I get asked this question daily, so I wanted to address some major bullet points.
I trialed the new AI Generative Fill that Photoshop released under Photoshop Beta. While it was great to see Photoshop get into the mix, I had some reservations while using the program itself. Here I will outline the pros and cons and why I ultimately deleted the program from my computer.
First of all, it took up 4 GB on my hard disk. That is a lot of precious space and in comparison to Photoshop itself, which is only 3.5 GB, I questioned why I was allowing anyone to take up so much space without my consent. 😛
I did like the outcomes better than prior Photoshop AI tools I have trialed. This includes in-house features like content aware fill (which is my premier tutorial that I labeled content aware FAIL and linked here) and other well-known platforms like Dall-E and Stable Diffusion. What’s beneficial about Dall-E and Stable is that they are also performed in-house as a plug-in and don’t require an additional app to download. The photos generated I presumed most likely came from the large Adobe Stock database that always comes up as top search results.
For artists that decide to find other sources of photo images and don’t rely on Adobe Stock except for adding to thumbnails and basic marketing covers, this can be a hassle but the world isn’t perfect, and I would give overall a positive review on the tool itself.
Pros:
- For the most part the image is blended into the existing one fairly well. However I do believe the “magic” behind the logic is in plain sight IMHO
- This is a great quick fix tool that has a higher success rate than others, except perhaps Dall-E. However it can be argued that utilizing Dall-E requires lots of additional teaching and image prompting in order to generate the image that you seek. It’s not 100% handy, whereas Generative Fill is able to guess what you are trying to do and doesn’t necessarily need a prompt.
Cons:
- Bulky. 4 GB is a lot of retail space when considered to be an ADDITIONAL tool that currently doesn't have much use outside of a hobby (see next bullet point)
- You currently cannot use any of the rendered images for commercial use. For someone who only creates art in Photoshop for commercial purposes, this is a hinderance and waste of my 4 GB of hard space.
- While the quality is good, it doesn’t render an image exactly as I’m expecting in my mind. I even re-did a prompt to fix some smoke that was generated, only to like the second version even less. (Example video I created is below)
- The image quality is not up to professional standards and is maxed at 1024 pixels. While that isn’t that bad for the average user, many Photoshop professionals may find that the extra steps necessary to generate a higher quality image may be enough of a deterrent to not use the product itself.
While Photoshop is always going to be at the forefront of what is new in design I have a hard time continuing this usage. And while I know that it is going viral on social media it is simply not that fun to generate these prompts. I got bored of it after 2 uses and I am anxiously awaiting for the ability to render the images in a smaller space, like in a stamp tool.
For everything else, there are long-established presets and actions that can fix lighting and toning, etc.
Watch my 5 AI videos:
1) Eminem in AI and
2) my Follow Up Analysis. I've also included additional AI videos that I created using Generative Fill.
3) Friends in AI - you can see that there were some unexpected surprises
4)You can see a side by side comparison between images I created on my own (that I’m allowed to sell) and
5) images created by AI.
You will find many more AI and non-AI edits on my YouTube channel. Feel free to visit and subscribe!