 kijai
|
e3afc7fc75
|
Add helper node for cfg scheduling and allow scheduling audio_cfg as well
|
2025-04-29 09:00:01 +03:00 |
|
|
kijai
|
1b3592f6b4
|
Update nodes.py
|
2025-04-29 07:34:36 +03:00 |
|
|
kijai
|
fe25d21fba
|
Fix audio cfg
silly mistake
|
2025-04-29 02:07:05 +03:00 |
|
|
kijai
|
e334f78124
|
Update nodes.py
|
2025-04-28 21:50:57 +03:00 |
|
|
kijai
|
c8a084d27a
|
Update nodes.py
|
2025-04-28 21:37:48 +03:00 |
|
|
kijai
|
8117c6f033
|
Context windows with FantasyTalking
|
2025-04-28 20:24:41 +03:00 |
|
|
kijai
|
5683d8306a
|
Support FantasyTalking
|
2025-04-28 19:10:32 +03:00 |
|
|
kijai
|
fc7ab666a2
|
support loading Fun camera model
Input not functional yet
|
2025-04-25 17:02:10 +03:00 |
|
|
kijai
|
e75e6d6933
|
Update nodes.py
|
2025-04-25 16:19:52 +03:00 |
|
|
kijai
|
031d8ac817
|
disable enhance always with DF
|
2025-04-25 15:47:25 +03:00 |
|
|
kijai
|
33b8f1c082
|
Support Fun control 1.1
Supports the new reference image input to the 1.1 control model
|
2025-04-25 15:44:18 +03:00 |
|
|
kijai
|
863829d083
|
dtype fixes and maybe allow fp8_fast to work
Unsure when this has changed but seems that _scaled_mm now works with same dtype? This fixes the quality degradation of fp8_fast in initial tests.
|
2025-04-25 03:12:10 +03:00 |
|
|
kijai
|
96a2172e13
|
Phantom + VACE testing
|
2025-04-23 20:34:28 +03:00 |
|
|
kijai
|
4602b9c885
|
Fix phantom start percent
|
2025-04-23 10:30:33 +03:00 |
|
|
kijai
|
e0e5fcf713
|
Add start/end percent for Phantom and other small fixes
|
2025-04-23 01:00:41 +03:00 |
|
|
kijai
|
24876cffc1
|
Update nodes.py
|
2025-04-22 22:34:45 +03:00 |
|
|
kijai
|
6fadcbd957
|
Support Phantom, refactor model dtypes, reduce DF model memory use
|
2025-04-22 21:13:27 +03:00 |
|
|
kijai
|
65f5505fca
|
Initial Skyreels DiffusionForcing model support
On it's own sampler at least for now while I figure out how it all works.
|
2025-04-21 18:13:51 +03:00 |
|
|
kijai
|
e5a326c981
|
Use penultimate_hidden_states instead when using native clip vision
The new Skyreels 1.3B I2V did not work otherwise, this seems to make it match the wrapper clip vision output better
|
2025-04-21 01:33:51 +03:00 |
|
|
kijai
|
18aa47cc74
|
Update nodes.py
|
2025-04-19 22:41:50 +03:00 |
|
|
kijai
|
20088d0fbf
|
Update nodes.py
|
2025-04-19 22:38:19 +03:00 |
|
|
kijai
|
3692f3580a
|
Update nodes.py
|
2025-04-19 21:15:24 +03:00 |
|
|
kijai
|
7b6c34e26d
|
Add options what to draw on dwpose detector
|
2025-04-19 19:47:09 +03:00 |
|
|
kijai
|
b9b8ec7bf8
|
Apply Fresca for cfg 1.0 as well
|
2025-04-19 17:50:17 +03:00 |
|
|
kijai
|
2cfae9fa8f
|
Make UniAnimate ref pose optional
|
2025-04-19 17:41:04 +03:00 |
|
|
kijai
|
90c9415d31
|
Add lcm scheduler, pad/truncate UniAnim poses if pose count doesn't match latent count
|
2025-04-19 17:24:18 +03:00 |
|
|
kijai
|
421e375f13
|
Strength controls for Unianimate
|
2025-04-18 23:06:49 +03:00 |
|
|
kijai
|
6d521a9f4e
|
Update nodes.py
|
2025-04-18 19:45:04 +03:00 |
|
|
kijai
|
7462356743
|
Support UniAnimate-DiT
https://github.com/ali-vilab/UniAnimate-DiT
|
2025-04-18 19:43:40 +03:00 |
|
|
kijai
|
00de4f5e0e
|
Detect (a guess for now) TeaCache coefficients for SkyreelsV2
|
2025-04-18 11:47:19 +03:00 |
|
|
kijai
|
761d188dba
|
Update nodes.py
|
2025-04-16 23:07:16 +03:00 |
|
|
kijai
|
7c81d7ce10
|
Better support for the official FirstLastFrame2Video -model (FLF2V)
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan2_1-FLF2V-14B-720P_fp16.safetensors
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan2_1-FLF2V-14B-720P_fp8_e4m3fn.safetensors
|
2025-04-16 19:40:04 +03:00 |
|
|
kijai
|
f919c4fd3d
|
Allow using custom sigmas, fix some cases with multiple VACE embeds
|
2025-04-13 16:35:40 +03:00 |
|
|
kijai
|
941acc434d
|
Add some schedulers
|
2025-04-12 17:18:53 +03:00 |
|
|
kijai
|
ea71dfda1a
|
restructure, add visualization node
|
2025-04-11 18:40:50 +03:00 |
|
|
kijai
|
94f59f09c8
|
map camera types
|
2025-04-10 23:27:36 +03:00 |
|
|
kijai
|
740fdd749a
|
initial recammaster support
|
2025-04-10 23:04:32 +03:00 |
|
|
kijai
|
6f2f2db9a5
|
Force some image inputs to be divisible by 16
|
2025-04-09 16:49:16 +03:00 |
|
|
kijai
|
f3f12aa6cf
|
Fix batched_cfg and some possible dtype tweaks to avoid recompiles
|
2025-04-08 20:08:33 +03:00 |
|
|
kijai
|
7de8ac25a6
|
Some VACE tweaks
|
2025-04-08 18:27:06 +03:00 |
|
|
kijai
|
96633b5101
|
Remove fp16 from the wrapper T5 loader as it ruins results.
|
2025-04-07 22:47:59 +03:00 |
|
|
kijai
|
7d23826136
|
tiled encoding fixes and add as option for VACE endore
|
2025-04-07 20:21:36 +03:00 |
|
|
kijai
|
f8155ab47f
|
Allow loading Fun Reward LoRAs
https://huggingface.co/alibaba-pai/Wan2.1-Fun-Reward-LoRAs/tree/main
|
2025-04-07 12:38:56 +03:00 |
|
|
kijai
|
34ee8efb9a
|
Allow more than 2 clip embeds
|
2025-04-05 18:56:28 +03:00 |
|
|
kijai
|
56a4221a38
|
Add extra latent to front instead (as it probably should)
|
2025-04-05 16:18:37 +03:00 |
|
|
kijai
|
36249e1a97
|
Basic Skyreels A2 support
converted model here:
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan2_1_SkyreelsA2_fp8_e4m3fn.safetensors
Encode the reference and input to the ImageToVideoEncode -node, take into account the extra latent when choosing frame count (should be -4 to normal), otherwise the standard I2V workflow.
|
2025-04-05 15:53:56 +03:00 |
|
|
kijai
|
ae40c3c131
|
Update nodes.py
|
2025-04-05 15:18:36 +03:00 |
|
|
kijai
|
d6a6629c08
|
Update nodes.py
|
2025-04-04 18:06:33 +03:00 |
|
|
kijai
|
2ed2c5cbc1
|
Fix non-vace (again)
|
2025-04-04 17:50:52 +03:00 |
|
|
kijai
|
04534e5eed
|
Allow using multiple VACE embeds, fix VACE blockswap error when vace blocks not set and blockswap is set
Experimental
|
2025-04-04 15:23:39 +03:00 |
|