A Realization on the Direction of Using Generative Services
I created a short video introduction course about the "Generative Fill" feature recently added to Photoshop 2024.
Personally, this is one of the most cost-effective features I've used while subscribing to Photoshop, and I believe it significantly boosts workflow productivity when combined with existing tools like the Stamp Tool, Healing Tool, and Content-Aware Fill.
However, after organizing this content, I suddenly had this thought.
'Ah, this is how I should use generative services.'
It's a bit of a strange story...
Looking at the flood of generative AI services these days, I find myself wondering how to use them.. Do I have to use them.. Will I fall behind if I don't? Many thoughts come to mind..
When using Photoshop's Generative Fill this time, I feel I have gained a sense of direction on how to use it at the very least.
Concerns and Unease About Generative AI.
I have been using generative image generators since around 2022 and held various concerns... I think this is because the functions are so powerful that I see many cases where they are used indiscriminately.
Especially for someone like me who has experienced infringement of rights to their own works, this is also a field where various concerns and thoughts arise..
Just as if to mock my concerns, since 2023, a wide variety of generative AI services have appeared in various fields, and recently, attempts to create videos themselves based on generative AI have even emerged.

It seems there is no field in content creation where AI-based generative services are not being used; they are being utilized across such a wide range of areas, and this trend is expected to deepen even further in the future.
In this environment, people like me who create content. Especially workers who have been using existing methods, I think they likely carry both anticipation and a sense of burden regarding how to utilize 'generative services'.
我也是如此,起初对 ChatGPT 等服务惊叹不已,但除了演示之外,由于在实际工作中难以应用,我几乎不再使用它们。同样,对于 Bing Image Creator 或 Dall-E 等图像生成器,我也担心版权问题,且结果并不总是令人完全满意,因此只在某些情况下使用。
It was during the process of creating a video introducing Photoshop's image fill technology recently that I had this realization.
It should be a choice rather than a necessity.
The Generative Fill feature added to Photoshop 2024 this time is a function that, when a certain area is selected within Photoshop and a prompt is entered or a blank is entered, creates an appropriate example image as shown below.

It looks similar to the services we often expect from AI-based generative services that we have seen frequently, but actually using it isn't necessarily the case.
First of all, it is certainly not an all-purpose feature, and the quality is often terrible.
However, as I continued to use it, I came to understand the method: you need to use Generative Fill based on well-organized data to get good results.
It goes without saying, but I believe there are hints on how to effectively utilize upcoming generative services from that perspective.
Actually, when organizing images with Photoshop to create thumbnails and similar images, I follow the following process.
- Thoughts on the desired result of the image to be created
- Organizing image resources I have or can use
- Identifying parts that are difficult to use from image resources or need to be deleted when applied to the result I envisioned
- Infinite repetition
- Correction and final finishing
- Infinite repetition if necessary
I assume it will be roughly the same... Here, No. 1 is the "concept" that the creator must possess, and I think it is the most important part. Of course, the form or result changes a lot during the work, but I view it as a part that represents the creator's personality.
Indeed, the parts that waste a lot of working time are numbers 2, 3, and 4.. Here, depending on your understanding of the tools or your skills, you can also shorten a lot of time.
I believe the "Generative Fill" feature of Photoshop 2024 plays a key role right here. Of course, unfortunately, everything cannot be done with Generative Fill alone, but when combined with existing Photoshop skills, it truly increases productivity in amazing ways.
I think that in any field, when there are background patterns or designs, the automation capabilities of artificial intelligence are truly far superior to manual human work in terms of efficiency.
And going back to points 5 and 6, I ultimately think they are a matter for humans, that is, a matter of the worker's choice. And that is what is being evaluated.
Therefore, what I realized this time is that properly utilizing "generative services" ultimately means using them well within the scope of areas that can be improved through specific services, based on one's existing working methods.
If you approach it thinking the service will handle all your tasks for you, you will likely encounter many failures for now. (Well, who knows if such a service that replaces human roles will appear this year... but that would be in the realm of work, not creativity.)
Ultimately, the most important thing is still human
When I worked at a past research institute, our organization was the company that purchased and used the programs of the most famous global design company the most domestically. Because of that, the company provided various technical support and came to our company first to introduce the new version coming out next year and the technology.
That is why our organization also had representatives of users of the technology, such as evangelists, and the company persistently asked them to write columns about the new technology.
However, my senior, who utilized the company's technology well, mostly rejected them. It was somewhat hard to understand for me, a 20-something who was obsessed with new technology and always thought new things were better, so I remember asking him seriously about the reason one day.
'If we are consistently using a company's software, it means that software already possesses the essential functions we need for our work. However, there is no reason to unconditionally use unverified new technology when we already have functions that work well. Upon closer inspection, while some technologies are quite useful, they only apply to very specific cases. It is not our job as actual workers to dig into new technology or excessively market it for cases that may or may not occur. That is the job of the company's marketing team or early adopters.'
At the time, I didn't quite understand it,
As I entered my 40s, having accumulated over a decade of content creation experience, I found this to be very relatable. If you already have a tool you use well, just stick with it; there is no need to constantly search for new tools just to use them every time.
If you want to draw, just get some paper and a pencil and draw.
However, if a new tool for coloring new materials has arrived, try it once: if the result fits the picture I want to draw, use it; if not, don't use it.
I believe that no matter how good a tool is, you don't need to entrust your work and creative results to it.
Based on how I feel about generative services as of 2024, this seems to be the right direction for using them effectively.