c8c24f1033
fix regex and output strings
2023-01-24 16:48:38 +03:00
c10c98c894
add use_noise_mask to the args
2023-01-24 16:42:15 +03:00
ae37a48ca3
better text v2
2023-01-24 12:16:06 +02:00
b68f9d6d4f
better text
2023-01-24 12:09:36 +02:00
8ae5cc0c3f
Final fixes + option to keep interp frames
2023-01-24 12:04:13 +02:00
93ab7f3db3
add missing params to settings.py
2023-01-24 11:18:20 +02:00
f381897174
remove old comments
2023-01-24 11:06:23 +02:00
bfe90b8562
Merge branch 'frame_interpolation' of https://github.com/hithereai/deforum-for-automatic1111-webui into frame_interpolation
2023-01-24 10:50:59 +02:00
c65df56220
final touches
2023-01-24 10:50:55 +02:00
e6c54e4a5d
Delete ffmpeg_replacer.py
2023-01-24 02:59:02 +02:00
245cf87188
global controls for ffmpeg crf and preset + cleanup p1
2023-01-24 02:57:51 +02:00
6c36adc9b8
almost ready, add cli-ffmpeg calls that work
2023-01-24 01:04:37 +02:00
0ac8bd32fa
tabs -> spaces fixup
2023-01-23 22:15:54 +03:00
32d3b6fa38
add the actual schedule to the args
2023-01-23 22:15:15 +03:00
f417cc33ff
consistent use of quotes
2023-01-23 09:37:03 -08:00
c52751f5c0
add checkpoint schedule
2023-01-23 09:24:05 -08:00
0f99d791c9
add 'everywhere' to usable vars
2023-01-23 19:48:24 +03:00
aa72ee7bb4
var -> match_object
2023-01-22 20:10:09 +03:00
e874b451bd
fixup imports
2023-01-22 15:18:05 +03:00
e2d13eb04b
Update ffmpeg_replacer.py
2023-01-22 13:33:05 +02:00
7b9abc66cd
PyAV ffmpeg usage to save interp vids
2023-01-22 04:27:57 +02:00
3c111cafb0
Update inference_video.py
2023-01-21 15:57:06 +02:00
8ed69c2430
change logic!
...
change logic to save pngs instead of using cv2 to directly save the video cuz it cannot do proper h2654 with opencv-python.
First step out of a few prolly.
2023-01-21 15:56:47 +02:00
7d617f3faf
dynamic model loading fixes
2023-01-20 21:26:14 +02:00
0a6960ff2b
better UI-text and add hints
2023-01-20 20:48:06 +02:00
78b5639227
better folder name
2023-01-20 20:34:02 +02:00
96635be7a5
Delete inference_img.py
2023-01-20 20:29:02 +02:00
8ff265f7db
remove print statement
2023-01-20 20:27:52 +02:00
ed20c17342
move license to correct file
2023-01-20 20:26:55 +02:00
7fe85a81ca
less files more happy, add rife 4.0, revert args
2023-01-20 20:23:55 +02:00
94aa4052c7
tiny fixes
2023-01-20 18:04:39 +02:00
46e02c063f
Delete README.md
2023-01-20 17:31:38 +02:00
4a53e275aa
Working properly v1
2023-01-20 17:29:38 +02:00
60363b555c
still not 100% there, getting closer
2023-01-19 10:25:59 +02:00
6517100c4d
Merge branch 'frame_interpolation' of https://github.com/hithereai/deforum-for-automatic1111-webui into frame_interpolation
2023-01-18 02:00:42 +02:00
2c30e1ae93
Getting there, not ready yet, add rife 4.3
2023-01-18 02:00:36 +02:00
4b5c44e4a6
Merge branch 'deforum-art:automatic1111-webui' into frame_interpolation
2023-01-17 18:17:23 +02:00
0a91f4db11
Initial GUI Accordion
2023-01-17 18:16:41 +02:00
39c2c6f1f6
make it actually work on autos env
2023-01-17 16:55:27 +02:00
c73a80b62d
Add Practical-RIFE files to src
2023-01-17 13:48:52 +02:00
a51c22c553
remove references to p.sd_model
...
it was removed as a param in e0e8005009Closes #228
2023-01-17 01:48:30 +03:00
6777e0d927
fix set difference
2023-01-16 20:14:23 +03:00
ebe657f630
Merge branch 'automatic1111-webui' into masking-overhaul
2023-01-16 19:43:31 +03:00
9ba8462efd
add defaults so the var keys are always found
2023-01-16 19:31:52 +03:00
446e3430a6
Fixes, -1 as default for unlimited frames and more checks
2023-01-16 18:12:14 +02:00
afa80cd231
more mask format conversion checks
2023-01-16 18:52:39 +03:00
19d106fc9c
convert all input masks to binary
2023-01-16 18:31:24 +03:00
4b311bf874
none to blank to simplity compose logic
2023-01-16 18:28:00 +03:00
32df52f511
Merge branch 'automatic1111-webui' into masking-overhaul
2023-01-16 18:25:01 +03:00
585b2e2051
mask schedule logic
2023-01-16 18:24:46 +03:00
9e0ae9ef26
add extract_from_frame and extract_to_frame to input vid options
...
The current code will always extract all of the input video frames.
This PR adds two new input parameters for extracting an input video:
extract from frame, and extract to frame.
Now we can only test frames 0 to 50, 50 to 100, etc.
2023-01-16 00:56:16 +02:00
c9f41faf3e
use clipseg inference to get word masks
2023-01-16 00:24:43 +03:00
898a261f32
Minor improvements +
...
better handling of wrong file/folder names, added function to delete all imgs from a folder (shorter code); now users can generate only the masks again even without having the 'generate_hybrid_frames' button selected
2023-01-15 22:39:47 +02:00
8d09bc7ddf
upload clipseg directly to src
...
as the current webui's transformers version (4.19) doesn't include it
2023-01-15 18:18:59 +03:00
31df9ed541
add parser for multiple masks boolean combinations
2023-01-15 17:40:14 +03:00
a1d5ed7752
Merge pull request #216 from hithereai/HumanMasking
...
Masking Humans automatically using RVM
2023-01-15 17:36:20 +03:00
1336e84e10
revert pathlib removal
2023-01-15 03:24:53 +02:00
fcafc74c1d
enable offline usage for 3D animation mode
...
It will fetch the model from cache if it's there, so only need to run 3D mode online once for it to work offline afterwards.
* Currently it will work only for windows users.
2023-01-15 03:13:15 +02:00
ecd35261b5
remove pathlib import
2023-01-15 02:53:16 +02:00
891b682c49
enable offline usage after initial run
...
This PR enables to use the new human_mask feature while offline. It only needs to be run once when online, then it will use its cache as first priority.
2023-01-15 02:51:36 +02:00
269d8011e6
fixed setting load/save and comments
2023-01-14 23:35:53 +02:00
e09eecfeb3
Update human_masking.py
2023-01-14 23:13:57 +02:00
fc1faadaae
not needed, being loaded to cache with torch.load
2023-01-14 23:12:26 +02:00
5abe55d1d4
Sync fps according to extract_nth_frame and more
2023-01-14 23:10:08 +02:00
18c89806c2
revert model to mobilenetv3
2023-01-14 17:19:11 +02:00
c9fe53b75f
use extracted input imgs instead of input vid
2023-01-14 16:55:01 +02:00
690deb04a9
Add RobustVideoMatting repo for human masking
2023-01-14 16:08:05 +02:00
5d5287d6cf
V01 of human masking
2023-01-14 15:49:52 +02:00
9636a7510b
Make parseq adapter directly search the JSON data for a likely match if an unrecognised series is requested. This ensures that, when new schedules are added to Deforum, only the Parseq UI needs to be updated to support them (and this can be improved in the future with support for custom fields).
2023-01-12 21:27:54 +11:00
7fc51359a8
add 'Init Video' soundtrack option
2023-01-09 22:43:48 +03:00
b4c764a90a
added missing imports for interpolation mode in prompt.py
...
added imports for 'pandas' and 'numpy'
2023-01-07 23:49:26 +02:00
aa9bffded9
lower default antiblur amount a bit
2023-01-07 13:54:03 +03:00
0b9ac68ce4
rounds steps to int to fix error
2023-01-06 18:26:26 +03:00
23657be814
removed untested code
2023-01-05 21:12:46 -06:00
8cf083b6ae
added optional step scheduling
2023-01-05 21:09:20 -06:00
2770a1f423
fixed typo
2023-01-05 11:01:30 -06:00
12bf30d9b5
remove file duplicates
2023-01-05 19:01:53 +03:00
f2d0bba323
Merge branch 'store-frames-in-ram' into automatic1111-webui
2023-01-05 18:51:50 +03:00
f4dce23721
Merge branch 'pr/204' into automatic1111-webui
2023-01-05 16:42:20 +03:00
d1715bcbd4
Merge pull request #198 from reallybigname/automatic1111-webui
...
Hybrid Video Compositing
2023-01-05 16:05:33 +03:00
beffcbeed6
fixed merge errors, added sampler scheduling
2023-01-04 21:49:16 -06:00
669401c059
reverted and resolved merge errors.
2023-01-04 21:48:27 -06:00
a739beb0e2
Merge branch 'automatic1111-webui' into Kitchen_add_sampler_scheduling
2023-01-04 21:25:13 -06:00
73e2ad6b56
added checkbox to enable/disable sampler scheduling
...
sampler names are now case insensitive
2023-01-04 21:16:44 -06:00
2b24935fe4
added schedule sampling
2023-01-04 20:16:34 -06:00
cc52c96a61
Hybrid Video Compositing - 3D working 2D broken
...
First commit. 3D mode works
2023-01-04 06:35:51 -08:00
9a43333267
Ensure parseq settings are included in the {timestamp}_settings.txt that is generated with a render beings. If the Parseq manifest is fectched from URL, include a summary so that you can always see exactly which parseq prompts and keyframes were used, even if the URL content changes.
2023-01-04 11:42:23 +11:00
8ec6e7755a
fixup gc import
2023-01-03 03:36:42 +03:00
354afa97bd
add gc collect to reset frames
2023-01-03 01:51:23 +03:00
b8f5b2eb2b
option to store frames in RAM instead of disc
...
for faster animation making on fast gpus at the cost of increased ram usage
2023-01-03 01:12:22 +03:00
b6cf3d110b
fix broken math support
2023-01-02 20:24:18 +03:00
ec52a1c6a2
Add files via upload
2023-01-01 13:00:30 -07:00
485d41a6cd
Merge remote-tracking branch 'upstream/automatic1111-webui' into frame-insertion
2022-12-31 21:13:21 -08:00
64b1b5c591
Getting a basic ui element added
2022-12-31 18:22:48 -08:00
5edb2c4e8c
Add files via upload
2022-12-31 14:02:50 -07:00
c6d018c830
Add files via upload
2022-12-31 14:02:05 -07:00
f06c05c82b
Add files via upload
2022-12-31 13:19:15 -07:00
2e462f049d
fixup saved settings loading
2022-12-30 15:27:35 +03:00
394bada75d
link perlin noise note to wikipedia
2022-12-30 15:25:40 +03:00
2720a116e3
add more useful links to the header
2022-12-30 15:17:01 +03:00
c730ab0a41
add note about Perlin noise to the header
2022-12-30 15:16:45 +03:00
7e375aae38
set perlin noise by default
...
it's just too good and addresses many beginner questions, especially related to frame coherence, detail loss and picture flattening as Perlin noise's organic nature counters all of them
2022-12-30 14:29:12 +03:00
125abc3114
make perlin octaves a slider
2022-12-30 14:22:46 +03:00
e059ecab7e
json when looping none or url when not
2022-12-28 19:15:27 -08:00
1cf9434a80
fixup args and tensor shapes to make it work
2022-12-28 20:38:56 +03:00
134130a3ac
typo
2022-12-28 19:10:42 +03:00
1b0fdd0937
safer noise_type loading from settings
2022-12-28 19:05:32 +03:00
aa2651260d
add noise params to args
2022-12-28 18:56:42 +03:00
f2ee0335bd
add 2D Perlin noise
2022-12-28 13:41:17 +03:00
0b60889aa3
Merge branch 'deforum-art:automatic1111-webui' into frame-insertion
2022-12-26 10:50:30 -08:00
48ba5d10b1
Resume animation
...
Color was being converted but was already a cv2 image
Start frame was calculating .txt files still causing resume to break
2022-12-25 15:40:52 -07:00
54dd79b36f
Add antiblur params to parseq, and fall back to normal Deforum anim series if a series isn't defined in Parseq.
2022-12-25 12:19:30 +11:00
a164a20a21
Merge branch 'automatic1111-webui' into get-closer-to-notebook
2022-12-24 14:26:08 +03:00
5cc7f69cbb
Merge branch 'deforum-art:automatic1111-webui' into frame-insertion
2022-12-22 16:54:50 -08:00
5aa9f38417
Merge pull request #159 from deforum-art/QOL-only-from-pr74
...
QoL-only features from #74 : blank frames reroll, anti-blur
2022-12-22 20:57:36 +03:00
aa69d74a1c
Merge branch 'deforum-art:automatic1111-webui' into frame-insertion
2022-12-20 18:29:55 -08:00
22b8d3d527
Merge pull request #161 from rewbs/allow-URL-for-parseq-manifest
...
Add support for reading parseq manifest from a remote URL.
2022-12-20 22:35:42 +03:00
e13e0e45fb
fixup prompts param name
2022-12-20 21:15:53 +03:00
9fdb78f122
Save prompts for batch settings
...
Save prompts for batch settings
2022-12-20 06:48:53 -07:00
2c2cc25681
Add support for reading parseq manifest from a remote URL.
2022-12-20 16:56:08 +11:00
b3ac894ebc
kernel size to int
2022-12-20 00:30:26 +03:00
64dbc9649e
fixup reroll dropdown loading from file
...
fixup reroll dropdown loading from file
38th html block fix
2022-12-20 00:26:33 +03:00
78ed389bbb
anti-blur
...
using unsharp mask, from #74
2022-12-19 23:51:00 +03:00
1dcd214623
cadence to number instead of slider
2022-12-19 23:27:40 +03:00
07229b44a0
blank frames reroll model fix from pr74
2022-12-19 23:25:23 +03:00
f99ffe7410
add blank frames reroll
2022-12-19 23:14:21 +03:00
10d4d46086
Merge branch 'deforum-art:automatic1111-webui' into frame-insertion
2022-12-19 10:41:20 -08:00
a4c42e58b1
Merge pull request #156 from phi-line/always-load-midas
...
Always load the midas model on predict_depths
2022-12-19 02:54:05 +03:00
e22d6edc57
Always load the midas model on predict_depths
...
Before, the midas model would only be loaded if the midas_weight
was 1.0 or higher. With this change, we sync the logic between the
main deforum-stable-diffusion v.6 repo and this one.
2022-12-18 13:25:18 -08:00
4a94690fae
fix resume issue w/ depth maps saved in output dir
...
Before, the depth maps would interfere with counting of the number
of frames. This change includes filenames that contain the resume
timestring, but does not include those that are depth maps.
2022-12-18 12:52:37 -08:00
0ede3a83da
detach 'deforum_helpers' from 'scripts'
...
Should prevent "No module named 'scripts.deforum_helpers'" error causing Deforum's tab not show up in some cases when you've might've broken the environment with webui packages updates
2022-12-18 19:23:46 +03:00
bf2eed031e
allow for frames to be inserted on any speed > 20
2022-12-17 09:09:26 -08:00
525f6077a8
Merge branch 'deforum-art:automatic1111-webui' into frame-insertion
2022-12-13 10:32:10 -08:00
a5b4b27e66
Merge from master
2022-12-13 15:32:18 +11:00
0542082ccf
Merge from master
2022-12-13 15:21:27 +11:00
535c203b2b
Cleaning up printouts and line numbering
2022-12-12 10:38:32 -08:00
601332c0f8
allowing as many frames as wanted, no 200 lock
2022-12-12 10:28:30 -08:00
992f4b0763
initial form of infinite loop
2022-12-12 09:38:53 -08:00
7b4d7d7eb1
get closer to notebook
2022-12-10 19:09:50 -07:00
0d15fb0c8d
get closer to notebook
2022-12-10 19:09:20 -07:00
f2fd6d553b
get closer to notebook
2022-12-10 19:08:40 -07:00
ff1769dab5
Merge pull request #111 from Funofabot/full-precision-midas
...
Full precision midas
2022-12-08 22:25:54 +03:00
c4b9bf6197
Added state.interrupted to vid2frames
2022-12-07 18:41:52 -07:00
17e7d81766
Update render.py
2022-12-07 18:29:51 -07:00
f37a670c3c
Added import function
2022-12-07 18:25:07 -07:00
cf7f681032
Changed next frame init for video
2022-12-07 18:23:58 -07:00
8466fff816
Fix less frames in mask than init video
...
Fix less frames in mask than init video
2022-12-07 18:21:31 -07:00
8920126faa
Add import
2022-12-07 18:14:20 -07:00
b68146d543
Update vid2frames
...
Changed the way that vid2frames reports problems
vid2frames addresses pathing errors
vid2frames checks if setting are changed.
vid2frames has input name + frame#
2022-12-07 18:12:32 -07:00
1f32a9d52b
Change convert to tensor.
2022-12-07 18:06:39 -07:00
8961efac9a
As discussed in https://github.com/deforum-art/deforum-for-automatic1111-webui/pull/112\#issuecomment-1333896001 , add option for histogram matching.
2022-12-05 12:56:12 +11:00
3ee58dc5ad
Update depth.py
2022-12-02 08:19:42 -05:00
78d0b5615d
add checksum for downloaded models
2022-11-30 20:27:02 -05:00
a05d9341f7
changes where code diverges from defourm main
2022-11-27 09:24:21 -07:00
9a5764cff4
Update render.py
2022-11-24 06:12:48 -07:00
dda94f75ca
Update depth.py
2022-11-24 06:11:18 -07:00
8f00e8b262
Update render.py
...
fix for midas fp32
2022-11-24 06:01:30 -07:00
0491256d16
Update depth.py
...
fixing not able to use fp32
2022-11-24 05:56:08 -07:00
34cc81a32b
Don't require toggle to enable pseudo-3D anim. Don't use deltas for fv.
2022-11-24 13:24:32 +11:00
44070d0fb7
sampler index -> sampler name (due to webui's api change)
...
Adapt to a1111 change to expect sampler name rather than sampler index (fix #102 )
2022-11-23 23:08:29 +03:00
2753859268
Ensure subseed and subseed_strength are correctly honouring in internal processing
2022-11-23 16:35:01 +11:00
66136300ba
Merge branch 'fix-to-use-sampler-name' into parseq-integration
2022-11-23 15:45:24 +11:00
954777dbde
Adapt to A1111 change to expect sampler name rather than sampler index ( cdc8020d13)
2022-11-23 15:26:39 +11:00
f370e9067d
Allow interpolation mode to pick up Parseq seed and subseed declarations.
2022-11-23 15:22:46 +11:00
c8cfab82e4
Merge support for prompt interpolation
2022-11-23 11:18:31 +11:00
d16a8f1e4b
Ensure prompts are sorted numerically by keyframe.
2022-11-23 10:52:41 +11:00
25af0a6dfd
Add prompt interpolation animation mode (using composable diffusion)
2022-11-23 10:38:40 +11:00
2676133b46
Merge branch 'deforum-art:automatic1111-webui' into parseq-integration
2022-11-22 15:07:45 +11:00
6995a16f14
Update depth.py
...
replace wget code
2022-11-19 19:32:38 -05:00
45e77020ec
Update depth.py
...
convert midas_depth to a tensor before adding to adabins_depth
2022-11-13 13:48:41 -05:00
9e55c3fad3
Merge branch 'automatic1111-webui' into parseq-integration
2022-11-10 20:59:14 +11:00
53e7b74797
move stuff to deforum_helpers to fix #76
2022-11-09 14:24:18 +03:00
kabachuha
hithereai
Ryan Scarbery
rewbs
Kitchen
reallybigname
Funofabot
Matisse
kishan
Zarxrax