conflicted merge by Geramy · Pull Request #4 · lemonade-sdk/stable-diffusion.cpp

Geramy · 2026-05-06T20:10:50Z

No description provided.

Co-authored-by: leejet <leejet714@gmail.com>

* inpaint: get max pixel max instead of single sample * inpaint: masked diffusion for inpainting models with inflated mask * refactor tensor interpolate nearest-like reduction paths and generalize max_pool_2d --------- Co-authored-by: leejet <leejet714@gmail.com>

…1372)

Co-authored-by: professor-moody <keys@nimbus.lan>

…tions (#1423)

…#1437)

)

…nal SLG (#593)

ci: revert CUDA artifact names to original sd-*-bin-* format

Co-authored-by: leejet <leejet714@gmail.com>

…ceeds (#1593)

…ion (#1576)

* refactor: img-cond->img_uncond * align APG and CFG++ with img-uncond CFG * set default img_cfg to 1.f --------- Co-authored-by: leejet <leejet714@gmail.com>

…uffer (#1601)

Signed-off-by: kei-g <km.8k6ce+github@gmail.com>

…1611)

wbruna and others added 30 commits April 5, 2026 23:52

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

687a81f

Co-authored-by: leejet <leejet714@gmail.com>

feat: add webm support (#1391)

7397dda

refactor: apply RAII ownership to examples (#1392)

359eb8b

refactor: split examples common into header and source (#1393)

5bf438d

chore: normalize text files to utf-8 without bom (#1394)

8afbeb6

fix: correct double increment on flow denoisers sigma calculations (#…

dd75372

…1372)

feat: add flux2 small decoder support (#1402)

e8323ca

refactor: simplify DiscreteFlowDenoiser (#1405)

be9f51b

chore: harden safetensors and gguf loading code (#1404)

118489e

Co-authored-by: professor-moody <keys@nimbus.lan>

feat: add sdcpp api support (#1407)

7ade90e

feat: use sdcpp-webui as embedded webui (#1408)

fd35047

docs: update readme

12a369c

docs: update readme

6b675a5

chore: allow building the embedded UI header separately (#1415)

ee5bf95

refactor: introduce shared tokenizer abstraction and split implementa…

9ac7b67

…tions (#1423)

feat: add left padding support to tokenizers (#1424)

c41c5de

feat: add ernie image support (#1427)

5c243db

feat: SDXS-09 support and update doc (#1356)

d73b419

feat: add er_sde sampler (#1403)

1b4e9be

fix: skip empty prompt segments around attention range (#1429)

84fc544

refactor: remove is_xl guard wrapper in get_sd_version (#1430)

a564fdf

fix: correct dpm++2s_a second model call (#1435)

2bcff67

fix: tune ernie-image default flow shift (#1433)

6a9cb31

feat: add DPM++ (2S) Ancestral implementation for flow models (#1428)

f3f69e2

feat(server): implement vid_gen async API and mode-aware capabilities (…

4d626d2

…#1437)

ci: skip docker image build job on pull requests (#1439)

3c99f70

chore: enable MSVC parallel compilation with /MP (#1438)

7d33d4b

feat: adapt LCM for flow models (#1413)

e77e4c4

fix: correct image to image DDIM and TCD (#1410)

7023fc4

wbruna and others added 7 commits May 30, 2026 18:31

fix: explicitly exclude f8, f64 and i64 tensors from mmap (#1575)

b54bd83

fix: prevent crash in case of a mem alloc error and graceful exit (#1566

b4ba55d

)

fix: split tokens before normalization (#1582)

d3b2cb0

fix: correct Gemma3 rope settings and vram limit propagation (#1583)

d2797b8

feat: add PiD support (#1585)

0982807

fix: remove kv padding from flash attention wrapper (#1453)

20901f6

feat: add support for APG (adaptive projected guidance) + uncondition…

be65ac7

…nal SLG (#593)

kenvandine pushed a commit that referenced this pull request Jun 1, 2026

Merge pull request #4 from Phqen1x/revert-cuda-artifact-names

2791d9d

ci: revert CUDA artifact names to original sd-*-bin-* format

stduhpf and others added 22 commits June 1, 2026 22:54

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

f8935d6

Co-authored-by: leejet <leejet714@gmail.com>

refactor: call CPU backend functions dynamically (#1591)

02f0637

Co-authored-by: leejet <leejet714@gmail.com>

fix(cmake): build HIP backend with PIC so the static-lib PIE link suc…

7948df8

…ceeds (#1593)

feat: --stream-layers for streaming weights from CPU during generat…

ed74577

…ion (#1576)

chore: embed server web UI in Docker images (#1597)

9c7f9a2

feat: make Wan2.2 5B FLF2V work (#1110)

2d40a8b

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

4513e3f

* refactor: img-cond->img_uncond * align APG and CFG++ with img-uncond CFG * set default img_cfg to 1.f --------- Co-authored-by: leejet <leejet714@gmail.com>

perf: keep chunk-K residency engaged with runtime LoRA (#1598)

a7f2e03

fix: zero Wan2.2 TI2V timesteps for fixed frames (#1604)

1f9ee88

perf: allocate CPU-offloaded params from runtime device pinned host b…

064001b

…uffer (#1601)

fix: Suppress spurious error output for --help (#1607) (#1608)

74f513d

Signed-off-by: kei-g <km.8k6ce+github@gmail.com>

perf: ratchet streaming budget so plan stops re-merging every step (#…

0648f44

…1611)

feat: add ideogram4 support (#1609)

b9254dd

refactor: unify model config detection (#1613)

cfbc19d

refactor: extract Wan VAE implementation (#1614)

dfb2390

refactor: reorganize src model layout (#1615)

f3fd359

chore: rename and reformat gits_noise.inl (#1617)

81abfb2

refactor: move photomaker into generation extension (#1618)

2a07540

refactor: split model loader from model definitions (#1619)

b3d56d0

perf: cap planner budget when model dwarfs the streaming budget (#1612)

17a2b4a

apg: normalize diff_norm calculation by tensor size (#1620)

138da14

feat: set tensor names on block params (#1622)

19bdfe2

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

conflicted merge#4

conflicted merge#4
Geramy wants to merge 133 commits into
lemonade-sdk:lemonadefrom
leejet:master

Geramy commented May 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

20 participants

Conversation

Geramy commented May 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

20 participants