 kijai
|
0b9beb1884
|
Faster VAE decoding, fix leapfusion I2V flashing issue
|
2025-01-31 23:18:52 +02:00 |
|
|
kijai
|
72fe241ea6
|
Update pipeline_hunyuan_video.py
|
2025-01-24 17:57:35 +02:00 |
|
|
kijai
|
5d7b0e4339
|
Support leapfusion img2vid
https://github.com/AeroScripts/leapfusion-hunyuan-image2video
|
2025-01-24 17:11:16 +02:00 |
|
|
kijai
|
88823b79b7
|
Allow using custom noise
|
2025-01-19 00:41:19 +02:00 |
|
|
kijai
|
fd2c196b48
|
Update posemb_layers.py
|
2025-01-17 17:10:56 +02:00 |
|
|
kijai
|
ea1b92cfc2
|
RoPE optimizations and fix
- small overall speed boost and fix freqs dtype
- allow not upcasting RoPE for slight speed boost
|
2025-01-17 16:57:37 +02:00 |
|
|
kijai
|
11ee15a91e
|
Support VHS animated previews
|
2025-01-07 00:39:09 +02:00 |
|
|
kijai
|
4fe87a3220
|
Some more schedulers
|
2025-01-06 19:17:46 +02:00 |
|
|
kijai
|
e3a2fa7ca7
|
remove noise from preview
|
2025-01-06 12:45:17 +02:00 |
|
|
kijai
|
1d67d090da
|
Fix preview
|
2025-01-06 11:38:40 +02:00 |
|
|
kijai
|
03801b426f
|
Add dpm++ scheduler
|
2025-01-06 02:27:29 +02:00 |
|
|
kijai
|
783d111459
|
fix
|
2025-01-06 06:46:53 +02:00 |
|
|
kijai
|
6c70beda52
|
Add TeaCache
|
2025-01-06 06:26:28 +02:00 |
|
|
kijai
|
f52a49f1d8
|
Add alternative cpu offloading
Thanks to the DiffSynth team:
https://github.com/modelscope/DiffSynth-Studio/blob/main/diffsynth/models/hunyuan_video_dit.py#L630
|
2024-12-23 17:47:09 +02:00 |
|
|
kijai
|
b9ea9bf7cd
|
Add Enhance-A-Video
https://github.com/NUS-HPC-AI-Lab/Enhance-A-Video/
|
2024-12-21 17:42:07 +02:00 |
|
|
kijai
|
da2fb54468
|
remove print
|
2024-12-19 17:03:10 +02:00 |
|
|
kijai
|
b6c1705f9f
|
simple context windows with freenoise shuffling
|
2024-12-19 11:56:53 +02:00 |
|
|
kijai
|
ec0fdf1fc4
|
consistently use safe_load
|
2024-12-19 01:53:00 +02:00 |
|
|
kijai
|
fb34e91f0e
|
Support official scaled fp8 model
|
2024-12-18 10:29:56 +02:00 |
|
|
kijai
|
46f569f571
|
clearer error on frame count mismatch
|
2024-12-18 00:34:12 +02:00 |
|
|
kijai
|
8f7710a77c
|
Avoid trying to create preprocessor with encoder only model
|
2024-12-16 21:12:02 +02:00 |
|
|
Dango233
|
fbfaa8d900
|
* TextImage Encoder now a seperate node.
* No more behavior change of original nodes
* Clean up
|
2024-12-15 14:14:28 +00:00 |
|
|
Dango233
|
a3fbc3df80
|
IP2V - Experimental implementation based on Kijai Nodes, Initial Commit
|
2024-12-15 09:02:33 +00:00 |
|
|
kijai
|
ecae5f270e
|
small fixes
|
2024-12-14 22:28:02 +02:00 |
|
|
kijai
|
4c5507ffc6
|
Add fp8 as quant choice for the LLM, fix using custom clip_l with cfg
|
2024-12-14 16:36:25 +02:00 |
|
|
kijai
|
4a81c5ec94
|
Add CFG as optional node
experimental, using CFG halves the inference speed but allows using negative prompt, possibly more useful when using LoRAs
|
2024-12-13 16:59:49 +02:00 |
|
|
kijai
|
706c3a07ef
|
Add nodes for text embed saving/loading
This way you can encode prompts in separate workflow to avoid ever having to load the text encoders for sampling if you're low on RAM
|
2024-12-13 13:07:34 +02:00 |
|
|
kijai
|
6ab3d0ae62
|
initial RF-inversion for testing
|
2024-12-11 16:27:10 +02:00 |
|
|
kijai
|
029636d2a7
|
Update pipeline_hunyuan_video.py
|
2024-12-10 09:41:35 +02:00 |
|
|
kijai
|
e7212f184f
|
VAE encode fixes
|
2024-12-09 21:03:16 +02:00 |
|
|
kijai
|
a01b0796be
|
VAE fixes and more options
|
2024-12-09 19:20:45 +02:00 |
|
|
kijai
|
fb00591ced
|
reset scheduler begin_index after using denoise on another sampler
|
2024-12-09 14:12:31 +02:00 |
|
|
kijai
|
9a4abbc535
|
Allow using comfy clip_l, fix fp8 fastmode mem use
|
2024-12-08 20:34:31 +02:00 |
|
|
kijai
|
ec00443395
|
Allow using custom prompt_templates
|
2024-12-08 17:08:06 +02:00 |
|
|
kijai
|
62d22a7ba9
|
support single image gen
|
2024-12-08 12:27:18 +02:00 |
|
|
kijai
|
d49a711cd8
|
Update models.py
|
2024-12-08 02:00:41 +02:00 |
|
|
kijai
|
f5d80350c0
|
remove prints
|
2024-12-08 01:47:06 +02:00 |
|
|
kijai
|
e9e058436e
|
block_swap tweaks
|
2024-12-08 01:44:16 +02:00 |
|
|
kijai
|
c64db95155
|
Update pipeline_hunyuan_video.py
|
2024-12-07 22:33:00 +02:00 |
|
|
kijai
|
929781949e
|
Add STG start/end percent
|
2024-12-07 21:51:37 +02:00 |
|
|
kijai
|
6655642ea6
|
add bitsandbytes nf4 quant as option for the llm
|
2024-12-06 18:58:02 +02:00 |
|
|
kijai
|
0cbcc20a89
|
initial STG testing
|
2024-12-06 17:09:56 +02:00 |
|
|
kijai
|
2e8f1f2fbd
|
fix flash_attn, make sdpa default
|
2024-12-06 11:12:10 +02:00 |
|
|
kijai
|
31950d0b55
|
possibly fix up sdpa
|
2024-12-06 03:15:02 +02:00 |
|
|
kijai
|
8e680e9fcb
|
Update pipeline_hunyuan_video.py
|
2024-12-04 13:17:39 +02:00 |
|
|
kijai
|
3197107b5e
|
fixup block_swap start
|
2024-12-04 03:53:51 +02:00 |
|
|
kijai
|
1073ef9156
|
custom block_swap to reduce VRAM use
thanks to 2kpr for the initial block swap code!
|
2024-12-04 02:13:48 +02:00 |
|
|
kijai
|
c81c67c5a3
|
expose flow_shift, add vid2vid
|
2024-12-04 00:55:53 +02:00 |
|
|
kijai
|
e834402b51
|
vae encode
|
2024-12-03 23:54:36 +02:00 |
|
|
kijai
|
8a4b535e49
|
expose vae temporal_tiling_sample_size
|
2024-12-03 19:38:03 +02:00 |
|