There are too many good AI models to test, and that may be the best problem VFX artists have had in a while.
This week on Fix It In Prompt, Alex Villabon and Doug Hogan try to keep up with FLUX 3, Seedance 2.5, MiniMax H3, LTX 2.5, MiniMax Music 3, and Qwen3.8 27B while asking what agents actually change in a production workflow. They get into MCP, ComfyUI, Blender, Hermes, Unsloth Desktop, local speech-to-text with Handy, and why an agent is more useful as a collaborator with tools than as a vending machine.
Then Doug shows MiniMax H3 video-to-video, face-replacement, character-placement, and background-restyling experiments using masks, audio-driven performance, reference images, and LLM-built prompts. Alex shows how to let an agent generate overnight tests from a prompt guide. The through-line is simple: use open weights to mix models, make failures visible, and let artists curate more iterations.
💬 What would you hand to an agent first: troubleshooting ComfyUI, building prompts, or generating overnight tests?
📬 CONTACT
Email us: fixitinpromptpodcast@gmail.com
Find Doug:
ComfyUI & VFX website: https://www.doughogan.com
ComfyUI Workflows: https://comfyui.sjv.io/OYydOG
LinkedIn: https://www.linkedin.com/in/doughogan
Find Alex:
VFX + AI newsletter: https://alexvillabon.substack.com/
LinkedIn: https://www.linkedin.com/in/avillabon/
🔗 LINKS MENTIONED IN THE EPISODE
MiniMax Music 3 — open-weights music model:
https://www.minimax.io/blog/minimax-music-3-0-next-generation-open-weights-production-ready-versatile-music-model
Qwen3.8 27B:
https://huggingface.co/Qwen/Qwen3.8-27B
Unsloth Desktop — local AI workspace for models, tools, and training:
https://unsloth.ai/docs/desktop
Higgsfield AI Film Production Pipeline — production guide:
https://drive.google.com/file/d/1MRQoNq7dJD3s1OGWi94BKxoTIac4w_33/view
Higgsfield Zephyr: Special — project and production assets:
https://higgsfield.ai/@higgsfield.studio/projects/zephyr-special
Handy — open-source, offline speech-to-text:
https://handy.computer/
MiniMax Physics Test:
https://www.reddit.com/r/StableDiffusion/comments/1vjxn9l/minimax_h3_physics_test_seedvr2_rtx_vsr/
Alex's Random TV Prompter for MiniMax H3:
https://www.youtube.com/watch?v=UHmmsGG1dg4
LTX 2.5:
https://ltx.io/model/ltx-2-5
⏱️ TIMECODES
00:00 Too many releases, too little sleep
02:08 Why FLUX 3 and MiniMax are fun
07:13 The model-release firehose
09:52 Agents, Comfy MCP, and a one-minute generation
12:01 What an agent actually is
14:54 Unsloth Desktop as a local AI workstation
17:15 Turning apps agent-centric with MCP
19:03 Finding repetitive work worth automating
24:42 The Higgsfield production guide and open-source project assets
30:08 Handy, WhisperFlow, and local speech-to-text
32:24 MiniMax H3 and the 32B text encoder
37:04 Face replacement, audio-driven performance, and masking
40:59 Open weights, model combinations, and community-built workflows
44:17 Video restyling with LLM-built prompts
48:08 Overnight experiments and what models can do
50:07 From RFPs to overnight concept exploration
52:14 LTX 2.5 and choosing the right model for the shot
55:38 Experiment, share, and stay curious
#DellProPrecision #DellProMax #DellTech #NVIDIA #AI #VFX #ComfyUI #MiniMaxH3 #LTX #FLUX3 #Qwen #OpenWeights #Agents #MCP #Filmmaking #VisualEffects
