feat(ltx2): LTX-2.3 video generation (M1+M2+M3) — conversion, C API, parity, Bare JS addon - #7

Open

64johnlee wants to merge 161 commits into

tetherto:masterfrom

64johnlee:ltx2-video-generation

Commits on Mar 9, 2026

feat: add spectrum caching method (#1322)
rmatif
authored
dea4980
refactor: remove ununsed encode_video (#1332)
leejet
authored
d6dd6d7

Commits on Mar 15, 2026

docs: add Anima2 gguf download link to anima.md (#1335)
Juste-Leo2
authored
6fa7ca9
feat: add generic DiT support to spectrum cache (#1336)
rmatif
authored
adfef62
chore: remove SD_FAST_SOFTMAX build flag (#1338)
wbruna
authored
f6968bc
refactor: move all cache parameter defaults to the library (#1327)
wbruna
authored
630ee03
ci: add CUDA Dockerfile (#1314)
Nauscar
authored
83eabd7
refactor: optimize the VAE architecture (#1345)
leejet
authored
acc3bf1
ci: avoid cuda docker build timeout by using -j16
leejet
committed
61d8331
feat: add embedded WebUI (#1207)
leejet
authored
862a658

Commits on Mar 16, 2026

Commits on Mar 29, 2026

Commits on Mar 30, 2026

Commits on Apr 1, 2026

feat: ancestral sampler implementations for flow models (#1374)

wbruna
and
leejet
authored
99c1de3
feat: add webp support (#1384)
leejet
authored
87ecb95

Commits on Apr 5, 2026

chore: make libwebp optional and support system libwebp (#1387)

wbruna
and
leejet
authored
687a81f
feat: inpaint improvements (#1357)

stduhpf
and
leejet
authored
9369ab7
feat: add webm support (#1391)
leejet
authored
7397dda

Commits on Apr 6, 2026

Commits on Apr 8, 2026

Commits on Apr 11, 2026

refactor: simplify DiscreteFlowDenoiser (#1405)
wbruna
authored
be9f51b
chore: harden safetensors and gguf loading code (#1404)

wbruna
and
professor-moody
authored
118489e
feat: add sdcpp api support (#1407)
leejet
authored
7ade90e
feat: use sdcpp-webui as embedded webui (#1408)
leejet
authored
fd35047
docs: update readme
leejet
committed
12a369c
docs: update readme
leejet
committed
6b675a5

Commits on Apr 15, 2026

Commits on Apr 16, 2026

Commits on Apr 18, 2026

Commits on Apr 19, 2026

Commits on Apr 22, 2026

feat: add sd-webui style Hires. fix support (#1451)
leejet
authored
c97702e

Commits on Apr 23, 2026

feat: add more built-in highres upscalers (#1456)
leejet
authored
b8bdffc

Commits on Apr 27, 2026

Commits on Apr 29, 2026

Commits on May 6, 2026

Commits on May 14, 2026

feat: display server url with "http://" prefix. (#1486)
cphlipot
authored
6ee0684
docs: add new go bindings for stable-diffusion.cpp (#1480)
l8bloom
authored
60477fd
feat: add Euler CFG++ and Euler-A CFG++ samplers (#1354)
daniandtheweb
authored
9d68341
feat: support for memory-mapping model weights (#1414)

authored
57ff2eb
fix: Use PkgConfig for WebP and WebM (#1400)
candrews
authored
eeac950
feat: add hidream o1 image support (#1485)
leejet
authored
0665a7f
docs: add CONTRIBUTING.md
leejet
committed
381e0df
docs: add .github/pull_request_template.md
leejet
committed
0b82969

Commits on May 16, 2026

Commits on May 17, 2026

ci: update ROCm builds for Windows and Linux to use ROCm 7.13 (#1504)
superm1
authored
3b4d26f
feat: add ltx2.3 support (#1463)
leejet
authored
67dda3f
fix: handle stereo format in sd_audio (#1489)

stduhpf
and
leejet
committed
cde20d5
feat: add ltxav latent2rgb projection matrix (#1502)
stduhpf
authored
06accf2
feat: add ltx2.3 flf2v support (#1505)
leejet
authored
e43b24c
refactor: split guidance composition (#1506)
leejet
authored
50134e5
feat: add Gradient Estimation sampler (#1484)
wbruna
authored
e7eb92f
refactor: minify vocab files (#1509)
leejet
authored
baf7eda
ci: add inactive pr clean up workflow
leejet
committed
82e03ef
docs: update news
leejet
committed
8308042

Commits on May 18, 2026

ci: add CUDA Docker image support for NVIDIA Spark GB10 (#1512)
ich777
authored
21fd4e6
feat: make negative max_vram control the amount of spare vram (#1503)
wbruna
authored
f683c88
fix: only enable TAE after successful load (#1517)
leejet
authored
b758b7d
fix: restore singleton dims for LLM outputs (#1518)
leejet
authored
b706d68
sync: update ggml (#1520)
leejet
authored
22c8c40
ci: add RDNA1 + RDNA2 targets for ROCm 7.13 (#1511)
sofiageo
authored
caa823a

Commits on May 19, 2026

Commits on May 20, 2026

Commits on May 21, 2026

Commits on May 22, 2026

Commits on May 23, 2026

fix: make macOS binaries use relocatable rpaths (#1552)
leejet
authored
72e512a
feat: add Longcat-Image / Longcat-Image-Edit support (#1053)

stduhpf
and
leejet
authored
a397e03

Commits on May 25, 2026

Commits on May 26, 2026

fix: skip permission denied errors in recursive_directory_iterator (#1564)

schirik
and
Serge F. Chirik
authored
07b2b18
feat: add microsoft lens support (#1560)
leejet
authored
92dc726

Commits on May 27, 2026

Commits on May 28, 2026

fix: correct tae for models that use the flux2 vae (#1571)
wbruna
authored
0e4ee04

Commits on May 31, 2026

Commits on Jun 1, 2026

feat: support img-cfg for edit models (#929)

stduhpf
and
leejet
authored
f8935d6
refactor: call CPU backend functions dynamically (#1591)

wbruna
and
leejet
authored
02f0637
fix(cmake): build HIP backend with PIC so the static-lib PIE link succeeds (#1593)
RapidMark
authored
7948df8

Commits on Jun 2, 2026

Commits on Jun 3, 2026

refactor: img-cond->img_uncond (#1594)

stduhpf
and
leejet
authored
4513e3f
perf: keep chunk-K residency engaged with runtime LoRA (#1598)
fszontagh
authored
a7f2e03
fix: zero Wan2.2 TI2V timesteps for fixed frames (#1604)
leejet
authored
1f9ee88

Commits on Jun 4, 2026

feat(ltx2): add conversion script, ltx2.h C API, and fix CPU VAE im2col crash

64johnlee
and
claude
committed
c63beba
fix(convert): vectorised bf16→fp32 and remove unused struct import

64johnlee
and
claude
committed
1524436
ci: add LTX-2 focused CI workflow (build, convert-script, macOS Metal)

64johnlee
and
claude
committed
2f8344c
fix(convert): replace safe_open with direct safetensors parser; fix bf16_to_fp32

64johnlee
and
claude
committed
efc3854
ci: opt into Node.js 24 for actions to silence deprecation warning

64johnlee
and
claude
committed
99ab65f
test: fix M2 script to expect .avi suffix from sd-cli output

64johnlee
and
claude
committed
1ee82af
ci: trigger LTX-2 workflow on shell-script changes too
64johnlee
committed
d9812d2
fix(convert): validate tensor buffer size vs shape in _iter_safetensors

64johnlee
and
claude
committed
5b2bc6b
fix(ltx2.h): correct API mismatches against actual sd_ctx_params_t

64johnlee
and
claude
committed
49b6e33
feat(ltx2): M2 parity harness + portable smoke test + Vulkan/parity CI

64johnlee
and
claude
committed
97d9a6c
feat(ltx2): M3 — Bare native addon for LTX-2 video generation (T2V/I2V)

64johnlee
and
claude
committed
5b48da2
fix(ltx2/bare): make the M3 addon build, install and load on linux-x64

64johnlee
and
claude
committed
af2fc6e