How I Use AI to Automate My Street Photography Workflow
Yo, what’s poppin’, people? Dante.
Today, I want to share with you some thoughts about automation and street photography and how we can use AI to enhance our workflows.
Studying Japanese Photography With AI
Here we have a study guide that I developed for a book titled First, Abandon the World of Certainty, which contains contributions from Kōji Taki, Takuma Nakahira, Daidō Moriyama, Takahiko Okada, Mike Amano, and Yutaka Takanashi.
This is something that I think is actually incredible about AI.
I have this rare book that I purchased in Tokyo titled First, Abandon the World of Certainty, which I’ll show you here.
I have it on my bookshelf.
The text is all in Japanese.
I picked this up in this pretty cool bookshop in Tokyo, and I was really intrigued because it contains the missing works, I suppose, from the Provoke movement.
I guess Provoke 4 and 5.
There is Provoke 1, 2, and 3.
And then I discovered this book.
The work itself inside is very interesting. Very blurry, very grainy, very out of focus, very mysterious work that I am just enamored by.
However, the philosophy behind Japanese photography really intrigues me, and this book specifically contains a lot of Japanese text.
And so I did the work of actually transcribing and translating all of the Japanese text within this book into English.
I used artificial intelligence with services from ChatGPT, Claude Code, and even ElevenLabs to create an audiobook and PDF files that are downloadable and legible.
I’ve created summaries of the book using these services and have even developed a printed study guide with some of the best quotes from the work, some of the most important ideas from the work, so that I can actually study more deeply the philosophy behind that kind of photography.
This is what’s incredible about AI. It allows me to take something that was previously inaccessible and transform it into something I can actually study and learn from.
Automating Photography Zines
As you can see here, I’ve been making documents for myself about how I make zines at home.
Pretty much cataloguing all the materials that I need and creating QR codes to open up links on Amazon for somebody to purchase if they’d like to.
I’ve got work that I’m developing automatically here inside these manila folders, where the work is automatically chronologically timestamped and sequenced using the metadata from my files, creating a fully formatted zine that I can print out automatically at home.
Essentially, you upload 36 photographs, and you make a zine automatically.
All you have to do is print it out at home and staple it.
Here we have a contact sheet that is automatically generated alongside the manifest document.
The workflow is easy and repeatable to the point where I’ve been cataloguing all of my photographs in a physical and digital archive that I store here in my home and online.
Building the Flux Photography Archive
The way that I’m storing my photographs online is actually really radical and interesting.
If you go to flux.dantesisofo.com, you can view my full archive.
There are around 1,200 photographs, and I actually have them all organized on a timeline.
On the left-hand side, you can pick out any month, any day, any time, and look at the photograph that I made on that particular day.
You can view the information with the camera settings.
You can download the JPEG file.
I even implemented a feature where you can add a photograph to a cart.
There’s a cart feature here on the side of the archive, and if you select 36 pictures and add them to a cart, anybody who uses my archive can generate one of these automatic zines.
However, with my personal backend, when I upload photographs to my own archive, the archive automatically ingests the work and creates full project pages.
The contact sheet, the sequencing, the physical PDF file, the project page, and the zine are created instantly.
I can click download PDF, access this PDF, and print it out at home.
And the craziest thing is, I’m not doing anything.
I’m simply uploading new work that I select to my website, and the project is automatically generated.
The URL is created.
The PDF is created.
The zine is created.
A ZIP file containing all the original contents is created.
A contact sheet is made.
And then all you’re left to do is print the work.
It’s automatically recognizing when 36 new pictures are uploaded to my website to create that project.
Anytime I have 36 new pictures, a full webpage and physical zine are generated.
And it’s open source and accessible to anybody in the world.
Archiving Philadelphia With GPS and AI
I’ve also created a system for archiving my city.
While I hit the streets of Philadelphia, implementing the Ricoh GR World app into my workflow is allowing me to create GPS timestamps for each file.
I can catalog different streets and automatically create projects where I’m archiving my city on a map.
Here we have the first street that I walked on, Broad Street.
You can click on any of the different points on the map and look at the photograph.
It’s pretty incredible what you can do with AI because these projects are automatically generated.
I have a full system that’s taking over the backend of my practice.
I’m making the selections still and picking what pictures to put within these projects.
However, the map that’s generated, the project page that’s generated, and the zine that’s generated from these projects can all be done automatically.
I have this full map of my city that I can update whenever I create a new project.
We have documentation of how many streets I’ve walked, how many photographs I’ve made, how many miles I’ve walked, and how much time I’ve spent working on the project.
Flux TV: Turning Old Video Into an Automated Archive
I have a TV that I’ve been working on.
I have a couple of different CRT TVs.
Here we have one that I showed inside of a gallery.
Essentially, the work that I’ve been making using this old-school Handycam is on display.
Here we have this old Sony Handycam.
What I’ve been doing is creating this system where I essentially have a Raspberry Pi, which is a small computer hooked up to this TV.
I’ve used software that I programmed this remote with so that I can actually cycle through different channels on the device.
You can see my name, the channel, and I can go through different videos that I’ve produced.
What’s going on in the backend is that these video files that are produced are weird MPG files or something.
In order to properly create videos, upload videos, and work with the videos, I needed to create a system to extract the video and convert the files to MP4.
And so I created a suite of software.
I’ve been using Claude Code to create my own software.
For instance, here you see Flux TV, which actually connects directly to the Raspberry Pi.
I have Flux Capture.
When I dock this Sony Handycam into my computer, Flux Capture automatically runs.
It automatically detects when there are new files to ingest, converts the files to MP4, and stores them on a NAS drive.
And so I have my own cloud.
Flux Extract: Contact Sheets, GIFs, and Video
Here we have the Flux Extract tool.
With this tool, I can essentially drag in any of the video files I make, and it will automatically create a screenshot and a contact sheet every second.
I can create short-form videos from this.
I can also create GIF files with little animations.
Actually, you can see here on the Flux TV archive.
When you go through, you can watch the video that’s presented on the TV, but you can also see that there are GIF files that are automatically extracted and generated from these videos that I’m producing.
Which is pretty cool.
One of the things I did recently is archive and put online all of my old home movies from my childhood.
You can see old home movie footage from stuff that my dad recorded on a Handycam when I was a little kid.
Which is pretty cool.
There’s footage of me rollerblading and stuff like that.
I’m basically using AI to help me with all these projects.
For instance, I had to take these video files from a DVD, extract them, and create digital files.
Then I used services to create all these GIFs using Claude and other tools.
Look at that.
Oh, that video clip is so good.
And so I have these little CRT TVs displaying the work.
Flux Contact Sheet Studio
Going back to the suite of software that I’m making, here we have Flux Contact Sheet Studio.
You can basically upload your photos into this little studio, and then it automatically generates contact sheets like this.
With these contact sheets, there are QR codes associated with each individual small thumbnail photograph.
You can scan a QR code and access the digital twin file directly on a webpage.
I can go through my work physically and then access the work digitally very easily.
One thing I do with these contact sheets is have them all cut out into these small little thumbnails.
Then you can sort them into little groupings of photos and find the digital files when needed.
Which is extremely helpful and a very innovative and interesting way that I’ve been using AI in my photography workflow.
One of the other automatic zine generators I made creates these smaller zines.
As you can see, we have this tiny little pocket-sized zine.
Flux Print Studio and the Flux Aesthetic
Back to the suite of software, we have Flux Contact Sheet Studio.
I even have Flux Print Studio, where you can print photos with captions and timestamps below, showing the date and time in the Flux style.
Using the Flux aesthetics.
All of my documents are using that same typeface and same style.
I’ve created a full knowledge base of what Flux is using Claude and creating websites using Amazon.
I basically have these websites built using Amazon and am creating them as subdomains on my main website using WordPress.org.
I can even have the tools here on the website.
There’s the Flux Generator.
You can upload your own 36 photos and generate your own issue.
You can even submit it to me and have it added automatically into a catalog.
People all over the world are submitting work to me.
This is some work from a different photographer over there in Germany.
Other people can even use the geotagging tool.
I still have that on there.
Open-Sourcing the Flux System
I have this wiki.
This wiki is essentially just open-sourcing everything that I’m working on.
I should definitely update this soon.
But basically, it’s a knowledge base on how I’m building the archive and how I’m building these different tools.
If anyone wants to run Flux or essentially adopt this system, hopefully the wiki contains enough information that you can feed it to your own AI and create your own agent that will help you adopt a similar streamlined workflow.
I’m trying to make Flux something that is open source, that anyone can use and benefit from in some way.
Using AI to Select Photographs
One of the radical things is that sometimes you have a bunch of different photos from a scene.
Let’s say you made a photograph of a fire hydrant and you have five different variations or ten different variations.
You can send them to ChatGPT and have it analyze the different frames for you and tell you the filename that it suggests you select.
And I’m going to be honest with you.
It’s still important for you to make the human decision at this point with AI until it learns your preferences.
This is something I’m going to continue pushing and working toward.
Making it so that the archive is studied by my agent, so that when I allow the AI to make selections for me, it will understand my aesthetic sensitivities and tastes.
But nine times out of ten, when I tell you the AI is going to select the photograph that you want anyway, it really helps with the fast workflow.
Keep the Photography Workflow Simple
In terms of workflow, it’s so simple, guys.
You don’t need any crazy camera.
I use the Ricoh GR.
I have an iPad Pro.
I use the built-in Photos app, and I simply tap and select the photographs that I want to keep.
I create simple folder structures using the Photos app.
I don’t use Lightroom.
I mean, I still have Lightroom because I have my old RAW files and stuff.
But as you can see, all of the dates here with the photos that I’ve made, I go through small thumbnails like this, point and click, favorite, and just use the Photos app.
It’s just so much more simple.
The greatest thing about Apple, macOS, and iOS is that you can just AirDrop shit back and forth between different devices.
It’s super nice.
I’m currently experimenting with the Mac mini using this setup right now because I have some ideas about how I can use the Mac mini.
But that’s pretty much it.
Just sharing what’s going on in my brain this morning.
I think AI is extremely powerful.
If you’re a photographer and you’re looking to enhance your workflow and help yourself make things, there are definitely some innovative solutions that I’ve found.
I’m sharing that with you so maybe you can also find ways to use AI.
Archiving My Thoughts and Uploading My Brain to the Internet
This is actually a folder.
There are multiple folders.
In these folders, there are thousands of pages of my actual, I guess you’d call it intellectual property, whatever you want to call it.
It’s my own thoughts. This is me.
All of the videos that I’ve been uploading, all of the YouTube archive, all of the website archive, have been transcribed directly in my voice and printed out into these documents so I can read through my thoughts.
I can revisit my thoughts.
Watching my old YouTube videos.
Reading my old blog posts.
This is an extremely powerful way to create.
Essentially, you just take the audio file and transcribe it.
You can even simply upload it to the Voice Memos app on your desktop if you’re using an Apple computer and have that automatically transcribe the video.
Then what I do is copy and paste the transcripts from my audio files and format them using ChatGPT into a Markdown file.
I use the iA Writer app on my iPhone, iPad, or computer to create PDF files and Markdown files.
I can then copy and paste those and put them onto blog posts.
Once these audio files are transcribed and made into text on my website, now I have this infinite archive.
My brain has been uploaded to my website.
And so if you ask ChatGPT:
What does Dante think about layering?
What does Dante think about health and fitness?
What does Dante do other than photography?
ChatGPT will send you the answer.
And so I’m cataloguing myself, archiving my work, archiving my videos, et cetera, for AI.
That way, in the future, I don’t know.
I just think this is the future.
I think that this is interesting.
The Digital Cyber Monk
Yeah, I’m a cyber monk.
Check this out.
I got a tatami mat recently.
I’ve been drinking tea and sitting on the floor.
Basically, this is the digital cyber monk setup.
Got the gym set up too.
Standing desk is pretty awesome.
You can just have a standing setup.
Basic laser printer.
We’ve got the Brother monochrome printer.
Got some label makers.
Phomemo sticker maker and label maker.
Canon SELPHY CP1500.
That’s pretty much all I need to make my zines, which is awesome.
You just need a pretty heavy-duty stapler.
Like a Bostitch heavy-duty stapler type jawn.
This one, I think, can staple up to 40 pages.
I don’t know what it is, but you need the heavier-duty stapler because the zines with 36 pictures are around 44 pages or so.
This is what I pull up onto the tatami mat with.
Check it out.
The AI cyber monk.
Got the cool teapot.
And then when it comes to how my brain is coming up with these ideas, I basically just fuel myself with burgers and meat.
Then I have the capacity to innovate with technology and my photography.
So that’s how it’s done.
The Future of Photography and AI
If you found this video useful or insightful or entertaining or whatever, visit the website.
Subscribe to the YouTube channel.
I’ve got a Substack now, so check the link in the description for that.
And also join the Discord server if you want to join the community.
With that being said, thanks for watching, and I’ll see you soon.
Peace.