I’ve just completed my latest AI filmmaking assignment with Urban Myth Productions. The challenge was to take three shots and run them through the complete filmmaking process — not just generate some AI footage and call it finished. I followed the same five stages used in traditional film production: 1. Development The idea, concept and creative direction. 2. Pre-production Planning the shots, character, framing, lighting and environment. 3. Production Creating the three shots using AI and directing the performance and visuals. 4. Post-production Editing the shots together, adding dialogue, sound and the finishing touches. 5. Distribution Preparing the finished film for an audience. The biggest lesson for me was that AI doesn’t replace filmmaking — it gives you a completely different set of filmmaking tools. You still need to understand story, performance, composition, lighting, continuity and editing. This was a small exercise, but it’s helping me build a repeatable workflow that I can eventually use for much larger AI film projects.

I've been trying something quieter after the driving shots: two people having a conversation in a cafe in Poznan's Stary Rynek. The people are AI-generated; the square is a Blender rebuild of the real place. FLUX 3 made the characters, and MiniMax H3 with Motion Context V2 carried the scene through four 8-second segments in ComfyUI. The image and video generation ran entirely locally on my 16 GB GPU with free workflows, without paying for cloud generations. The 32-second film took about 78 minutes to generate. It wasn't a clean first attempt. H3 turned the town hall into a grey block until I added a real reference photo. I kept that failure in the breakdown because it explains more than the final shot alone. https://www.youtube.com/watch?v=xqRHKUBqsD8 The film is first, then the process. Does the conversation feel like one continuous moment to you, or can you spot where the segments meet?

The Higgsfield API is made for code, so I built a free app for it: pick a model, add your images, write a prompt and press Generate. It works in Chrome, Edge and Brave, and you pay Higgsfield per generation from your API balance, no subscription. The app, a README and a step-by-step install video are in the Apps and Tools classroom: https://www.skool.com/aishotstudio-8297/classroom/55e13e57?md=27652a36ffbe43b196584ffca4b2a127

New video is up. Claude installed and connected Blender for me with one paste in PowerShell and one app restart, and in the second part I show the setup step by step with a clean prompt for each step. After that you just tell Claude in plain words what you want, and it builds it in Blender. THE PROMPTS Prompt 1 (install): I want to use Blender with you through the Blender MCP connector (GitHub ahujasid/blender-mcp, package mcp-for-blender). Check what is already installed on my Windows PC, then give me one PowerShell line that installs Blender and uv with winget. I will paste it myself. Prompt 2 (connect): Blender and uv are installed. Add the Blender MCP server to my Claude desktop config file: command "uvx", args ["mcp-for-blender"], env DISABLE_TELEMETRY = "true". Keep every existing entry in the file. Then install and enable the Blender MCP add-on in Blender, save the preferences, and check that it listens on port 9876. Tell me when I should restart the app. Prompt 3 (test, after the restart): I restarted the Claude app. Check that the Blender connector is running, build a simple grey-box test room in Blender and show me a viewport screenshot. First build prompt (night street): Build me a small night street for a film shot: wet road, a few buildings with lit windows, two street lamps and a parked car. Then add a camera that slowly moves toward the car for 5 seconds.

Hi everyone, I'm Can. I saw the invitation to share what we're working on, so here's a recent Blender and AI showcase I'd like some fresh eyes on. I rebuilt Devrim Stadium at METU and Poznan's Old Market Square, then cut between the real footage, clay models and finished scenes. The stadium section explores driving shots, camera flyovers and changes in light and atmosphere. At 0:59, the same setting moves through different lighting treatments. Around 3:02, the Poznan comparisons begin. https://www.youtube.com/watch?v=_EJuGTBYT8k Which shot has the strongest sense of depth, and where would you change the camera or lighting? Specific notes would be very welcome.