asagi4
054134b5d5
Anima Attention Couple (VERY EXPERIMENTAL, SEE README)
...
Code stolen and adapted from ppm's node pack again.
I probably introduced bugs.
2026-05-12 20:07:37 +03:00
asagi4
24240aac2d
Make github tests work again
2026-02-04 16:39:50 +02:00
asagi4
49b761ae5b
Working importing for v3 nodes
2026-02-04 16:39:50 +02:00
asagi4
d009bb782b
v3: nodes_hooks.py
2026-02-04 16:39:50 +02:00
asagi4
eaf5fa20c4
v3: nodes_base.py
2026-02-04 16:39:50 +02:00
asagi4
2becb5cdc6
Initial v3 migration
2026-02-04 16:39:50 +02:00
asagi4
805ffbf463
Use pytest tests
2026-02-04 16:39:50 +02:00
asagi4
d14f5b789a
Convert tests to pytest
...
Not 100% sure these fully work yet
2026-02-04 16:39:50 +02:00
asagi4
deba0a0642
Ruff fixes etc.
2026-02-04 16:39:50 +02:00
asagi4
1c50ae5297
Disable the cache hack for now
2025-08-19 19:53:23 +03:00
asagi4
cea1e5b30f
Initial embedded documentation
2025-06-21 13:15:58 +03:00
Pam
e4f64837ef
Add PCAttentionCoupleBatchNegative;
...
Revert some optimizations in AttentionCoupleHook
2025-06-03 03:17:47 +05:00
asagi4
c4ac37333d
Switch Attention Couple implementation to one based on ppm
...
Also removes compatibility code with older ComfyUI
2025-05-26 20:11:36 +03:00
asagi4
3745bc6879
doc and reformat
2025-01-12 15:56:25 +02:00
asagi4
99c815a3b6
Mechanism for using attention masks via PCTextEncode and a Hook node
...
This is experimental and may still change
2025-01-12 15:56:01 +02:00
asagi4
f7d78e54d5
Change logging format
2024-12-15 15:34:32 +02:00
asagi4
25990cb17e
Cache hack for performance
...
Set PROMPTCONTROL_ENABLE_CACHE_HACK=1 in your environment to enable
2024-12-15 15:34:27 +02:00
asagi4
2f5d62b46b
Tag a non-broken release
2024-12-13 20:24:00 +02:00
asagi4
e64a71fc6a
Make the description a bit less terse.
2024-12-13 19:38:34 +02:00
asagi4
83f859bc36
Refactoring
2024-12-12 21:57:05 +02:00
asagi4
2fae4c0bc8
Add nodes for attaching masks to a CLIP instead of a separate schedule object
2024-12-12 20:18:40 +02:00
asagi4
192f7e3efd
Disable schedule nodes for now
2024-12-12 19:31:08 +02:00
asagi4
eab2cc09dd
Change types to mark them as explicitly incompatible
2024-12-11 22:43:21 +02:00
asagi4
de1c39a74a
Reorganize nodes
2024-12-11 22:36:30 +02:00
asagi4
7bfd6790df
Clean up legacy stuff
2024-12-11 21:50:11 +02:00
asagi4
da502219ad
Documentation reorganization
2024-12-11 17:08:51 +02:00
asagi4
df72d2c478
Add some display name mappings etc.
2024-12-11 17:08:51 +02:00
asagi4
13696a11b7
Add experimental PCEncodeLazy
2024-12-11 17:08:47 +02:00
asagi4
58fe45eb87
Add LoRA loading optimization. I don't think I like this interface though...
2024-12-07 21:11:38 +02:00
asagi4
4ea73cf4ec
Fix imports
2024-12-06 14:30:38 +02:00
asagi4
2e60c904c8
PCEncodeSingle node for encoding a single prompt
2024-12-05 23:55:04 +02:00
asagi4
e4254828f5
Handle old ComfyUI versions
2024-12-05 21:48:36 +02:00
asagi4
751af8cabb
Initial nodes using the new hooks mechanism recently merged
2024-12-03 14:05:54 +02:00
asagi4
81cced3776
IMASK and PCScheduleAddMasks
...
See #44
2024-05-23 18:48:00 +03:00
asagi4
e8cb0a853b
Node for wrapping guiders, usable with SamplerCustomAdvanced
...
Very lightly tested, so undocumented for now.
2024-04-16 00:07:02 +03:00
asagi4
a7cd6462a1
Fix some pylint issues
2024-02-18 18:18:12 +02:00
asagi4
373d3e9d1a
PCPromptFromSchedule, fixes ##28
2024-02-13 21:42:16 +02:00
asagi4
cf65324df7
Nodes for setting default parameters for SDXL() and mask sizes
...
in the schedule.
This integrates better with other nodes that calculate sizes.
2024-01-31 20:14:26 +02:00
asagi4
8953752088
Make logging a bit nicer
2024-01-09 02:56:34 +02:00
asagi4
d49ba9caa1
Forgot to remove nodes from __init__.py
2023-12-06 18:48:40 +02:00
asagi4
ec67906f01
Use a separate node to toggle split scheduling
2023-09-03 19:29:20 +03:00
asagi4
78ff0aad55
Add a simple wildcard node
...
Not documenting yet, as it doesn't save its state in metadata.
Also improved AIO node performance a bit
2023-08-09 22:17:52 +03:00
asagi4
816dcc9fdc
Add experimental node for linear interpolation between conds.
...
I have no idea if the math is correct for anything other than
the simplest cases, but it seems to work.
The harder part seems to be figuring out how to make it part of
prompt scheduling without breaking everything.
2023-08-04 20:31:23 +03:00
asagi4
49c439a7d7
Add a simple all-in-one combo node
2023-08-04 02:45:19 +03:00
asagi4
b558473d28
Add FilterSchedule
2023-08-01 01:52:27 +03:00
asagi4
29262e9c6f
Restructure and add new nodes
2023-07-31 21:04:21 +03:00
asagi4
0ef52350e1
Tag filtering
2023-07-30 20:48:43 +03:00
asagi4
8826ed3051
Shuffle code around
2023-07-30 16:02:07 +03:00
asagi4
c1a08f31af
Add ConditioningCuitoff node
2023-07-28 07:15:27 +03:00
asagi4
4ae97bea43
Reformat code
2023-07-28 01:45:41 +03:00
asagi4