AI as a Starting Point vs. Manual Craft: How Do You Balance the Two? Let's talk Harmonize

Harmonize is probably doing pretty much the same thing that we used to do manually, but doing it very much quicker, and likely doing a better job of it.  It has to be 100% better at a local level (pixel by pixel as opposed to the entire composited element, or roughly masked areas of that element.

We might have created Actions or Adjustment Layer presets to streamline the process, or even scripts if you are clever.  Adobe have simply given us a new tool we can use in place of those semi-automated processes we used to use, but with the massive benefit of Ai analysis of the background and introduced elements at a pixel level.

So you are on shaky ground when clinging on to the reigns of your high horse.  Are you going to insist on using Quick Mask and the Magic Wand? Is the Healing Brush OK, but the Remove Tool is the work of the devil?   Do you feel that Content Aware Fill is OK but Generative Fill is cheating? 

That last one was a trick question, because Content Aware Fill uses Ai, so now whadda ya' gonna do?

Ahhh. I see AI tools like Harmonize as a great enhancer — they speed up the repetitive parts and give a cleaner starting point, but the real value still comes from the manual precision and personal style we bring in. For me, it’s about using AI as an assistant, not a replacement.

I recently designed custom arcade wraps for a friend who builds video machines. Harmonize was a lifesaver — lining up shadows and colors on those big side panels in seconds, so I could focus on making the artwork pop!

Well put AmyBeth. I agree with you. 

I appreciate the insight in your video. Though hesitant about AI in art--in general--placing an emphasis on AI as a "helper" to significantly reduce time on setting up tasks is appealing. After all, I want to spend more time on making the art and less searching around for ways to achieve a look I'm after. It's all in the tools, right? For instance, with a painting we're restricted by the paint tubes, canvas, and brushes. We've gotta' make due with what we have. The digital art world is so much different because access to tools is as fast as a quick search and a few button clicks--not multiple trips to the store or getting stuck in an Amazon rabbit hole. With digital I don't have to spend a lot of time on foresight to avoid constantly leaving my work area. I can jump in and start drawing knowing that a stroke, a brush, whatever can be replaced at any time. One may argue that THAT'S part of the art process, choosing materials and implementing new ones along the way, and impeding that with an AI shortcut is detrimental to that process. At any rate, I try to avoid anything (including AI and AI shortcuts) that takes the personal journey out the creation process. My approach is AI is like having an assistant available to make suggestions, and it can even do a task or two from time to time that doesn't interfere with the process: e.g., recommend a color pallet, give ideas on a particular character, etc. If it keeps my digital pen on the screen and eyes away from distraction, I'm down to try it. 

Lovely Video! And I love the point it makes.
Sometimes I wonder what I was doing before AI. Now I use AI as a co creator.
I can do more things for fun like photoshopping myself in "Time Travelling Photos" (And yes I know these can use further polish but this is fun!)

I do not and will not use Generative AI.

I like AI driven tools, they can be genuinely faster and more accurate than their non-AI counterparts, but I try and keep Generative AI out of my worksflow. Outside of the quality of the images, which can vary, the problems over copyright and ownership of images made with generative AI, it would seem that many commercial workflows prefer to keep away from it at present, as reflected in contract terms.
Then of course there is the 'soul', experience, and thinking behind an image. AI does not have that experience and soul and unless it ever starts to think like a human and draw on those experiences, it never will. Generative AI feels like commisioning someone, or something, to produce an image rather than picking up the tools and producing one. Typist vs artist?

Dave

Jane and I did Zoom group with Adobe NJ with Lisa Carney last year in which she talked about using Firefly and Generative Fill to dramatically speed up the artwork she does for the Hollywood movie and TV industry. She later did a session at MAX with Jesus Rameriz.  Jesus works with her and is apparently a wizard at creating and extending backgrounds (necessary because their artwork has to fit multiple formats and devices).  This really bought home the real world benefits of these Ai tools, and how you'd struggle to stay competitive if you ignored their time saving abilities.  They also talked about using the Ai to generate ideas.  If you follow Dave's SFTW I am sure you will suddenly get a new idea after seeing other people's uploads.  

This has become significantly more relevant with Google's Gemini 2.5 Flash.  Being able to describe the area you want to change rather than making complex selections can be a huge time saver, and the results are getting crazy good.  We still need to fine tune, and we will have to learn new skills like how to write effective promts.  People should at least try all the new tools before judging them or considering them cheating.  Then they can decide if they still want to be left behind.

https://www.adobe.com/max/2024/sessions/advanced-photoshop-techniques-from-two-hollywood-s6305.html

I think AI works best as a kickstarter—it saves time and gives you a rough draft. But the human touch is what makes the final piece feel real and unique. For me, it’s about letting AI do the heavy lifting, then adding my own craft to harmonize everything.