Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity
1,622 Commits 274 Branches 239 Tags 397 MiB
320cbdd62d
Commit Graph

1622 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Lincoln Stein
d25bf7a55a cut over from karras to model noise schedule for higher steps 
The k_samplers come with a "karras" noise schedule which performs
very well at low step counts but becomes noisy at higher ones.

This commit introduces a threshold (currently 30 steps) at which the
k samplers will switch over from using karras to the older model
noise schedule.
 2022-10-27 17:06:49 -04:00
Damian at mba
3539f0a1da slightly more verbose docs 2022-10-27 22:50:32 +02:00
Damian at mba
737a7f779b tweak prompt syntax docs 2022-10-27 22:48:06 +02:00
Damian at mba
71dcc17fa0 fix prompt syntax doc table error 2022-10-27 22:45:59 +02:00
Damian at mba
a90ce61b1b fix broken images 2022-10-27 22:43:21 +02:00
Damian at mba
d43167ac0b improve documentation of "attention weighting" syntax 2022-10-27 22:41:06 +02:00
Damian at mba
245cf606a3 be more forgiving about prompts with ((words)) 2022-10-27 22:36:33 +02:00
Lincoln Stein
943616044a Merge branch 'switch-ksampler-noise-scheduler-adaptively' into development 
- This sets a step switchover point at which the k-samplers stop using the
  Karras noise schedule and start using the LatentDiffusion noise schedule.
  The advantage of this is that the Karras schedule produces excellent
  results at low step counts but starts to become unstable at high
  steps.

- A new command argument --karras_max, lets the user set where the
  switchover occurs. Default is 29 steps (1-29 steps Karras),
  (30 or greater LDM)

- Tildebyte, sorry to do a fast forward three-way merge for this
  but rebasing was just too painful due to extensive recent
  changes to the diffuser code.
 2022-10-27 16:11:26 -04:00
Lincoln Stein
943808b925 add threshold for switchover from Karras to LDM noise schedule 2022-10-27 15:50:32 -04:00
Damian at mba
30745f163d add one more test case 2022-10-27 21:18:08 +02:00
Damian at mba
e20108878c fix attention weight inside .swap() 2022-10-27 21:17:23 +02:00
Damian at mba
f73d349dfe refactor hybrid and cross attention control codepaths for readability 2022-10-27 19:40:37 +02:00
Damian at mba
dc86fc92ce fix crash parsing empty prompt "" 2022-10-27 19:01:54 +02:00
Lincoln Stein
aa785c3ef1 ready for merge after documentation added 2022-10-27 11:55:00 -04:00
mauwii
fb4feb380b update docker docs 2022-10-27 11:51:36 -04:00
mauwii
9b15b228b8 add action to build the container 
it does not push the container but verify buildability
 2022-10-27 11:51:36 -04:00
mauwii
99eb7e6ef2 add conda env for linux-aarch64 
- neither environment.yml nor environment-mac.yml was working
 2022-10-27 11:51:36 -04:00
mauwii
bf50a68eb5 update entrypoint 
- when run without arguments it starts the web-interface
- can also be run with your own arguments
  - if u do so it does not start web unless u do
 2022-10-27 11:51:36 -04:00
mauwii
67a7d46a29 add script to easily run the container 2022-10-27 11:51:36 -04:00
mauwii
3e2cf8a259 add env.sh with variables shared in run and build 2022-10-27 11:51:36 -04:00
mauwii
624fe4794b add build script which also creates volume 
it needs a token from huggingface to be able to download the checkpoint
 2022-10-27 11:51:36 -04:00
mauwii
44731f8a37 add .dockerignore to repo-root 
since transfering  the 2GB context would not be rly productive
 2022-10-27 11:51:36 -04:00
mauwii
b2a3c5cbe8 update Dockerfile 2022-10-27 11:51:36 -04:00
Taylor Kems
e9f690bf9d Update IMG2IMG.md 2022-10-27 11:16:15 -04:00
Lincoln Stein
0eb07b7488 Merge branch 'outpaint' of https://github.com/Kyle0654/InvokeAI into Kyle0654-outpaint 2022-10-27 09:16:40 -04:00
Lincoln Stein
16e7cbdb38 tweaks to documentation and call signature for advanced prompting 2022-10-27 08:30:09 -04:00
Damian at mba
135c62f1a4 fix issue with hot-dog, improve () suppression 2022-10-27 07:37:48 -04:00
Lincoln Stein
582e19056a Merge branch 'development' of github.com:invoke-ai/InvokeAI into development 2022-10-27 02:06:25 -04:00
Lincoln Stein
52de5c8b33 documentation fix 2022-10-27 01:58:20 -04:00
Lincoln Stein
799dc6d0df acceptable integration of new prompting system and inpainting 
This was a difficult merge because both PR #1108 and #1243 made
changes to obscure parts of the diffusion code.

- prompt weighting, merging and cross-attention working
  - cross-attention does not work with runwayML inpainting
    model, but weighting and merging are tested and working
- CLI command parsing code rewritten in order to get embedded
  quotes right
- --hires now works with runwayML inpainting
- --embiggen does not work with runwayML and will give an error
- Added an --invert option to invert masks applied to inpainting
- Updated documentation
 2022-10-27 01:51:35 -04:00
Damian at mba
79689e87ce fix crash making embeddings from too-long prompts with attention weights 2022-10-26 22:42:17 -04:00
Lincoln Stein
0d0481ce75 inpaint model progress 
- working with plain prompts, weighted prompts and merge prompts
- not tested with prompt2prompt
 2022-10-26 22:40:01 -04:00
Lincoln Stein
869d9e22c7 documentation fix 2022-10-26 22:37:30 -04:00
Lincoln Stein
3f77b68a9d fix mishandling of embedded quotes in prompt 2022-10-26 18:27:35 -04:00
Lincoln Stein
2daf187bdb working with 1.4, 1.5, not with inpainting 1.5 2022-10-26 18:25:48 -04:00
Lincoln Stein
9b7159720f resolve conflicts between PR #1108 and #1243 2022-10-26 15:37:24 -04:00
Kyle Schouviller
e7c2b90bd1 Merge branch 'outpaint' of https://github.com/Kyle0654/InvokeAI into outpaint 2022-10-26 12:12:17 -07:00
Kyle Schouviller
d05373d35a Force RGB for img2img 2022-10-26 12:12:08 -07:00
Kyle Schouviller
bd8bb8c80b Adding outpainting implementation (as part of inpaint). 2022-10-26 12:12:08 -07:00
Kyle Schouviller
dac1ab0a05 Better inpainting color-correction 2022-10-26 12:12:08 -07:00
Kyle Schouviller
2a44411f5b Force RGB for img2img 2022-10-26 12:09:38 -07:00
Lincoln Stein
2f1c1e7695 Merge branch 'fix-prompts' of https://github.com/damian0815/InvokeAI into merge-prompt-and-inpaint-model 2022-10-26 08:50:55 -04:00
Lincoln Stein
2b6d78e436 minor cleanups 
- remove --fnformat from canonicalized dream prompt arguments
  (not needed for image reproducibility)
- add -tm to canonicalized dream prompt arguments
  (definitely needed for image reproducibility)
 2022-10-26 08:32:54 -04:00
Lincoln Stein
b1da13a984 minor cleanups 
- change default model back to 1.4
- remove --fnformat from canonicalized dream prompt arguments
  (not needed for image reproducibility)
- add -tm to canonicalized dream prompt arguments
  (definitely needed for image reproducibility)
 2022-10-26 08:29:56 -04:00
Kyle Schouviller
eaf6d46a7b Adding outpainting implementation (as part of inpaint). 2022-10-26 00:39:36 -07:00
Lincoln Stein
906dafe3cd make variations work with inpainting model 2022-10-26 00:18:31 -04:00
Lincoln Stein
d3047c7cb0 do not encode init image in starting latent 2022-10-25 22:44:42 -04:00
tyler
62412f8398 fixing aspect ratio on hires 2022-10-25 21:28:50 -05:00
Kyle Schouviller
f1ca789097 Better inpainting color-correction 2022-10-25 17:10:28 -07:00
Lincoln Stein
4104ac6270 copied workflows from main to dev 2022-10-25 17:27:38 -04:00