[EXPERIMENT][WIP] Add LTX-Video 2.3 OpenVINO notebook

[mlukasze]

⚠️ AUTOMATICALLY GENERATED BY OMEGA AGENT — REQUIRES HUMAN REVIEW ⚠️
This PR was created by an AI agent as part of automated model enablement.
A human maintainer must review and approve it before it can be considered for merge.
Do NOT merge without human review and sign-off.

Summary

This PR adds a Jupyter notebook demonstrating OpenVINO conversion and inference for Lightricks/LTX-Video 2.3.

Model

• Model: Lightricks/LTX-2.3
• Architecture: 22B DiT (Diffusion Transformer) for video generation
• Library: Diffusers (not standard Transformers)
• Components: transformer, vae_encoder, vae_decoder, text_encoder (Gemma3)

OV Conversion Results

Component	Size	CPU	GPU	Inference
transformer	18.88B params	✅	✅	✅
text_encoder	Gemma3	✅	✅	✅
vae_encoder	-	✅	-	✅
vae_decoder	-	✅	-	✅

Conversion Path

Lightricks/LTX-2.3 (42.98 GB safetensors)
  → convert_ltx2_to_diffusers.py (official Lightricks conversion script)
  → diffusers format
  → ov.convert_model()
  → OpenVINO IR

Notes

• Uses dg845/LTX-2.3-Diffusers (community diffusers format conversion) for easier loading
• Full end-to-end video generation pipeline wiring still in progress
• Related issue: intel-sandbox/applications.ai.openvino.meat#903

Status

⚠️ This is an experimental PR. The conversion works but full pipeline wiring (connector → transformer input mapping) needs final validation with real prompts.

Labels: experimental, do-not-merge

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB