by hand.
Basically I recreate the outside of the last stroke before the background. And for stuff like the “V” in hair it’s pretty quick to detail it with a tiny eraser, like 2px or 3px, and then go back with a 6px or larger to get the rest of it.
With a lot of the images that I grabbed to help with this collab, doing it by hand seems more appropriate than using a mask, because they were traditional art or stylized digital art where the outline wasn’t contiguous - like where clothes join the figure, and there were lots of “gaps” in the outline that a mask would turn into deep gouges in the edge of the figure.
Sometimes smoothing a mask’s edge gets those, but for me it’s easier to just go over it by hand.
Then once I have an outline around the figure I zoom back out and grab the remainder of the background with the wand set to 128 or even 255 and then expand that selection 3px, and delete it.
Then I copy the remainder and paste it into a new file - this helps me see if there’s any dots that got missed and it resized the canvas to the new image.
If it’s a rough original, I’ll add smoothing to the eraser tool, which helps take out some of the jaggies.
On some images using an antialiased eraser works best, bit I try to keep the antialiasing off, and the smoothing off, unless the original calls for it.
I’ve got a huge Huron tablet, so that helps it go fast.
I know that masks can do the same thing, or even better, but I’ve been doing this since the 80’s to remove the backgrounds from scans of line art, or to take out the backgrounds from model shots - so much “remove the background but keep the detail in the hair” in my life - so it’s just faster for me to do it by hand most of the time.
I can do it with masking sometimes, but it’s super rare that masking it doesn’t take longer for me than doing it by hand.
I’ve always envied people who could do it with mask-fu though. That seems like an incredibly useful skill.
For me, in these images the edge seems more important than the body of the figures, because they’ll be laying on top of each other, and up you don’t know what colors it might end up laying on, so making sure the edge is as clean as possible is a real worry, because it effects not just the image you’re working on, but also the 2 or maybe even more images it will be on top of.
