Revolutionizing Portrait Photography: How I Use AI

I’m a tech nerd and often consider myself an early adopter of new technologies. It’s why I usually have the latest Apple OS downloaded on my iPhone and MacBook, even if it makes some of my programs buggy until they catch up. Because of that, I’m constantly researching and testing new tools to make my life easier, to remove the mundane, and to let me focus on what I love – creating and working with people. There are new tools that implement Artificial Intelligence (AI), which are powerful and can be leveraged in amazing ways. In this blog post, I want to walk through how I use AI tools for portrait photography editing in my business, both to be educational and transparent.

AI Portrait Retouching

The new AI tools for retouching and working with images are insane. I love creating my pictures, but I never enjoyed retouching them. I’d rather be working with my lights and camera, talking to the human before me, than sitting on my computer trying to remove stray hairs from a patterned shirt. It just doesn’t excite me. So, I’m always looking for tools that allow me to spend more time working with my clients.

AI Culling With Aftershoot

The first step after a photoshoot is to go through the images and remove the duplicates, blinks, process & test shots (if I tweak my lighting, I’ll snap a few frames to make sure it looks good before posing my client), or just bad images (and yes, every photographer takes them).

In the past few years, I’ve become more intentional when pressing the shutter button, which means fewer images are taken during a portrait session and fewer images to review on my computer. Unfortunately, it’s still a lot. While I may show a client 60 proof photos, I’ve already removed 300. This process of sorting through images is called Culling. I hate it. Culling a session takes 45 minutes to an hour, depending on the session.

Now, I only need 15 minutes. Several companies have recently introduced AI software that automatically sorts, culls, and ranks your images. With Aftershoot AI editing, I only need 15 minutes. My personal favorite is Aftershoot for the simple reason that it works with Capture One, my raw image processing software. Every other AI culling software only works with Adobe Lightroom. 

Aftershoot will import my image catalog and then sort it based on the kind of session (wedding, portrait & headshot, event, and more!). Once sorted, I’ll go through its selections and tweak where I may disagree (and it’s usually a very subtle difference of opinion). For a typical portrait session, Aftershoot takes about 5 minutes to complete the culling process and then 10 minutes for me to review and check it. Honestly, I could probably skip my check process and still be happy with the results 95% of the time.

Aftershoot has also recently added global retouching to its AI platform. You can purchase existing editing presets or upload your previous images and have the program learn your editing style. Once it’s learned your style (they recommend giving it 5,000 images or more), it’ll cull and edit all in one go. It’s really incredible. I haven’t used the AI edits feature yet because my editing style changes depending on the shoot. Maybe I’ll try it one day, but doing global edits has never been a significant time suck for me.

Capture One for AI Editing & Global Changes

Most photographers use a DAM (Digital Asset Management) program specifically for handling images. These include well-known players such as Lightroom, Bridge, and Capture One. The benefits of these programs are that they allow you to organize and catalogue massive libraries of images based on dates and other metadata tags, and they allow you to do a lot of foundational editing work. They excel at what I call Global Editing, AKA if I move a slider, the whole image is affected. These programs enable photographers to apply all global edits to a single image and then apply those same modifications to other pictures taken in the same lighting setup, dramatically decreasing the time required to edit files. 

Recently, they’ve all upped their game by adding really powerful masking tools and some portrait retouching features.

Capture One has been excelling in this area with its AI features. One of the coolest features is the drop-down of selection masks. In a given portrait, I can select a subject, or any part of their hair, eyes, mouth, body, skin, and more with a simple click. Masking is incredibly powerful because I can now make adjustments to just that selection. For instance, if I want to make a subject’s eyes pop more, I can select ‘Eyes’ from the quick selection menu and then, once the mask is created, tweak the brightness so that only the subject’s eyes are affected. The generated masks are fantastic. With recent updates, I can now apply the edited layer mask to other images via Copy/Paste, and Capture One will intelligently select the eyes of all portraits. I don’t have to create the selection/mask manually – Capture One does it automatically. Now I can make detailed exposure, contrast, and color adjustments in batches.

CRAZY!

Capture One has also started adding AI Portrait Retouching, which is a game-changer. You can now do skin smoothing, eye bag removal, blemish removal, and dodging & burning. And just like the intelligent masks above, I can copy and paste the edits to other images, and they’ll be applied via the AI. They’re actively working on expanding the capabilities of this tool so photographers can stay in Capture One for all their retouching needs, which is incredibly exciting.

Retouch4Me For AI Portrait Editing

It’s time for detailed editing once culling and global edits are knocked out. Detailed editing is super time-consuming. Removing skin blemishes, fixing color casts in the skin, removing flyaway hairs, cleaning up & shaping hair in general, removing wrinkles and loose threads in clothing, cleaning up backgrounds of dirt and debris, and then adding contouring via a process called Dodging and Burning is tedious and can take a full hour for a single portrait. It makes a massive difference in how the final image looks, but it’s mind-numbing work, and honestly, sometimes I just don’t have the patience to do it all.

Meet Retouch4Me.  A series of standalone Photoshop Plugins, these AI portrait retouching modules are simply mind-blowing.

• Dodge & Burn
• Portrait Volumes
• Portrait Healing (skin retouching)
• Eye Vessel Removal
• Eye Brilliance (makes eyes POP)
• Backdrop Cleanup (only good for studio backdrops)
• Skin Tone
• Whiten Teeth
• Fabric (removes wrinkles)
• Skin Masking (creates selective masks for face, non-face, and all skin)
• Mattifier (removes highlights)

And I can do them all in just one click. Two minutes later, my image is fully retouched, with all these steps performed, or just a selection of a few. I can also adjust how much each one applies the effect.

At first, I thought, “There’s no way it can be that good.” After comparing the results to work I would have done or some retouchers I’ve hired – yes, it is. Getting as good, or better, results for my clients as if I did the work myself, but in a fraction of the time?

If I don’t outsource my edits (usually for images with more unique retouching needs), I’m always using these actions to shorten my retouching time from an hour per image to approximately 15 minutes.

AI Portrait Retouching with Evoto

After discovering Retouch4Me, I thought we had reached the pinnacle of AI retouching. Then I found Evoto AI. Damn.

Evoto works like Capture One, in that I can make edits to one image, and then the program will intelligently apply those changes to additional photos. The difference is that the available editing options are a smorgasbord of tools specifically designed for portraits.

There are more options than Retouch4Me, and you have more instantaneous control of each editing tool’s effect intensity. You have to be careful, though; you can definitely overedit and make your portrait subject look abnormal. However, with a careful touch, you can achieve some truly incredible, detailed retouching in a flash.

I love the hair clean-up tools for removing wisps outside the head profile and the crosshairs within the hair profile. Cleaning up hair is the bane of my existence. Evoto fixes that with just three sliders.

Because Retouch4Me and Evoto are so similar, I typically edit in Capture One first, then export to Evoto before importing the images into Photoshop, where I use Retouch4Me tools as plugins. I see them as complementary tools vs competing.

Some key things to keep in mind are that Evoto is a standalone program that charges per image edited. Each Retouch4Me tool is a one-time purchase that works as a standalone or a plugin within Photoshop, and then it’s yours forever (updates included). Additionally, if you don’t like an Evoto edit and want to redo it, you must run the image through the program again, which means incurring another edit charge. Thankfully, it’s like 13 cents per edit export, so it’s not a huge deal.

Photoshop AI – Generative Fill & Generative Expand

If you follow my work, you’ll know I love having random objects flying through the air. Something about that chaos is just fun for me, and possibly a subconscious outlet for the madness behind my super-organized brain.

Anyways…

One of the challenges I often encounter as a creative portrait photographer is that some props and locations are nearly impossible to acquire or require a Herculean effort to integrate. If I couldn’t get that prop in real life, I’d have to do a very time-consuming and tedious composite. If not done right, it looks awful and fake.

Just released, as of this article, is Photoshop’s Generative Fill feature, a great example of how AI enhances portrait photography. My composting days just got easier, and my ability to step up the creativity in my images has exploded — Generative Fill leverages Adobe Firefly, a massive stock image database. When used, Photoshop will search the database and match color, lighting direction, texture, and other elements, and place that content within your selection.

For example, I extended the background floor on this image, which made it look empty and bland. I made a selection in Photoshop, chose the Generative Fill tool, and then typed in what I wanted (in this case, green muslin fabric and again for a wooden pallet). BAM – I have objects in the image that match the rest of the scene and look totally normal.

Another example is this self-portrait. I wanted books to fall into the background, but I didn’t want to risk damaging any of them and lacked the extra hands to do the dropping action. So I used Generative Fill and told it “falling book” as a prompt. Without risking damage to any books or my back, I created the image in my mind.

Some creators take advantage of this feature to do some insane compositing swaps. You can entirely change someone’s outfit with scary accuracy and fit. I think that’s taking it too far, but I love the potential. This means I can create unique, gorgeous, and complicated portraits for my clients without spending a fortune on props or locations, while simultaneously minimizing risk.

Additionally, there is Generative Expand. One of my favorite print products is Canvas. I love the texture of the material and the wrap-around effect so that the image can be appreciated in 3D. The downside of Canvas prints is that you can sometimes lose critical aspects of an image because some of it has to be wrapped around the frame.

With Generative Expand, I can make an image “bigger” and Photoshop will fill in the empty space with contextually accurate data. For instance, if I want to post a picture with Instagram’s 4:5 ratio, I’ll need to crop off the legs or head of my subject because camera sensors have a 2:3 ratio (longer and narrower than 4:5). With Generative Expand, I can change the crop to 4:5, and Photoshop will add background to make it look like the original image. It’s pretty incredible and does it instantly.

AI For Blogging

Okay, this might be a little meta to read about AI for blogging while reading a blog post, BUT – this post had zero AI used for the writing (just help in two adjacent spots that I’ll reveal in a moment).

I genuinely enjoy writing. I’m weird that way. Even doing technical documents for my engineering job was something I didn’t mind. Still, it’s time-consuming, and there are certainly days when I’ve sat at my desk and stared at my monitor, unsure how to write or frame a specific item. AI has been beneficial in reducing those burdens.

I use ChatGPT for my blog writing, which has sparked a range of mixed feelings. I use it primarily as a research and structure tool, though sometimes I ask it for help with writing content itself.

AI for Blog Research

One of my favorite uses for ChatGPT is to provide it with a body of text, a hyperlink, an article, a URL, or a book, and ask it to summarize the content in a series of bullet points. If I’m in a time crunch and want to know the critical details of some text without reading it, ChatGPT does an outstanding job condensing it to the core points. It’s a great AI tool for content research.

When brainstorming for a new blog post, I’ll often ask ChatGPT for details about a topic. It’s like having a research intern who will gather information for me, allowing me to determine if it’s something I can leverage for my next post. From there, I’ll try to add my twist, perspective, and some additional thoughts based on my experiences. This could also include asking ChatGPT how most people search a topic in Google: “What common search terms are used for someone looking for an Atlanta corporate headshot photographer?”

Once I’ve selected a topic to blog about, I’ll often ask ChatGPT to help me write a catchy, SEO-optimized Title and Meta Description for that post, leveraging a specific keyword. This prompt optimizes the SEO of the post, which improves its ranking on various search engines. For this post, I did exactly that!

AI for Blog Outlines

Once I know the topic I want to blog about and the keywords, I’ll write a list of the main points I want to list as headers. From there, I can use ChatGPT for SEO to develop an outline using those keywords and headings. ChatGPT will create SEO-friendly headers that cover the discussion points and do so in a way that feels conversational vs. keyword stuffing (don’t do this).

Depending on the blog post, this may mark the end of my ChatGPT usage. I’ll then write out the content (body text) and add pictures with alt-text before posting.

AI For Blog Writing

I’ve done it. I’ve used ChatGPT to write blog posts for me. It’s incredibly helpful because sometimes proofreading a first draft and making modifications is much easier than writing a post from scratch.

And that’s a key point – I use ChatGPT as my crappy first draft. It gets the basics down before I add my personality and perspective. It shortens the overall writing process while providing good content and simultaneously conducting extensive research that’s included in the draft and likely in the final submission.

If I’m going to leverage ChatGPT for my first draft, I’m going to do it smartly. The goal is to reduce the time spent, not increase my workload! I achieved this goal by asking ChatGPT to create a draft based on specific criteria:

• Topic & keywords
• Number of words
• Inclusion of headings
• Based on my writing style (I ask it to sample previous blog posts)
• Perspective, audience, and tone

I had to conduct extensive experimentation to develop a prompt that met all of these criteria.

Worth it.

I can get a 2,500-word blog post with SEO-friendly keywords and headers in a matter of minutes. From there, I’ll review ChatGPT’s draft and make edits. This process typically takes between 30 minutes and an hour. That may seem like a lot, but it cuts down the overall writing process by about 3 to 5 hours, as I prefer writing fewer but longer blog posts.

What really makes this work well is creating custom GPTs. Custom GPTs are trained AI models tailored to the content I feed them. In this case, I fed ChatGPT a number of my previous blog posts and instructed it to learn my writing style so that it could replicate my voice. These custom GPTs cut down on editing time because I don’t have to add my language quirks – they’re already there! I can focus on the details of the content.

The other fantastic thing about custom GPTs is that you can build them around client avatars, AKA your target audience. My creative and conceptual clients are different than my brand portrait clients, so they are attracted to different language. A brand portrait client is concerned about elevating their brand to attract higher-paying clients. A conceptual portrait client cares about expressing themselves. To help craft blog posts and even web pages that speak specifically to each client, I trained those custom GPTs to respond like that specific avatar. No longer do I have to fumble with which language resonates with a target audience – my custom GPTs keep that in mind, and I simply have to curate and refine.

Custom GPTs require the paid version of ChatGPT, which may be worthwhile for at least a few months if you’re updating your website copy or refreshing your blog posts. 

Grammarly for Content Editing

While outputs from ChatGPT, Gemini, Claude, and Perplexity are usually grammatically correct, I still write most of my content, and I’m not. I’ve been a Grammarly user since 2020. The original platform did not and still doesn’t need AI. It uses machine learning to help correct any and all grammar mistakes, including punctuation, clarity, sentence structure, misplaced or unclear modifiers, and word repetition. I’ve become a better writer from using it because I pay attention to what it keeps asking me to change!

Recently, Grammarly added AI tools that enable you to adjust the tone of your text to be more professional, friendly, or other desired styles. I’ve found the recommendations to be typically solid, though I usually reserve this feature for my engineering day job.

You can get basic grammar fixes for free with Grammarly, but the paid version is worth its weight in gold.

AI and Website Development

I use ChatGPT very similarly for my website as I do for blogging. Having said that, I work with a copywriter 99% of the time (Krista Walsh Copywriter is fantastic and highly recommended, by the way). ChatGPT is particularly helpful when conducting competitor research to determine how I can effectively leverage my marketing and keywords.

One of my favorite tricks is to copy all the Google Reviews from a photographer and paste them into ChatGPT. I’ll then ask it to generate a bullet list of common words or phrases describing that photography service from those reviews.

I also used this on client reviews for my portrait services to help craft language for my website. 

It’s brilliant.

Lastly, if I do write new content for my website, I’ll often throw those paragraphs into ChatGPT and ask it to shorten my text. I’m a bit wordy, and most people want something brief on a website.

AI for Image Alt Text, Captions, and File Names

After I finish writing a blog post or updating website copy, the last thing I ever want to do is come up with file names, captions, and alt text for images. I’m guilty of writing the alt text but skipping the other two. While these three elements aren’t crucial to helping your webpage copy rank, they do help, and in the ever-competitive world of SEO, they are worth the few minutes of effort.

AI makes this even easier. Once I’ve finished writing some copy, I’ll link to it in ChatGPT, upload the images I want to use on the page, and then ask ChatGPT to create file names, alt text, captions, and recommendations for placement within the post for each image. I receive an Excel table as an export, which I can tweak as needed and then copy/paste where applicable. What typically took me 15 minutes now takes me two.

AI for Writing Emails

Before becoming an engineering project manager, my communication skills were solid, but they could have been more exemplary. They’re much better now, but I feel I can always do better. I leverage ChatGPT and Grammarly to help me craft messages that are concise, empathetic, and engaging, all while being to the point.

As I mentioned, I’m wordy, and I’ve often found that more text gets read less. So once I create my first draft of an email, I’ll plug it into ChatGPT and ask it to revise with a specific goal in mind. I get my message across, and the chance of my clients reading all that information increases substantially. It’s a win for everyone.

What I Don’t Like About AI

Yes, even though I leverage AI a lot in various aspects of my portrait photography business, there are still many aspects of AI that I don’t like. Frankly, it’s not so much AI itself but its usage that frustrates me. I’ve seen many people assume that they can let AI do everything for them, from blog writing to image creation and editing to organizing their lives. I’m not a fan of that kind of reliance or exploitation. 

Yes, AI is great for doing menial work and lets people do what’s terrific about us — combining and creating unique and novel concepts. However, completely offloading all work to AI is irresponsible and lazy, in my opinion. That’s how you end up with lousy information getting spread, like lawyers using arguments based on imaginary legal precedents. 

I also worry that it makes creators lazy. There’s something to be said about learning all the technical aspects of your craft so you know how to create something. With AI, creators can create amazing images, but there’s a concern about where the line gets drawn between digital art and photography (or similar for other artistic ventures).

I like using AI as a tool to help me create, not to do all the work for me.

Conclusion

I think we’re still in the early stages of what AI can do. It’s both disturbing and inspiring. With responsible usage, I believe AI will bring unprecedented improvements to the world. I look forward to learning and leveraging it to continue bringing the best experience to my clients. And take away some of the mundane aspects of my everyday life so I can focus on spending more time with friends and family. What do you think? Is AI worthwhile? Are you using it yet?