Generational Fill - My Thoughts

Hello folks. This was going to be a podcast episode but the script framework that I wrote was upon review by co-host Gordon would offended listeners so I scrapped it in favour of a personal opinion piece.

Allow me to be clear on the use of technology to help the creative improve his or her results. I am in favour of it, particularly if it helps a creative hone his or her work to that person’s greater pleasure and success. Second, the oft-referred to AI tools are not AI at all, they are content infusion or injection tools that draw upon a large sample set and a powerful set of algorithms to create content based on what is in the sample set, the source of such samples nearly always being the intellectual property of another person, used without approval, accreditation or compensation. I refer to this as theft and those who use such things as looters. This last statement is my personal conclusion, you may choose to agree, disagree or not care one way or the other. I find the idea morally repugnant.

Adobe has for many years worked to make it easier for users to remove elements from an image, such as a power line, or some other element that is distracting or non-supporting of the image. Up until the release of Photoshop Beta v24.6, the most powerful option was content aware fill. With the Generative Fill added in the Beta release of Photoshop available to all subscribers to Photoshop, this function is even better than it was according to those who have tested it.

I have not installed the Beta. I use my computers and software for production purposes and choose not to create unnecessary risk by installing a product that is not commercially ready and the installation of which could cause breakage in the tools that are installed and working. Due to the licensing model of subscription software, I can no longer keep a test machine active for playing with beta products. You may certainly install the Beta if you have the proper subscription and try it yourself, but be cautious as I have seen a multitude of user issues as a result of this.

From the perspective of Generative Fill being the next level of Content Aware Fill, I have no issue as I am working under the presumption that the filling is done with elements of your own image.

However, what Adobe, it’s promoters and those who are paid and unpaid shills or influencers are trumpeting is the ability to create something in the image that does not exist. Whether that is an extension in one or two dimensions to the image, or the creation of an object or subject that does not already exist in the image, and therefore MUST come from an alternate source, is where the line is drawn for me. Adobe is certainly not the only software company engaged in this unauthorized property use (aka theft) but they are the hippo in the bathtub. If it is ok for Adobe, others will follow.

If you are ok with the use of another person’s intellectual property without explicit consent and due compensation, then we are in opposition. You are completely free to make this decision, but by my moral standards, it makes you a looter. If that makes you decide to no longer read or listen to my work, that is completely your decision. That I consider the practice to be looting is my moral imperative and I stand by it.

Are there worse things in the world? Certainly. However evil wins when good people stand by and do nothing, and thus I invite a pox on all businesses that steal other people’s property for their own gain, or by their own stupidity. Stupid is not an excuse for theft.

While I respect Adobe’s desire to add perceived value to their products, I won’t use those functions and am giving strong consideration to not using their products as a forward step. What I do is not your concern. What you do is not my concern either. Personally I would prefer an option on installation to choose not to use other’s work as part of the software but I expect that this will never happen.

I have every expectation of being in the minority here and since I do not live by the judgement of others, do not really care if I stand alone, but I believe that if you do not stand for something, you will fall for anything, to paraphrase an historic quotation.

I understand that Generational Fill’s ability to create from a prompt varies. This in unsurprising as it is the same engine used in Adobe’s not really AI image generator called Firefly that is also in public beta. All of these application types fail at things like eyes, finger count and even the number of limbs in their creation of a representation of a live entity. They are readily recognizable and while they will get better, there are already good and honest humans working on other software to detect this kind of fakery. Whether that prevents businesses, users and whatever from actively using content generated from stolen works is unknown, but given my personal opinion of humanity in general, it’s unlikely.

Questions are being asked by smart creatives already. The US Copyright Office has already ruled that works created with these kinds of tools cannot be copyrighted. Professional and amateur creatives are asking who actually has title to the work when it contains pieces that are not the property of the creator. I ask whether misnamed AI sharpening, noise reduction and scaling software when used should also limit the ability to enforce copyright on the finished image. I have my own view here, but as I do not engage in politics or governmental affairs, my opinion will remain my own and not hold sway over others.

Thanks to those who choose to read or listen here. If you are offended, that is your decision, and has nothing to do with me. If you care to support my work, one effective way is to use the link to B&H Photo Video when shopping for gear. There is no cost to you and I receive a small commission on your purchases which is sincerely appreciated. I’m Ross Chevalier and until next time, I bid you peace.