6cdc3bba7e
switch to pillow-heif
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-26 07:49:26 +02:00
144c00efdd
Merge pull request #5060 from vladmandic/fix/queue-lock-exception-suppression
...
fix(queue): stop the queue lock from suppressing exceptions
2026-08-26 07:31:39 +02:00
cb06447426
fix(video): report a cancelled generation as 499
...
process_images swallows the interrupt assertion, so a cancel reaches the shared
video core as an empty result and was raised as 'processing failed' with 500,
leaving clients unable to tell a cancel from a crash. Mirrors the LTX path,
which already returns 499.
2026-08-26 04:28:59 +01:00
78a49eab33
fix(queue): stop the queue lock from suppressing exceptions
...
Queue.__exit__ returned _queue_lock, and a truthy __exit__ return suppresses the
exception in flight, so every `with queue_lock:` block discarded exceptions and
resumed with locals from the aborted block unassigned. The bare threading.Lock it
replaced returned None.
2026-08-26 03:36:15 +01:00
3f80bb05b0
docs(ltx): correct the refine default comment and stale file references
...
The refine-default comment claimed Condition variants are excluded from two-stage
refine, but supports_two_stage_refine has no such carve-out and ltx_process builds
a second condition set so conditioning survives the upsample. Two file references
had also drifted: ltx_process.py:179 moved, and the offload hook keying now lives
in sd_offload_balanced. Name the symbols instead of the line numbers.
2026-08-25 23:21:21 +01:00
99112c9be1
switch rocm nightlies to rocm 7.14
2026-08-25 12:34:55 +03:00
1946892b7f
improve memory reporting, explicit interpolation and minimax tracking
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-25 09:11:46 +02:00
5159e8b1a5
experimental minimax chunking
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-24 18:51:30 +02:00
1714657cf7
change minimax defaults
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-24 18:51:30 +02:00
f6777b168d
SDNQ Atten: add quantize_fp32
2026-08-24 19:34:54 +03:00
3c2236c4ed
fix ideogram4
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-24 14:57:58 +02:00
fb5584d3fe
error handle vae decode
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-24 10:26:28 +02:00
8eefb62173
Merge branch 'master' into dev
2026-08-24 09:14:02 +02:00
c2ce67418c
fix prompt enhance api
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-24 09:00:02 +02:00
0543fc185d
update kernel autotune reporting
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-23 14:08:21 +02:00
c81cfa1bb9
Merge pull request #5053 from vladmandic/dev
...
refresh master
2026-08-23 12:02:09 +02:00
9adccd6188
Merge branch 'master' into dev
2026-08-23 12:01:50 +02:00
1f1eab3b7a
update changelog and todo
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-23 11:52:42 +02:00
38324a198e
more api cleanup
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-23 11:40:53 +02:00
f4aa9d1b86
batch processing run register after script
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-23 11:34:34 +02:00
c5e29e3d7b
api schema updates
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-23 11:28:50 +02:00
cf3dfebc98
fix file logger traceback
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-23 09:59:38 +02:00
0325bf517b
Merge pull request #5049 from vladmandic/dev
...
refresh master
2026-08-22 20:10:53 +02:00
b21fb577c9
Merge branch 'master' into dev
2026-08-22 20:10:45 +02:00
e17efc19f7
update upscaler workflow
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-22 20:10:02 +02:00
2cafc60e92
Merge pull request #5048 from vladmandic/dev
...
refresh master
2026-08-22 18:12:10 +02:00
9b5cf7a73b
Merge branch 'master' into dev
2026-08-22 18:11:58 +02:00
9920004dec
update file logger and requirements
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-22 17:52:49 +02:00
2fe5049c04
Merge pull request #5047 from vladmandic/dev
...
refresh master
2026-08-22 16:15:42 +02:00
a6b6a472c9
Merge branch 'master' into dev
2026-08-22 16:15:32 +02:00
69d8f4f7c4
handle torchsde
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-22 16:08:58 +02:00
5c91f58c24
update todo/changelog
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-22 11:45:06 +02:00
f66650f913
Merge pull request #5046 from vladmandic/dev
...
fix missing export
2026-08-22 11:39:46 +02:00
25edd3387f
Merge branch 'master' into dev
2026-08-22 11:39:38 +02:00
87cb999713
fix missing export
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-22 11:38:51 +02:00
9053a3d8b9
Downgrade pre-commit CI action version to v1.1.0
2026-08-22 11:36:04 +02:00
100a4da109
Merge pull request #5045 from vladmandic/master
...
refresh dev
2026-08-22 11:30:32 +02:00
864e9a497f
Merge branch 'dev' into master
2026-08-22 11:30:18 +02:00
63a5722429
prepare queue mgmt
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-22 11:24:41 +02:00
367f863661
Update GitHub Actions to use latest action versions
2026-08-22 11:24:19 +02:00
dfc399c7af
Rename lint workflow to 'Lint Project'
2026-08-22 11:20:56 +02:00
c68d993b75
Rename workflow to 'Build GitHub Pages'
2026-08-22 11:20:31 +02:00
dcdfa5534f
Rename workflow to 'Edit Readme Sponsors'
2026-08-22 11:20:07 +02:00
2ba7f9d1c8
update js deps and submodules
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-22 09:30:49 +02:00
8cc52889a3
lint
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-21 20:35:15 +02:00
84c1ed66b3
video upscaling
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-21 20:28:44 +02:00
9dd1d73f6d
video upscaling using spandrel
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-21 16:21:12 +02:00
d239bfcde1
placeholder video upscale
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-21 10:26:17 +02:00
52f5c17435
update spandrel integation
...
Signed-off-by: Vladimir Mandic <mandic00@live.com >
2026-08-21 09:35:15 +02:00
25b7961e4e
fix(lora): refuse a network whose deltas do not fit the model
...
A delta that does not fit its target module cannot apply, and applying only
the layers that do fit leaves the model in a state nothing was trained for,
so try_load_chain drops the whole file when any family reports a mismatch.
Bias deltas were never checked against the target bias and could only surface
at apply time; a module with no bias stays a non-mismatch, since whole
architectures are built bias=False.
- check bias deltas against the module bias in the lora, norm and full loaders
- carry the mismatch count on the network so the chain can refuse the file
- record refused writes in the infotext so a partial apply is not read as clean
- point the krea2 full-diff test at a module that has a bias
2026-08-21 02:12:16 +01:00
Vladimir Mandic
CalamitousFelicitousness
Dity0