ONE YEAR of Work for TEN SECONDS of Film

That's a Higgsfield submarine ad. You should watch how Corridor Crew uses and evangelizes the tech, as it's far more honest: link ▶ this video link link (These are all fascinating.) Or an AI video tech reviewer like Theoretically Tim who covers the space as it evolves: link

I've been a photons-on-glass filmmaker for over ten years, and I've been developing ArtCraft for myself, my friends, and my colleagues. All of my film school friends have a lot of ambition, but the production pyramid doesn't allow individual talent to shine easily. 10,000 students go to film school, yet only a handful get to helm projects they want with full autonomy - and almost never at the blockbuster budget levels that would afford the creative vision they want. There's a lot of nepotism, too. AI is the personal computer moment for film. The DAW. One of my friends has done rotoscoping with live actors: link The Corridor folks show off a lot of creativity with this tech: link link ▶ this video We've been making silly shorts ourselves: link link The secret is that a lot of studios have been using AI for well over a year now. You just don't notice it, and they won't ever tell you because of the stigma. It's the "bad toupee fallacy" - you'll only notice it when it's bad, and they'll never tell you otherwise. Comfy is neat, but I work with folks that don't intuit node graphs and that either don't have graphics cards with adequate VRAM, or that can't manage Python dependencies. The foundation models are all pretty competitive, and they're becoming increasingly controllable - and that's the big thing - control. So I've been working on the UI/UX control layer. ArtCraft has 2D and 3D control surfaces, where the 3D portion can be used as a strong and intuitive ControlNet for "Image-to-Image" (I2I) and "Image-to-Video" (I2V) workflows. It's almost like a WYSIWYG, and I'm confident that this is the direction the tech will evolve for creative professionals rather than text-centric prompting. I've been frustrated with tools like Gimp and Blender for a while. I'm no UX/UI maestro, but I've never enjoyed complicated tools - especially complicated OSS tools. Commercial-grade tools are better. Figma is sublime. An IDE for creatives should be simple, magical, and powerful. ArtCraft lets you drag and drop from a variety of creative canvases and an asset drawer easily. It's fast and intuitive. Bouncing between text-to-image for quick prototyping, image editing, 3d gen, to 3d compositing is fluid. It feels like "crafting" rather than prompting or node graph wizardry. ArtCraft, being a desktop app, lets us log you into 3rd party compute providers. I'm a big proponent of using and integrating the models you subscribe to wherever you have them. This has let us integrate WorldLabs' Marble Gaussian Splats, for instance, and nobody else has done that. My plan is to add every provider over time, including generic API key-based compute providers like FAL and Replicate. I don't care if you pay for ArtCraft - I just want it to be useful. Two disclaimers: ArtCraft is "fair source" - I'd like to go the Cockroach DB route and eventually get funding, but keep the tool itself 100% source available for people to build and run for themselves. Obsidian, but with source code. If we got big, I'd spend a lot of time making movies. Right now ArtCraft is tied to a lightweight cloud service - I don't like this. It was a choice so I could reuse an old project and go fast, but I intend for this to work fully offline soon. All server code is in the monorepo, so you can run everything yourself. In the fullness of time, I do envision a portable OSS cloud for various AI tools to read/write to like a Github for assets, but that's just a distant idea right now. I've written about roadmap in the repo: I'd like to develop integrations for every compute provider, rewrite the frontend UI/UX in Bevy for a fully native client, and integrate local models too.

I've been a photons-on-glass filmmaker for over ten years, and I've been developing ArtCraft for myself, my friends, and my colleagues. All of my film school friends have a lot of ambition, but the production pyramid doesn't allow individual talent to shine easily. 10,000 students go to film school, yet only a handful get to helm projects they want with full autonomy - and almost never at the blockbuster budget levels that would afford the creative vision they want. There's a lot of nepotism, too. AI is the personal computer moment for film. The DAW. One of my friends has done rotoscoping with live actors: link The Corridor folks show off a lot of creativity with this tech: link link ▶ this video We've been making silly shorts ourselves: link link The secret is that a lot of studios have been using AI for well over a year now. You just don't notice it, and they won't ever tell you because of the stigma. It's the "bad toupee fallacy" - you'll only notice it when it's bad, and they'll never tell you otherwise. Comfy is neat, but I work with folks that don't intuit node graphs and that either don't have graphics cards with adequate VRAM, or that can't manage Python dependencies. The foundation models are all pretty competitive, and they're becoming increasingly controllable - and that's the big thing - control. So I've been working on the UI/UX control layer. ArtCraft has 2D and 3D control surfaces, where the 3D portion can be used as a strong and intuitive ControlNet for "Image-to-Image" (I2I) and "Image-to-Video" (I2V) workflows. It's almost like a WYSIWYG, and I'm confident that this is the direction the tech will evolve for creative professionals rather than text-centric prompting. I've been frustrated with tools like Gimp and Blender for a while. I'm no UX/UI maestro, but I've never enjoyed complicated tools - especially complicated OSS tools. Commercial-grade tools are better. Figma is sublime. An IDE for creatives should be simple, magical, and powerful. ArtCraft lets you drag and drop from a variety of creative canvases and an asset drawer easily. It's fast and intuitive. Bouncing between text-to-image for quick prototyping, image editing, 3d gen, to 3d compositing is fluid. It feels like "crafting" rather than prompting or node graph wizardry. ArtCraft, being a desktop app, lets us log you into 3rd party compute providers. I'm a big proponent of using and integrating the models you subscribe to wherever you have them. This has let us integrate WorldLabs' Marble Gaussian Splats, for instance, and nobody else has done that. My plan is to add every provider over time, including generic API key-based compute providers like FAL and Replicate. I don't care if you pay for ArtCraft - I just want it to be useful. Two disclaimers: ArtCraft is "fair source" - I'd like to go the Cockroach DB route and eventually get funding, but keep the tool itself 100% source available for people to build and run for themselves. Obsidian, but with source code. If we got big, I'd spend a lot of time making movies. Right now ArtCraft is tied to a lightweight cloud service - I don't like this. It was a choice so I could reuse an old project and go fast, but I intend for this to work fully offline soon. All server code is in the monorepo, so you can run everything yourself. In the fullness of time, I do envision a portable OSS cloud for various AI tools to read/write to like a Github for assets, but that's just a distant idea right now. I've written about roadmap in the repo: I'd like to develop integrations for every compute provider, rewrite the frontend UI/UX in Bevy for a fully native client, and integrate local models too.

I've been a photons-on-glass filmmaker for over ten years, and I've been developing ArtCraft for myself, my friends, and my colleagues. All of my film school friends have a lot of ambition, but the production pyramid doesn't allow individual talent to shine easily. 10,000 students go to film school, yet only a handful get to helm projects they want with full autonomy - and almost never at the blockbuster budget levels that would afford the creative vision they want. There's a lot of nepotism, too. AI is the personal computer moment for film. The DAW. One of my friends has done rotoscoping with live actors: link The Corridor folks show off a lot of creativity with this tech: link link ▶ this video We've been making silly shorts ourselves: link link The secret is that a lot of studios have been using AI for well over a year now. You just don't notice it, and they won't ever tell you because of the stigma. It's the "bad toupee fallacy" - you'll only notice it when it's bad, and they'll never tell you otherwise. Comfy is neat, but I work with folks that don't intuit node graphs and that either don't have graphics cards with adequate VRAM, or that can't manage Python dependencies. The foundation models are all pretty competitive, and they're becoming increasingly controllable - and that's the big thing - control. So I've been working on the UI/UX control layer. ArtCraft has 2D and 3D control surfaces, where the 3D portion can be used as a strong and intuitive ControlNet for "Image-to-Image" (I2I) and "Image-to-Video" (I2V) workflows. It's almost like a WYSIWYG, and I'm confident that this is the direction the tech will evolve for creative professionals rather than text-centric prompting. I've been frustrated with tools like Gimp and Blender for a while. I'm no UX/UI maestro, but I've never enjoyed complicated tools - especially complicated OSS tools. Commercial-grade tools are better. Figma is sublime. An IDE for creatives should be simple, magical, and powerful. ArtCraft lets you drag and drop from a variety of creative canvases and an asset drawer easily. It's fast and intuitive. Bouncing between text-to-image for quick prototyping, image editing, 3d gen, to 3d compositing is fluid. It feels like "crafting" rather than prompting or node graph wizardry. ArtCraft, being a desktop app, lets us log you into 3rd party compute providers. I'm a big proponent of using and integrating the models you subscribe to wherever you have them. This has let us integrate WorldLabs' Marble Gaussian Splats, for instance, and nobody else has done that. My plan is to add every provider over time, including generic API key-based compute providers like FAL and Replicate. I don't care if you pay for ArtCraft - I just want it to be useful. Two disclaimers: ArtCraft is "fair source" - I'd like to go the Cockroach DB route and eventually get funding, but keep the tool itself 100% source available for people to build and run for themselves. Obsidian, but with source code. If we got big, I'd spend a lot of time making movies. Right now ArtCraft is tied to a lightweight cloud service - I don't like this. It was a choice so I could reuse an old project and go fast, but I intend for this to work fully offline soon. All server code is in the monorepo, so you can run everything yourself. In the fullness of time, I do envision a portable OSS cloud for various AI tools to read/write to like a Github for assets, but that's just a distant idea right now. I've written about roadmap in the repo: I'd like to develop integrations for every compute provider, rewrite the frontend UI/UX in Bevy for a fully native client, and integrate local models too.

The Corridor Crew [1] are luminaries in our field, and they are incredibly bullish on this tech. They've made dozens of essays and done tons of experiments showing that they think AI is going to be great for our field: link (scrub through the timelines to the end of these videos to see) ▶ this video link link Listen to them. Our entire industry pays attention to them, and they're right! [1] link

> More of this, less AI The comment was great without this part. I'm sorry I'm going to go on a rant, but seeing this day in and day out really gets to me. I'm a filmmaker. AI is an incredible tool that is making VFX affordable to those of us with budgets. My friends and I are already using it in roto/comp workflows: link link link (NSFW, bloody/gory - the effects look bad because we did this for a "film hackathon" and had an extremely short amount of time with no chance to fix the editing, script, lack of locations, etc. Under four hours for the VFX work.) Corridor Crew are making incredible stuff and teaching people how to do it: link link ▶ this video There's so much use by actual artists. I don't have a list in front of me, but I'll start compiling that to share next time this comes up. I am really salty about this because I am receiving hate and death threats because of this. I've had a film of mine booed in a live theater because I worked on a Blender + Mocap + AnimateDiff short film. I'm deep in the Atlanta film scene. I know a lot of the people that were in the audience personally. This was fresh off of the strikes, and the looming AI fears had people worried, so I understood the apprehension. But it still sucks to have your hard work and labor shunned like that in person. In any case, I won an award for that film. A year later, I did the same thing and everyone clapped. (It's not in my portfolio. I left a link in one of my previous comments if you care to dig.) It's super easy to say "AI sucks", "AI bros", "kill AI artist", etc., but it misses all of the hard work being done by those using the tools. I one day hope to have our small studio making high fantasy and science fiction films. I covet the art direction of Denis Villeneuve and Peter Jackson. I always have. My director friends that went to film school hoped to make films like that when they were younger, as I did I. The industry is very pyramid shaped, and unless you get really lucky or have nepo baby connections, you'll never get the chance to helm a project like that. Tens of thousands of students go to film school every year, yet most of their dreams wither on the vine. That's incredibly sad. We only get a small slice of human creativity, and what we do see is usually pretty generic. We have our Arris - that part has been democratized for a while. We have distribution. But we could never afford a $100 million VFX budget. AI changes the game and gives us so much hope that we can finally tell the stories we really want to tell in the way that we want to tell them. Instead of thinking of AI as "just prompting", think of it as an exoskeleton for practitioners. One analogy is that cameras can be used by anyone to make all kinds of junk - selfies, food pics, "butt dialed" accidental shots. Yet professional photographers are capable of making incredibly captivating art and immortalizing Pulitzer Prize winning moments. There are AI coding assistants for people who can't code (and I think this is awesome for accessibility and letting people act on their dreams - it's a gateway!), and there are AI coding models for full-time engineers. It's a lot like that. I've made a lot of experiments in this space. Huge ComfyUI workflows, custom trained models frankensteined from stitched together modules, a lot of Unreal and Blender-inspired work. I'm currently working on this [1] to enhance my own work. My desire right now is making it really fast to visually mold starting frames (which are like a director and DP blocking out scenes before shooting) and to precisely get the image you want before generating it. This year I'm going to make full AI animations as well as a bunch of rotoscoped human + AI films thi

This is a trip down memory lane! I remember when DeepDream first came out, and WaveNet not long after. I was immediately convinced this stuff was going to play a huge role in media production. I'm a big hobbyist filmmaker. I told all of my friends who actually work in film (IATSE, SAG, etc.) and they were so skeptical. I tried to get them to make an experimental film using DeepDream. This was about the same time Intel was dabbling in 360 degree filmmaking and just prior to Epic Games / Disney working on "The Volume". I bought a bunch of Kinects and built a really low-fidelity real time version of what Intel was working on. The sensors are VGA resolution, so it's not at all cinematic. When Stable Diffusion came out, I hooked up Blender to image-to-image and fed it frames of previz animations to convert to style transferred anime. Then IP Adapter. Then Animate Diff. My friends were angry with me at this point. AI was the devil. But I kept at it. I built an animation system for the web three years ago. Nonlinear timeline editing, camera controls, object transformations. It was really crude and a lot of work to produce simple outputs: link It was ridiculously hard to use. I typically film live action for the 48 Hour Film Project (twice annual film "hackathon" that I've done since I was a teenager). I used Mocap suits and 3D animation, and this is the result of 48 hours of no sleep: link We won two awards for this. The audience booed us. The image-to-video models came out right after this and immediately sunk this approach. Luma Dream Machine was so easy and looked so much better. Starting frames are just like a director and DP blocking out a scene and then calling action - it solved for the half of the problem I had ignored, which was precisely controlling for look/feel (though this abandons temporal control). There was a lot of slop, but I admired the work some hard-working people were creating. Those "movie trailers" people were criticizing were easily 10 hours of work with the difficulty of the tech back then. I found use in model aggregation services like OpenArt and FreePik. ComfyUI is too far removed for me - I appreciate people who can do node magic, but it's not my thing. I've been working on ArtCraft ( link ), which is a more artist-centered version for blocking out and precisely articulating scenes. My friends and I have been making a lot of AI films, and it's almost replaced our photons-on-glass filmmaking output. (We've done some rotoscoped AI + live action work.) link (live action rotoscoped film) link (EbSynth sketch about The Predator) link (Robot Chicken inspired Superman parody) link (JoJo grinch parody) We're going to do a feature length film at some point, but we're still building up the toolbox. If you're skeptical about artists using AI, you should check out Corridor Crew. They're well respected in our field, they have been for over a decade, and they love AI: link link link ▶ this video They're big ComfyUI fans. I just can't get used to it. Real filmmakers and artists are using this tech now. If you hate AI, please know that we see this more as an exoskeleton than as a replacement. It enables us to reach the look and feel of a $100+ million dollar Pixar, Star Wars, or Marvel film without the budgets we could never have without insane luck or nepotism. If anything, this elevates us to a place where we will one day be competing with Disney. They should fear us instead of the other way around.

I've been a photons-on-glass filmmaker for over ten years, and I've been developing ArtCraft for myself, my friends, and my colleagues. All of my film school friends have a lot of ambition, but the production pyramid doesn't allow individual talent to shine easily. 10,000 students go to film school, yet only a handful get to helm projects they want with full autonomy - and almost never at the blockbuster budget levels that would afford the creative vision they want. There's a lot of nepotism, too. AI is the personal computer moment for film. The DAW. One of my friends has done rotoscoping with live actors: link The Corridor folks show off a lot of creativity with this tech: link link ▶ this video We've been making silly shorts ourselves: link link The secret is that a lot of studios have been using AI for well over a year now. You just don't notice it, and they won't ever tell you because of the stigma. It's the "bad toupee fallacy" - you'll only notice it when it's bad, and they'll never tell you otherwise. Comfy is neat, but I work with folks that don't intuit node graphs and that either don't have graphics cards with adequate VRAM, or that can't manage Python dependencies. The foundation models are all pretty competitive, and they're becoming increasingly controllable - and that's the big thing - control. So I've been working on the UI/UX control layer. ArtCraft has 2D and 3D control surfaces, where the 3D portion can be used as a strong and intuitive ControlNet for "Image-to-Image" (I2I) and "Image-to-Video" (I2V) workflows. It's almost like a WYSIWYG, and I'm confident that this is the direction the tech will evolve for creative professionals rather than text-centric prompting. I've been frustrated with tools like Gimp and Blender for a while. I'm no UX/UI maestro, but I've never enjoyed complicated tools - especially complicated OSS tools. Commercial-grade tools are better. Figma is sublime. An IDE for creatives should be simple, magical, and powerful. ArtCraft lets you drag and drop from a variety of creative canvases and an asset drawer easily. It's fast and intuitive. Bouncing between text-to-image for quick prototyping, image editing, 3d gen, to 3d compositing is fluid. It feels like "crafting" rather than prompting or node graph wizardry. ArtCraft, being a desktop app, lets us log you into 3rd party compute providers. I'm a big proponent of using and integrating the models you subscribe to wherever you have them. This has let us integrate WorldLabs' Marble Gaussian Splats, for instance, and nobody else has done that. My plan is to add every provider over time, including generic API key-based compute providers like FAL and Replicate. I don't care if you pay for ArtCraft - I just want it to be useful. Two disclaimers: ArtCraft is "fair source" - I'd like to go the Cockroach DB route and eventually get funding, but keep the tool itself 100% source available for people to build and run for themselves. Obsidian, but with source code. If we got big, I'd spend a lot of time making movies. Right now ArtCraft is tied to a lightweight cloud service - I don't like this. It was a choice so I could reuse an old project and go fast, but I intend for this to work fully offline soon. All server code is in the monorepo, so you can run everything yourself. In the fullness of time, I do envision a portable OSS cloud for various AI tools to read/write to like a Github for assets, but that's just a distant idea right now. I've written about roadmap in the repo: I'd like to develop integrations for every compute provider, rewrite the frontend UI/UX in Bevy for a fully native client, and integrate local models too.