Aurel Manea
Aurel Manea
Generative AI Systems Architect | Applied AI Filmmaker | Technical Art Innovator
PROFESSIONAL SUMMARY
Early generative AI adopter and systems builder with 20+ years bridging real-time rendering, procedural systems, and machine learning workflows. Founder of one of the earliest large Stable Diffusion communities (70,000+ members).
Specialized in building practical AI pipelines using ComfyUI, WAN 2.2, LTX-2 and optimized local GPU infrastructure (RTX 3090 environments).
Focused on applied AI deployment, synthetic media systems, creative automation, and next-generation AI filmmaking workflows.
AI & GENERATIVE SYSTEMS LEADERSHIP
Founder – Stable Diffusion Community (70K+ Members)
- Built one of the earliest large communities around Stable Diffusion
- Educated thousands on LoRA training, diffusion workflows, GPU optimization
- Early experimentation with Disco Diffusion before mainstream adoption
- Structured technical breakdowns of quantization, memory constraints, and inference trade-offs
Beta Tester – Midjourney
- Early access contributor providing structured model feedback
- Comparative experimentation across diffusion architectures
- Analysis of prompt token behavior and aesthetic steering
Custom AI Pipeline Development
- Advanced workflow design in ComfyUI (multi-branch graph systems)
- Tiled VAE decoding for memory-constrained video generation
- GGUF quantization experiments (Q4, Q8)
- FP16 vs BF16 performance testing
- Lightning LoRA acceleration workflows
- Local + cloud GPU orchestration (e.g. RunPod)
- Optimization for RTX 3090 (24GB VRAM environments)
AI FILM & SYNTHETIC MEDIA SYSTEMS
- Directed fully AI-generated short films and music videos
- Built structured prompt-to-video pipelines using WAN 2.2 and LTX-2
- Developed cinematic motion control frameworks (camera rotation tokens, parallax sequencing)
- Experimented with motion consistency and diffusion timing stabilization
- Created reproducible visual style systems using LoRA stacking
- Designed image-to-video pipelines that preserve structural coherence
Focused on transforming generative AI from experimentation into repeatable production workflows.
TECHNICAL DEPTH SNAPSHOT
Models: Stable Diffusion, WAN 2.2, LTX-2
Inference: GGUF, FP16, BF16, Tiled VAE, Lightning LoRA
Frameworks: Advanced ComfyUI node systems
Hardware: RTX 3090 optimization (24GB VRAM), 64GB RAM environments
Cloud: RunPod GPU deployment
Core Concepts:
- Quantization trade-offs
- Memory bottleneck mitigation
- Diffusion step optimization
- Model loading constraints
- Performance tuning for large video generation
REAL-TIME RENDERING & ENGINE FOUNDATIONS
Extensive background in real-time systems using Unity and shader programming.
- Pioneer of early Romanian VR application development
- Built real-time volumetric and parallax systems
- Developed mobile SSAO solutions for early mobile hardware
- Created interactive 3D experiences using custom rendering pipelines
- Designed procedural and performance-optimized visual systems
This foundation directly informs current AI pipeline architecture and optimization work.
SELECTED INNOVATION PROJECTS
- Early Romanian VR application (Samsung Gear VR)
- Large-scale AI visual experiments integrating diffusion + motion
- Public technical breakdowns of AI workflows
- GPU performance experimentation across quantized models
- Synthetic narrative experiments merging AI + cinematography
CORE STRENGTHS
- Systems thinking across art + engineering
- Early adoption mindset
- Practical AI deployment (not surface-level usage)
- Performance optimization under hardware constraints
- Synthetic media pipeline design
- Community leadership and technical education
WEBSITE & PORTFOLIO
https://aurelm.com
Full AI experiments, film work, and technical breakdowns available online.
Key Achievements:
First Romanian VR Application
I developed the first Romanian VR application that combined my skills in photography, 3D, and
development. It was created in under three weeks, during which I taught myself Unity and the necessary C#.
Typically, I learn just what I need for a project, then move on and forget the rest. Review of the app (10 years ago): https://www.youtube.com/watch?v=nTjPjjyvpKM&t=121s&ab_channel=Otonka
Adaptability across challenges and disciplines , very technical and creative at the same time
experimented with hundreds of AI tools and adapted and used as needed the best one for the job and always keeping up-to-date. Very good intuition on how LLMs work and able to even hack them to say things they are not allowed to say. Learned intuition of the latent space and how symbols are connected to other symbols.
Pioneer in Generative AI
I was among the very few experimenting with generative AI before it became mainstream, starting in 2021 with Disco Diffusion and early beta Stable Diffusion. This early edge in AI tools, including technical mastery of platforms like ComfyUI and LoRA training, has brought me national and international media attention.
Example: https://petapixel.com/2022/08/16/these-are-not-photos-beautiful-landscapes-created-by-new-ai/
Speaker at multiple conferences about GENAI long before the hype (Example devplay 2022, Bucharest :
https://www.youtube.com/watch?v=2frapV4JZz0&ab_channel=Dev.PlayConference )
Beta Tester for Midjourney, Stable Diffusion, Dall-e. Had contact with the community and created the Stable Diffusion facebook group with more than 70k users : https://www.facebook.com/groups/stablediffusion
Viral Creative Work
My creative work has led to several viral videos, including what became the most viral video ever in
Romania, one featuring a local celebrity where I trained the models and voice, generated music, wrote the
text myself, also trained a model on myself also. The viral video became a sensation and it was covered and also I was invited on a lot of television stations and picked up by the whole press.
https://www.digi24.ro/stiri/un-clip-cu-ion-iliescu-in-2099-a-devenit-viral-eu-sunt-tot-aici-timpul-nu-m-a-putut-invinge-2914759
First Mobile SSAO Shader
I developed the first mobile Screen Space Ambient Occlusion shader fast enough to run on phones.
https://aurelm.com/2014/08/16/ssao-for-the-masses/
Universal Prototyping Shader
Created a 3ds Max prototyping shader used worldwide. https://aurelm.com/envo/
https://polycount.com/discussion/100187/envo-realtime-environment-shader-for-3d-studio-max
First documented handheld long exposure astrophotography in history:
Since I am a tech enthusiast and artist I always play around and experiment with new tech.
https://petapixel.com/2017/05/19/can-shoot-milky-way-hand-held-olympus-e-m1-mark-ii/
Strong Lighting & Artistic Vision
Informed by extensive experience in photography and video: https://aurelm.com/
Speaker about AI at DevPlay in Romania in 2023
https://dev-play.ro/speakers/aurel-manea/
Introduced AI in the companies I worked for.
In Gameloft especially since 2022 I held internal workshops training the artists in using open source generative AI, stable diffusion in that particular case since it was cutting edge in 2022. In Fortis Games, the last company I worked with I worked on implementing AI on every aspect of the creation pipeline.
General Ability to predict technologies and early adopt them.
Very enthusiastic about anything new, always eager to experiment with it and separate signal from noise.
Crossing the medium boundaries
My Parallax photography project that started first as video (no drones back then) converting photos manually to 3d meshes and reconstructing everything in a 3d scene to bring them to life:
https://www.youtube.com/watch?v=I3IXnZrV1-k&ab_channel=AurelManea ( tutorial here )
then move on to interactive web experience by learning javascript and three.js (project not working anymore now and it’s a real pain to try and debug after so many years 🙁
And then the final step, VR, to let the user experience the place I tried to capture (see at the beginning of the document).
First viral microscopy video 🙂
Since I experiment with anything I 3d printed an adapter for the mobile phone to my microscope (phone was a galaxy s5). The results were so good that it went viral with millions of views 13 years ago and I sold the rights for usage for a a lot of movie studios. So yeah, my blood is everywhere now 🙂
https://www.youtube.com/watch?v=9va0KPrVExs&t=71s&ab_channel=AurelManea . Info article here.
Skills
Advanced gen-ai skills and a time advance being one of the first to early adopt.
Experience with more that 100 AI Tools and services open source or closed source used in an interconnected way to get the job done in the best possible way. Training models, LORA’S, etc…
Unity, Unreal Engine, GLSL, C#, Blueprints
Shader writing, material systems, VFX, rendering
AI tools: Stable Diffusion, ComfyUI, LoRA training
3D modeling
Pipeline supervision
Advanced photographic, lighting & composition skills
Professional Career Timeline
Fortis Games – Senior Technical Artist / GFX Artist September 2023 – March 2025 – Bucharest, Romania
Unity shaders development
VFX creation
AI integration in company workflows
AI research and development
Gameloft – 18 years 8 months – Bucharest, Romania
Senior Technical Artist (Nov 2021 – Sep 2023) / 3d Technical Director (Nov 2019 – Nov 2021) / Lead 3d Artist (2008 – 2019) / 3d Artist (2005 – 2008)
Shaders for environment, postprocessing, Unreal blueprints
Generative AI for concept art (weapons, characters, environment) using Stable Diffusion, Midjourney
3D Technical Director (Nov 2019 – Nov 2021)
Supervised technology in the 3D rendering pipeline
Provided technical solutions for visual requirements
Prototyped new technologies
Oversaw 3D art department
Shader writing, materials, scripting, GFX
Studio Lead 3D Artist (Feb 2008 – Nov 2019)
Managed 3D arts department
Supervised visual quality of the Bucharest studio
Shaders, postprocessing, GFX development
3D Artist (Feb 2005 – Feb 2008)
3D modeling
Tool and engine development
Technical advice
Games worked on at Gameloft:
Gangstar: New York (PC) – Senior tech/VFX artist (Unreal Engine materials, postprocessing, weather
system, day/night cycle, car damage system)
Asphalt Streetstorm (Mobile) – Studio lead 3D artist (GLSL shaders, postprocessing, visual supervision)
Modern Combat 5 (Mobile) – SSAO shader (first performant SSAO for mobile, presented at GDC),
postprocessing
The Dark Knight Rises (iOS/Android) – Lead technical artist (shader systems, special FX)
N.O.V.A. 3 (Mobile) – Lead technical artist (shader & GFX dev, tools, tech supervision)
N.O.V.A. 2 / N.O.V.A. 1 / Terminator 4 / Shrek Karting (iPhone) – Modeling, tool & engine development,
technical supervision
Asphalt 4 (iPhone) – Supervision, technical advice
Ferrari Challenge (iPhone) – Engine development, technical artist, modeling, GFX
Wild West Guns (Wii) – Lead 3D artist, engine development, modeling, animation, GFX
Assassin’s Creed: Altair’s Chronicles (DS) – Lead 3D artist
Brothers in Arms (DS) – Lead 3D artist
Brothers in Arms 3D (Vcast) – Lead 3D artist
Vijay Singh Pro Golf 2005 3D (Vcast) – 3D artist
Image Promotion – Graphic Designer / 3D Artist
August 2000 – November 2004
3D work for commercials, TV shows, printed advertising, web design
Education
Self schooled, still learning everyday
important note: If you are looking for a coder I am not it. If you are looking for a multidisciplinary architect yes.