Skip to content
Navigation Menu
Sign in
Products
Platform
GitHub
Repository hosting, code review, and collaboration
Copilot
AI coding agent for tasks, projects and automated work
Build & Ship
Actions
CI/CD and automation
Code Quality
Quality analysis, fixes, and enforcement
Codespaces
Cloud development environments
Secure
Code Security
Vulnerability detection and remediation
Secret Protection
Secret scanning and leak prevention
All Features
Developers
Learn
Documentation
Training
Certifications
Build
Build with Copilot
Customize Copilot
Integrations
Explore
Trending repos
GitHub Sponsors
Events
Changelog
Changelog
Your source for the latest features, improvements, and fixes across GitHub products, all in one place.
Businesses
By Stage
Enterprises
Team
Startups
Nonprofits
By Industry
Healthcare
Financial services
Manufacturing
Government
See all industries
Evaluate GitHub
Why GitHub
Customer Stories
Business insights
Trust Center
Add-ons
GitHub Advanced Security
Enterprise-grade code security
Copilot for business
Enterprise-grade AI features
Premium Support
Enterprise-grade 24/7 support
Resources
Learn & Connect
Documentation
Resource center
Blog
Events & webinars
Community Forum
Customer Stories
Integrations
Support
Customer support
Training
GitHub Status
Company
About
Partners
Changelog
GitHub Universe
GitHub Universe
The event for developers, by developers — talks, workshops, and what’s next for GitHub.
Pricing
Search
/
Sign in
Sign up
Appearance settings
You signed in with another tab or window.
Reload
to refresh your session.
You signed out in another tab or window.
Reload
to refresh your session.
You switched accounts on another tab or window.
Reload
to refresh your session.
Dismiss alert
{{ message }}
leejet
stable-diffusion.cpp
Repository navigation
Code
Issues
230
(230)
Pull requests
41
(41)
Discussions
Actions
Projects
Wiki
Security and quality
6
(6)
Insights
More
items
feat: add LTX-2 video generation support
- #1459
#1459
Closed
mudler
wants to merge 28 commits into
leejet:master
leejet/stable-diffusion.cpp:master
from
mudler:feat/ltx-video
mudler/stable-diffusion.cpp:feat/ltx-video
Copy head branch name to clipboard
Conversation
Commits
28
(28)
Checks
Files changed
Closed
feat: add LTX-2 video generation support
#1459
mudler
wants to merge 28 commits into
leejet:master
leejet/stable-diffusion.cpp:master
from
mudler:feat/ltx-video
mudler/stable-diffusion.cpp:feat/ltx-video
Copy head branch name to clipboard
Commits
Commits on Apr 23, 2026
feat: port LTX-Video (Lightricks) from diffusers
Show description for 9cd646f
mudler
committed
9cd646f
View commit details
Copy full SHA for 9cd646f
Browse repository at this point
docs: add LTX-Video status and testing guide
mudler
committed
2fc91df
View commit details
Copy full SHA for 2fc91df
Browse repository at this point
feat(ltxv): pivot port to LTX-Video 2.0 (video-only)
Show description for aa6909f
mudler
committed
aa6909f
View commit details
Copy full SHA for aa6909f
Browse repository at this point
feat(ltxv): align transformer tensor names with LTX-2.3 22B checkpoint
Show description for 57a3871
mudler
committed
57a3871
View commit details
Copy full SHA for 57a3871
Browse repository at this point
feat(ltxv): full LTX-2.3 22B structure: EmbeddingsConnector + 9-block VAE
Show description for 6de6a3a
mudler
committed
6de6a3a
View commit details
Copy full SHA for 6de6a3a
Browse repository at this point
feat(ltxv): add split RoPE (LTX-2.3 rope_type=split)
Show description for ebe038a
mudler
committed
ebe038a
View commit details
Copy full SHA for ebe038a
Browse repository at this point
feat(ltxv): LTXV2Conditioner stub + ignore extra checkpoint prefixes
Show description for 4f1155e
mudler
committed
4f1155e
View commit details
Copy full SHA for 4f1155e
Browse repository at this point
feat(ltxv): match checkpoint dtype for CausalConv3d weights (BF16 on LTX-2.3)
mudler
committed
3b0d5bf
View commit details
Copy full SHA for 3b0d5bf
Browse repository at this point
docs(ltxv): add e2e test script (load → convert q8_0 → vid_gen)
mudler
committed
fbe393e
View commit details
Copy full SHA for fbe393e
Browse repository at this point
fix(ltxv): match actual LTX-2.3 tensor prefixes in version detection
mudler
committed
8aee894
View commit details
Copy full SHA for 8aee894
Browse repository at this point
fix(ltxv): VAE decode — F16 conv3d (BF16 breaks cuda im2col_3d) + drop st_t-1 frames per temporal upsample
Show description for 2b52f4e
mudler
committed
2b52f4e
View commit details
Copy full SHA for 2b52f4e
Browse repository at this point
fix(ltxv): wrap VAE decode output as 5-D [W,H,T,C,N] for sd.cpp pipeline
Show description for f790864
mudler
committed
f790864
View commit details
Copy full SHA for f790864
Browse repository at this point
docs(ltxv): update status — end-to-end pipeline validated on DGX GB10 (BF16 + q8_0 GGUF)
mudler
committed
1f61cfa
View commit details
Copy full SHA for 1f61cfa
Browse repository at this point
fix(ltxv): keep scale_input=true for VAE decode, revert incorrect permute
Show description for fb94de7
mudler
committed
fb94de7
View commit details
Copy full SHA for fb94de7
Browse repository at this point
Commits on Apr 24, 2026
wip(ltxv): tensor stats logging + VAE PerChannelRMSNorm restored
Show description for bac2524
mudler
committed
bac2524
View commit details
Copy full SHA for bac2524
Browse repository at this point
fix(ltxv): massive quality improvements — connector pre-norms, VAE normalization, depth-to-space
Show description for a3c6205
mudler
committed
a3c6205
View commit details
Copy full SHA for a3c6205
Browse repository at this point
fix(ltxv): add final norm_out — transformer output std 57 → 1, produces real frames
Show description for 28656c2
mudler
committed
28656c2
View commit details
Copy full SHA for 28656c2
Browse repository at this point
docs(ltxv): update status — 8-step distilled produces photo-realistic frames
Show description for 0f0bc9f
mudler
committed
0f0bc9f
View commit details
Copy full SHA for 0f0bc9f
Browse repository at this point
feat(gemma3): Phase 1 — architecture skeleton + SentencePiece BPE tokenizer
Show description for b14ee37
mudler
committed
b14ee37
View commit details
Copy full SHA for b14ee37
Browse repository at this point
feat(gemma3): Phase 2+3 — forward pass matches HuggingFace to within bf16 precision
Show description for 5c26bbd
mudler
committed
5c26bbd
View commit details
Copy full SHA for 5c26bbd
Browse repository at this point
feat(gemma3): Phase 4 — 49-layer concat + per-token RMSNorm matches HF
Show description for 84d8c28
mudler
committed
84d8c28
View commit details
Copy full SHA for 84d8c28
Browse repository at this point
feat(gemma3): Phase 5 — wire Gemma-3-12B into LTXV2Conditioner
Show description for 52b1483
mudler
committed
52b1483
View commit details
Copy full SHA for 52b1483
Browse repository at this point
fix(gemma3): correct 49-layer concat layout + LTX connector to reference
Show description for 185f81c
mudler
committed
185f81c
View commit details
Copy full SHA for 185f81c
Browse repository at this point
fix(gemma3): drop duplicate 1/sqrt(head_dim) scale on Q — prompts now work
Show description for f79bcf8
mudler
committed
f79bcf8
View commit details
Copy full SHA for f79bcf8
Browse repository at this point
feat(ltxv): wire LTX-2.3 distilled sigma schedule for 8-step runs
Show description for e685b11
mudler
committed
e685b11
View commit details
Copy full SHA for e685b11
Browse repository at this point
docs(ltxv): reflect Gemma-3 text conditioning + distilled schedule
mudler
committed
81f8ebe
View commit details
Copy full SHA for 81f8ebe
Browse repository at this point
fix(ltxv-vae): use patchify convention for decoder's final unpatchify
Show description for 6fc619a
mudler
committed
6fc619a
View commit details
Copy full SHA for 6fc619a
Browse repository at this point
docs(ltxv): note the unpatchify-convention fix + checkpoint config check
Show description for 8363e79
mudler
committed
8363e79
View commit details
Copy full SHA for 8363e79
Browse repository at this point
You can’t perform that action at this time.
