Hi everyone,

It's been quite a while since I last posted here, so I thought I'd share a few images from my latest AI wetlook project.

As many of you know from my old real-life WAM photography work, wet hair has always been my favorite aspect of the genre. For me, it's all about the transformation: seeing perfectly styled, voluminous hair slowly change into a glossy, shiny, soaked look. Every hairstyle reacts differently, and capturing those textures convincingly has always fascinated me.

One thing that has amazed me over the last couple of years is how much AI has improved. It went from producing obviously fake-looking wet hair to creating images where the shine, strand definition, and overall texture can look remarkably convincing. My eye still spots plenty of room for improvement, but the progress has been incredible.

Videos are a different story. While AI can generate great still images relatively well now, making hair behave naturally over time--getting the movement, water interaction, and consistency right--is still one of the biggest challenges. But with every new generation of tools, it gets a little better, and that's what keeps me experimenting.

I've attached a few images below from my latest Drenched in Style collections. If you're interested in seeing where this journey is heading, I've also started sharing exclusive wetlook collections, videos, and behind-the-scenes experiments on my Patreon.

https://www.patreon.com/wamphotoai

I'd love to hear what you think--and I'm curious whether anyone else here has been experimenting with AI to recreate the kinds of wet hair transformations we've enjoyed for years.

Cheers,

Chris
WAM Photography
WamPhoto AI

WOW!!!

These are great! I would love to know what AI you are using for these. I'm assuming something that you are running localy. Reve was my go-to for wetlook for a long time. It was hit or miss, but it was hit at least 50% of the time, and hits were hits. Reve 2.0 replaced the old model a few weeks ago, and it's horrible now.

It's not one AI but it's layers of various tools (depending what I need to achieve how annoying moderation is - especially with Veo3) and the images are extracted out of my video production workflow. But even then they are not merely screenshots but they again go through a workflow of their own before posting.

I have never used Reve, got to take a look. As for replacing models and tools, I feel you. I worked with grok a lot but recently the rolled out grok imagine 1.5 and practically killed wetlook with it. With no fallback option to the old model it was rendered useless for me.

When these AI image generators burst on the scene 3-4 years ago I think it really messed with people's heads, the ability to type in a few words and get an image 80-90% of what you were thinking seemed impossible. They've gotten incrementally better over the years but honestly from what I've seen progress has more or less stalled the last 10-12 months.

As it sits right now, on the photo-real side of things, I think they are firmly in the uncanny valley. At a glance they are pretty much perfect, good enough to fool rubes, high enough quality to satisfy some people, but ultimately they don't really pass for the real thing. It's the last 1-2% that makes all the difference. The 1-2% that Hollywood has spent decades and billions of dollars trying to recreate, and mostly has failed in doing so. I believe Deadpool 3 has the best CGI head replacement (for anyone who cares) and that really only works because of the weird circumstances in the film. Honestly I'm fairly optimistic that last 1-2% are impossible to overcome, I think society would start to break down if you could type a few words and get something indistinguishable from the real thing.

At the end of the day AI images are fast food, quick and easy, good enough in some cases, you can trick yourself into thinking you're eating something high quality. Then you eat real food and laugh to yourself at the thought of even comparing the two. And that's not even getting into the morality of it all, which is obviously atrocious.

It doesn't really matter what I think though, this is what the billionaires have decided to do with their time and money. Eventually there will be an inflection point, a high profile court case, someone being wrongfully accused with AI generated images or something, and maybe the tides will turn. But for the time being, we are in the AI age.

That's unfortunate to hear... I was going to bite the bullet and pay for Grok, but not if they neutered it's ability to do wetlook. I used to use it before they killed the usage for free. It was my favorite. But when they killed the free tier they also had a lot of other issues going on and I just didn't want to pay for it yet. Reve has an AI Agent built in, and I was working with it to try to get the new model to make wetlook anywhere near as good as the old one. The agent said it could do it, but after several attempts the agen said "Huh... I see what you mean. That's frustrating. Newer AI models are trending more towards "idealistic and perceived beauty" over "realistic", and I think that is what we are seeing here." It WAS pretty cool. You could build galleries of references and do a simple prompt of "Create an image of @charactername in a pool on a hot summer day. Her skin and clothing are dripping wet. Her hair is wet and slicked sleek and flat, glistening with water, limp with no volume like @poolhair. Candid. 2:3 aspect ratio" and it would churn out some awesome stuff. Now it churns out what reminds me of SD or Midjourney a year ago or so... Ugh.

It's good to see underwater hair but from what I've seen it's not very good with AI