Inside LTX-2.5 with the LTX Team | Fix It In Prompt Ep. 005

This week on Fix It In Prompt, Alex Villabon and Doug Hogan are joined by LTX VP Applied Research Ofir Bibi and Technical Director Mohamed Oumoumad to unpack LTX-2.5: how better data and Gemma 4 improve prompt understanding, why older LoRAs can survive model upgrades, and how Diffusion Fidelity Rendering spends compute where the final pixels actually matter. We get into open weights, LTX Trainer, IC-LoRAs, world models, real-time and 360 generation, alpha and depth outputs, multi-pass EXRs, physical AI, and why production teams need repeatable controls instead of a one-prompt movie button. 💬 If you could pull one production pass or control out of an AI video model today—alpha, depth, motion, normals, clean plate, or pose—which would change your pipeline most? 📬 CONTACT Email us: fixitinpromptpodcast@gmail.com Find Doug: ComfyUI & VFX website: https://www.doughogan.com ComfyUI Workflows: https://comfyui.sjv.io/OYydOG LinkedIn: / doughogan Find Alex: VFX + AI newsletter: https://alexvillabon.substack.com/ LinkedIn: / avillabon 🔗 LINKS MENTIONED IN THE EPISODE LTX-2.5 open-source overview: https://docs.ltx.io/open-source-model... Run LTX-2.5 with ComfyUI or Python: https://docs.ltx.io/open-source-model... LTX Trainer overview: https://docs.ltx.io/open-source-model... LTX LoRA and IC-LoRA guide: https://docs.ltx.io/open-source-model... ComfyUI-LTXVideo nodes and workflows: https://github.com/Lightricks/ComfyUI... ⏱️ TIMECODES 00:00 Meet the LTX team: Ofir Bibi and Mohamed Oumoumad 01:39 LTX-2.5 closes the speed-versus-quality gap 04:49 Data is king: captions, filtering, and reinforcement learning 09:15 Better prompting can mean saying less 13:07 LoRA compatibility across LTX generations 16:33 What makes LTX a world model? 20:42 Diffusion Fidelity Rendering: spend compute where it matters 32:21 Mohamed’s journey from VFX artist to model trainer 41:00 Open weights, community, and the LTX ethos 44:43 One base model cannot fit every production 49:02 Real-time diffusion, 360 video, and the road to LTX 5.0 54:44 Can a world model output VFX passes? 01:04:44 Open models versus building production on moving sand 01:07:17 Why VFX studios should fine-tune their own models #DellProPrecision #DellProMax #DellTech #NVIDIA #AI #VFX #LTX #LTX25 #ComfyUI #OpenSourceAI #LoRA #AIVideo #Filmmaking