Morph MORPH
SIGN IN SIGN UP
dhruvhead / stable-diffusion.cpp UNCLAIMED
0 0 108 C++

COMMITS

master
September 10, 2026
H
fix: enable VAE decode tiling fallback without auto-fit (#1932)
Hmission committed
b68d586
L
refactor: split generation pipeline out of stable-diffusion.cpp (#1957)
leejet committed
14eddb3
S
docs: reflect GGML_MAX_NAME value change in rpc docs (and in ggml_extend assert) (#1950)
stduhpf committed
469fc49
L
refactor: unify model source and weight lifecycle management (#1956)
leejet committed
6b47fec
September 7, 2026
W
fix: correct SDXL embeddings loading (#1939)
Wagner Bruna committed
d04e895
N
fix: preserve K-quantized embedding weights (#1936)
nan committed
9cdb6b6
L
refactor: split ggml extensions and move implementations to cpp files (#1945)
leejet committed
31ab2b2
September 6, 2026
L
fix: reuse graph cut plans across CFG passes (#1943)
leejet committed
d8fb10c
L
feat: enable single-GPU auto-fit with tiered parameter placement (#1942)
leejet committed
80bac2d
L
feat: add verbose logging and log-level selection (#1941)
leejet committed
dbb6112
L
refactor: unify runner lifecycles and weight residency (#1940)
leejet committed
462d675
A
feat: prefetch streamed layers during compute (#1905)
assouan committed
6c57cc3
August 30, 2026
L
feat: generalize temporal tiling across video VAEs (#1926)
leejet committed
6b3edaa
L
fix: use carrier sampling for MiniMax H3 audio (#1924)
leejet committed
40e605f
V
feat: support numbering for preview images (#1895)
vmobilis committed
9029655
V
feat: additional `--preview-interval` values (#1915)
vmobilis committed
d9b6e27
L
sync: update ggml
leejet committed
134c821
L
feat: use backend-native FP8 matmul when supported (#1916)
leejet committed
2540a4f
J
fix: correct MiniMax H3 audio Euler steps (#1908)
jk212h20 committed
dc4000d
J
fix: correct MiniMax H3 reference audio encoding (#1886)
jk212h20 committed
c797899
P
feat: add LTX-2.5 support (#1893)
Piotr Wilkin (ilintar) committed
afd5306
L
fix: match exact weights in LLM config detection (#1923)
leejet committed
2c92949
August 27, 2026
L
feat: load scaled FP8 weights without upfront conversion (#1913)
leejet committed
be0e344
August 25, 2026
L
feat: support standard Qwen3-VL weights for MiniMax-H3 (#1910)
leejet committed
50d6405
V
feat: configurable image / video compression (#1909)
vmobilis committed
0a565f2
V
fix: use sd_get_preview_interval() (#1907)
vmobilis committed
12ee60d
August 19, 2026
L
chore: format code
leejet committed
97d2990
V
fix: guard against missing sampler/scheduler names (#1887)
vmobilis committed
16304cc
V
fix: make max_order of lms sampler configurable (#1885)
vmobilis committed
760717a
W
fix: mark graph cuts with both a prefix and a suffix (#1883)
Wagner Bruna committed
88b044b