# FreakyJolly Is Back


## Posts

- [MiniMax H3 Prompt Guide: Best Prompt Templates for T2V, I2V, FL2V and Reference-to-Video (ComfyUI)](https://freakyjolly.com/minimax-h3-prompt-guide-comfyui/): MiniMax H3 prompt guide for ComfyUI: the official field structure, copy-paste templates for T2V, I2V, FL2V and reference-to-video, camera moves, dialogue tags and fixes.
- [MiniMax H3 Out of Memory in ComfyUI: 16 GB GPU Fixes That Work](https://freakyjolly.com/minimax-h3-out-of-memory-comfyui-16gb-fix/): Fix MiniMax H3 out of memory in ComfyUI on 16 GB GPUs: tested settings for model load, Qwen text encoder, sampler and VAE decode on an RTX 5080.
- [Local Background Removal: Best Practices with U2NETP on CPU](https://freakyjolly.com/local-background-removal-u2netp-cpu-best-practices/): A tested local U2NETP CPU workflow with an original teapot fixture, transparent PNG checks, real edge analysis, and reusable scripts. Includes the limitations behind the numbers.
- [Add Sound Effects and a Music Climax to a Short Without Re-encoding the Video](https://freakyjolly.com/ffmpeg-sound-effects-music-climax-video-copy/): Build a licensed sound-effects library and music reveal with FFmpeg and Python while preserving video. Includes tested scripts, four CC0 samples and real verification reports.
- [Best MiniMax H3 Reference-to-Video Workflow for ComfyUI (Consistent Characters, 16 GB, Free JSON)](https://freakyjolly.com/best-minimax-h3-reference-to-video-comfyui-workflow/): My MiniMax H3 reference-to-video ComfyUI workflow for consistent characters on 16 GB: ref2va, 6-step Turbo, RIFE 72 fps + 1080p finish. Free JSON.
- [Best MiniMax H3 Image-to-Video Workflow for ComfyUI (First/Last Frame, 16 GB, Free JSON)](https://freakyjolly.com/best-minimax-h3-image-to-video-comfyui-workflow/): The MiniMax H3 image-to-video and first/last-frame ComfyUI workflow I run on a 16 GB RTX 5080: 6-step Turbo, synced audio, free JSON.
- [Best MiniMax H3 Text-to-Video Workflow for ComfyUI (16 GB VRAM, Free JSON)](https://freakyjolly.com/best-minimax-h3-text-to-video-comfyui-workflow/): My tested MiniMax H3 text-to-video ComfyUI workflow for 16 GB VRAM: 6-step Turbo LoRA, 4B ClipProj encoder, synced audio. Free JSON download.
- [ComfyUI Pixal3D Workflow: Image to 3D with a Real Example](https://freakyjolly.com/comfyui-pixal3d-trellis2-image-to-3d-workflow/): Turn one image into a textured static GLB with this ComfyUI Pixal3D workflow. Follow an actual character example, exact baseline controls, required models, and output checks, with optional ungenerated experiments kept separate.
- [Voicebox Tutorial: Free Local AI Voice Generation on Windows](https://freakyjolly.com/voicebox-local-ai-voice-generator-windows-tutorial/): Set up Voicebox on Windows, choose the right local text-to-speech engine, create a voiceover, and explore authorized voice cloning, expressive delivery, and multilingual narration.
- [MiniMax H3 ComfyUI reference to image: Consistent Character Workflow (Free JSON)](https://freakyjolly.com/minimax-h3-comfyui-reference-to-image-ref2va/): Turn MiniMax H3 ref2va into a still-image workflow with length 5 and ImageFromBatch. Includes tested settings, honest limitations, and a free API-format JSON.
- [Best Local AI Video Models in 2026: Text-to-Video, Image-to-Video and Reference-to-Video](https://freakyjolly.com/best-local-ai-video-models/): An evidence-linked practical countdown of local video model families, including text, image and reference workflows, hardware trade-offs and licensing cautions.


[comment]: # (Generated by Hostinger Tools Plugin)