Credit: digital compositor

It was great to work on this A24 film, even though I worked on only about six shots. My work included enhancing water dripping from a hole in the ceiling and removing unwanted elements from several shots.

Credit: Compositing supervisor and compositing artist

The LA28 handover celebration took place on August 11, 2024, during the finale of the Paris 2024 Olympic Games Closing Ceremony. Tom Cruise lands on LA and the show starts on the beach with Billie Eilish, Red Hot Chili Peppers, and Snoop Dogg.

The performance was recorded in secrecy just hours before the end of the Paris ceremony. During an intense crunch, we delivered invisible and beauty visual effects for around a dozen shots in record time. I coordinated artists remotely across seven time zones, transferred 500 GB of data within a few hours, and supported a team of 18 visual-effects artists who had only 90 minutes to complete the work.

It was an honor to be part of this.

A new ComfyUI for VFX training course is available today on Borgo.tv Training. The course is in Portuguese and covers everything from installation to advanced Stable Diffusion workflows for image and video VFX work.

Topics covered:

• Local installation, interface, prompts, and prompt conditioning
• Stable Diffusion 1.5, SDXL, templates, checkpoints, LoRAs, ControlNets, inpainting, outpainting, IP-Adapters, upscaling, and sampler refinement
• Text-to-image, image-to-image, and animation with Stable Diffusion and AnimateDiff
• SDXL in depth: CLIP_G, CLIP_L, base and refiner workflows, resolutions, SDXL Lightning, SDXL Turbo, and PAG (Perturbed Attention Guidance)
• ProPainter and guidance for problematic or advanced ComfyUI installations
• FLUX.1, FLUX.1 ControlNet models, and inpainting with FLUX
• CogVideoX video generation

EDIT (Updates since the original post date):

- Wan2.1 video generation: text-to-video, image-to-video, Fun image-to-video with ControlNets, Advanced WAN, granular memory management, and accelerators such as SageAttention, TeaCache, and Enhance-A-Video - Wan2.1 VACE: video manipulation with inpainting, outpainting, ControlNets, first and last frames, and video upscaling - Loops for automation and extension of video workflows - Krea2 for artistic image generation, LoRA use, image referencing, ControlNets, and custom LoRA creation with ai-toolkit