A first glimpse at horror within world models.
Testing scary Doom-esque landscapes with FPS movement with @overworld_ai's Waypoint world model. I am interested in developing a scary world👿
I generated the reference image with GPT Image 2. Upscaled with @topazlabs. Running on a RTX 4090.
#worldmodels #doom #fps #overworldai #waypoint
This is crazy: 2 minutes long continue consistent POV walkthrough with seedance 2. Real immersive horror lost place experience. Watch the full 12min film (YT link in the comments)
I've been building a tool that can generate an entire pre viz package, an easy way to present your vision. Cinematic, anime and pixel style. Drag and drop character and prompt. I used my profile picture as a character reference. Here are the assets:
Is there such a thing as “vibe filmmaking”? After all, the best stories come from real life and aren't planned. Many filmmakers say that the story is everything and has to be planned down to the last detail. But maybe that's just a relic of a bygone era when resources were more limited? We're on the verge of creating real-time films. I'm thinking of projects like *The Blair Witch Project*. And I'm experimenting with that '90s VHS look.
A first glimpse at horror within world models.
Testing scary Doom-esque landscapes with FPS movement with @overworld_ai's Waypoint world model. I am interested in developing a scary world👿
I generated the reference image with GPT Image 2. Upscaled with @topazlabs. Running on a RTX 4090.
#worldmodels #doom #fps #overworldai #waypoint
My one complaint about GPT Image 2 is that it actually does not get fingers correctly sometimes. It's rare but it does happen, and that's annoying because it's supposed to be state of the art lol
#gptimage2
Shoutout to @arcjet, excellent product if you want to secure your app at runtime against prompt injection, bots, bulk signups, etc! @darkframeai was having an issue with bulk signups to farm free credits and @arcjet stopped it in it's tracks!
Italian Horror LoRA preview has dropped on @huggingface! Dropping in comments. That makes two LoRA previews across two horror subgenres. I am expanding @darkframeai beyond a web app: open source and a podcast on the horizon!
Initial results from an Italian Horror LoRA training using @ostrisai. The way it captured the lighting is really impressive. The contrasting lights, shadows, red and blue hues. 😁
Model: Z-Image-Turbo
Steps: 1,000
Dataset: @OpenAI GPT Image 2
Total training time: 25m23s
Hardware: RTX 4090
Italian Horror LoRA Preview dropping on @huggingface soon!
Exploration of Italian Horror giallo actresses holding a variety of weaponry, shades of purple, psychedelic backdrops and a dash of mod fashion 💜
Images: GPT Image 2
Prompts: Gemini Flash 2.5
#aihorror#aifashion#italianhorror#italiangiallo#giallo
I’ve been stress testing the capabilities of what Seedance can do especially in the hybrid space. As a filmmaker I love being on set and working with actors. AI can’t replicate the happy accidents you create while filming. While it’s not perfect I can see how smaller indie films can start utilizing this motion transfer technique to tell more grand stories. Let me know what you think.
Created using @runwayml thanks to my CPP @c_valenzuelab
@carolletta Feel free to DM me, I have integrated many models including Seedance 2.0 into my platform @darkframeai. It's primarly for indie creators making scary movies, might be good for your monster creation ❤️😌
Have voice generation, music, etc.