[@Guido76]
The main point of my comment is that after 14 years of steadily improving, these types of software are not a joke. Even so, they were not intended on being the “only” solution for removing the background or removing the foreground and rebuilding with the known background.
>”But in the meantime, let’s stop saying that this software or its equivalents do their job “wonderfully”.”
“Wonderfully” would be an opinion so if anyone claims these types of programs do their job wonderfully, it would be a moot point. Maybe the developers might use that wording, I wouldn’t go that far. From the standpoint of what they are intended to do, such as you mention, “very blue sky, or a very smooth beach” they do a very good job and can speed up the entire process I used to do by cloning and rebuilding the background manually.
As for removing a solid background without losing the tiny strands of hair, they can do that too. There is a tutorial for that very thing on one of the first products in this category of software. Keep in mind, I stated, removing the background from a “solid background.”
Granted, when the background is complicated, again as you mentioned, “an animated street or forest background” of course these programs require tweaking of the “selected” areas, no different than GIMP, Photoshop, or any other good photo editors that use layers. Those that do not have layers do not have the “masking” ease of the others, but it is doable given enough time. I worked with photos of a couple’s hike where the husband and wife took turns photographing the others. They wanted to frame one of the photos. Unbeknownst to them, I created one that had everyone in the photo. After choosing it as their favorite shot, then they asked, “Wait, who took this photo?” I created the photo using an early version of MS Paint and printed it on a MIPS printer. Zoomed in, there are no odd or blotchy pixels. They did not know that sort of thing was possible. So given enough time, every pixel can be manipulated to perfection.
I started doing photography in the 60s, so I am probably about a decade older. I thought that I was going to be more in graphics until computers intervened. My career has been as a software developer, first on mainframes, then on the very first PCs (running CP/M before DOS came along) In those very early days of the PC, other software developers often asked other developers to “beta-test” their software as opposed to using the public. Software developers felt a kinship with each other, so they felt that another developer would be more trustworthy and understand the need for a good test. As a result, I got to see several early graphic programs developed before Windows came along, and many of those never got to the production level. I know that I was lucky to get exposure to so many graphics programs. I created a lot of graphics and photographic artwork using the best tools available at the time and still do. I am quite confident in my ability to know and understand these types of programs.
In the early 80s, a photo was considered as substantial proof in a court of law, but that had to change due to the new ability to manipulate photographs using photographic editing software.
My first Pinhole and View Cameras had no intelligence at all; everything depended on the operator. As someone that takes photos, you are aware that modern digital cameras in an Auto mode can recognize when the subject is a portrait or otherwise, and thus can adjust the camera settings for the “most likely” best outcome. Nikon even that that technology in the 70s before digital media was used. Now, look at what cameras and software can do today. Cameras can recognize many more situations. Like bats get a 3D understanding of what is in front of them, some cameras can use sound or light to determine distances. As a result, they can recognize not only a simple portrait but whether it is a group shot, people sitting on a couch, people standing, etc. Software can make the background of a portrait that has everything in focus, look out of focus except for the person, obviously very closely related to what Cutout can do. Video is not left out either. Watermark removal generally leaves a blotchy area behind. The watermark location does not move but the background does. I am working on software that “predicts” what will be in the same area and uses it instead, leaving a clean un-blotched image. Now, with deep-fake technology, any actor’s face (or other body parts) can be replaced. The technical world is changing at a fast pace.
As for “Give me a link to a site that sells software like Cutout and demonstrates real clipping against an animated street or forest background, and then I’ll watch it with attention and interest,” that is not what I do. That will be your task. What I can tell you is that we both know that sort of complicated background removal is currently not possible as an “automated” solution, but I have seen some that can do a pretty good job and faster than I can get to the same point; then I can take over for additional rebuilding. What these background remover software solutions are doing is improving and will continue to advance.
So, let’s let these opinions cook for another five to ten years, then we can revisit the topic to see what has changed. We both might be surprised what these programs can identify and rebuild behind the removed item.
Advancement is going to happen, and that is no joke.