Lincoln Stein
0d1aad53ef
Merge branch 'development' into development
2022-09-13 10:57:05 -04:00
Any-Winter-4079
d0a71dc361
Update attention.py for 16-32GB M1 performance ( #540 )
...
Code cleanup and attention.py einsum_ops update for M1 16-32GB performance.
Expected: On par with fastest ever from 8 to 128GB for 512x512. Allows large images.
2022-09-13 10:53:45 -04:00
jspraul
e1a6d0c138
web server does not supply embiggen options yet ( #535 )
2022-09-13 08:29:14 -04:00
Peter Baylies
e1f6ea2be7
Merge branch 'development' of https://github.com/pbaylies/stable-diffusion into development
2022-09-12 18:35:25 -04:00
Peter Baylies
5941ee620c
* check for missing parameters.
2022-09-12 18:35:10 -04:00
Peter Baylies
0a6c98e47d
Merge branch 'development' into development
2022-09-12 17:58:27 -04:00
Mihai
dedf8a3692
Remove pointless del statements in diffusionmodules.model. ( #520 )
2022-09-12 17:39:06 -04:00
Peter Baylies
5e15f1e017
Merge branch 'development' into development
2022-09-12 16:59:15 -04:00
Mihai
0bc6779361
Disable autocast for cpu to fix error. Remove unused precision arg. ( #518 )
...
When running on just cpu (intel), a call to torch.layer_norm would error with RuntimeError: expected scalar type BFloat16 but found Float
Fix buggy device handling in model.py.
Tested with scripts/dream.py --full_precision on just cpu on intel laptop. Works but slow at ~10s/it.
2022-09-12 16:55:21 -04:00
Peter Baylies
2a292d5b82
* Fallback in case we don't have get_noise() (using the base class for some reason...)
2022-09-12 16:52:22 -04:00
Peter Baylies
4a5a228fd8
Merge branch 'development' into development
2022-09-12 16:34:10 -04:00
Travco
dbf2c63c90
Add Embiggen automation to upscale-cut-img2img-stitch and achieve high res without extra VRAM ( #437 )
...
* Add Embiggen automation
* Make embiggen_tiles masking more intelligent and count from one (at least for the user), rewrite sections of Embiggen README, fix various typos throughout README
* drop duplicate log message
2022-09-12 15:37:26 -04:00
Peter Baylies
ea60d036d1
Merge branch 'development' into development
2022-09-12 10:37:49 -04:00
Lincoln Stein
9fa1f31bf2
fix opencv and realesrgan dependencies in mac install
2022-09-12 07:07:05 -04:00
Peter Baylies
77db46f99e
Merge branch 'development' into development
2022-09-12 04:50:40 -04:00
Any-Winter-4079
25d9ccc509
Update model.py
2022-09-11 22:37:45 -04:00
Any-Winter-4079
9cdf3aca7d
Update attention.py
...
Performance improvements to generate larger images in M1 #431
Update attention.py
Added dtype=r1.dtype to softmax
2022-09-11 22:36:58 -04:00
Henry van Megen
aac5102cf3
Disabled debug output ( #436 )
...
Co-authored-by: Henry van Megen <hvanmegen@gmail.com>
2022-09-11 22:34:35 -04:00
Peter Baylies
45673e8723
Merge branch 'development' into development
2022-09-11 19:03:29 -04:00
Lincoln Stein
7708f4fb98
slight efficiency gain by using += in attention.py
2022-09-11 16:03:54 -04:00
chromaticist
4951e66103
Adding support for .bin files from huggingface concepts ( #498 )
...
* Adding support for .bin files from huggingface concepts
* Updating documentation to include huggingface .bin info
2022-09-11 15:44:26 -04:00
Lincoln Stein
93c242c9fb
make gfpgan_model_exists flag available to web interface
2022-09-11 14:47:58 -04:00
Lincoln Stein
4923118610
Merge branch 'development' of github.com:lstein/stable-diffusion into development
2022-09-11 12:51:25 -04:00
Dominic Letz
defafc0e8e
Enable upscaling on m1 ( #474 )
2022-09-11 12:51:01 -04:00
Lincoln Stein
16f6a6731d
install GFPGAN inside SD repository in order to fix 'dark cast' issue #169
2022-09-11 12:47:26 -04:00
Lincoln Stein
70aa674e9e
merge PR #495 - keep using float16 in ldm.modules.attention
2022-09-11 10:34:06 -04:00
Peter Baylies
737a97c898
Merge branch 'development' into development
2022-09-11 10:30:23 -04:00
Lincoln Stein
839e30e4b8
improve CUDA VRAM monitoring
...
extra check that device==cuda before getting VRAM stats
2022-09-11 10:10:24 -04:00
Peter Baylies
c6be8f320d
Merge branch 'development' into development
2022-09-10 10:31:32 -04:00
Lincoln Stein
5c43988862
reduce VRAM memory usage by half during model loading
...
* This moves the call to half() before model.to(device) to avoid GPU
copy of full model. Improves speed and reduces memory usage dramatically
* This fix contributed by @mh-dm (Mihai)
2022-09-10 10:02:43 -04:00
Lincoln Stein
817c4a26de
remove -F option from normalized prompt; closes #483
2022-09-10 09:54:26 -04:00
Peter Baylies
103b3e7965
Merge branch 'development' into development
2022-09-10 07:43:54 -04:00
Lincoln Stein
723d074442
Allow ctrl c when using --from_file ( #472 )
...
* added ansi escapes to highlight key parts of CLI session
* adjust exception handling so that ^C will abort when reading prompts from a file
2022-09-09 18:49:51 -04:00
Peter Baylies
424f4fe244
Merge branch 'development' into development
2022-09-09 13:15:43 -04:00
Lincoln Stein
10db192cc4
changes to dogettx optimizations to run on m1
...
* Author @any-winter-4079
* Author @dogettx
Thanks to many individuals who contributed time and hardware to
benchmarking and debugging these changes.
2022-09-09 09:51:41 -04:00
Peter Baylies
c41599746d
Merge branch 'development' into development
2022-09-09 00:19:17 -04:00
Lincoln Stein
c85ae00b33
fix bug which caused seed to get "stuck" on previous image even when UI specified -1
2022-09-08 23:57:45 -04:00
Peter Baylies
7f0cc7072b
Merge branch 'development' into development
2022-09-08 22:54:26 -04:00
Lincoln Stein
33874bae8d
Squashed commit of the following:
...
commit afee7f9cea6531446171f8db171f8db742d7e67b62f0
2022-09-08 22:16:29 -04:00
Lincoln Stein
653144694f
work around unexplained crash when timesteps=1000 ( #440 )
...
* work around unexplained crash when timesteps=1000
* this fix seems to work
2022-09-08 20:41:37 -04:00
blessedcoolant
c33a84cdfd
Add New Logo ( #454 )
...
* Add instructions on how to install alongside pyenv (#393 )
Like probably many others, I have a lot of different virtualenvs, one for each project. Most of them are handled by `pyenv`.
After installing according to these instructions I had issues with ´pyenv`and `miniconda` fighting over the $PATH of my system.
But then I stumbled upon this nice solution on SO: https://stackoverflow.com/a/73139031 , upon which I have based my suggested changes.
It runs perfectly on my M1 setup, with the anaconda setup as a virtual environment handled by pyenv.
Feel free to incorporate these instructions as you see fit.
Thanks a million for all your hard work.
* Disabled debug output (#436 )
Co-authored-by: Henry van Megen <hvanmegen@gmail.com>
* Add New Logo
Co-authored-by: Håvard Gulldahl <havard@lurtgjort.no>
Co-authored-by: Henry van Megen <h.vanmegen@gmail.com>
Co-authored-by: Henry van Megen <hvanmegen@gmail.com>
Co-authored-by: Lincoln Stein <lincoln.stein@gmail.com>
2022-09-08 20:39:51 -04:00
Peter Baylies
c71d8750f7
Merge branch 'development' into development
2022-09-08 10:23:23 -04:00
Lincoln Stein
7996a30e3a
add auto-creation of mask for inpainting ( #438 )
...
* now use a single init image for both image and mask
* turn on debugging for now to write out mask and image
* add back -M option as a fallback
2022-09-08 07:34:03 -04:00
Peter Baylies
ab39bc0bac
Merge branch 'development' into development
2022-09-07 23:15:09 -04:00
Lincoln Stein
5c6b612a72
fix bug that caused same seed to be redisplayed repeatedly
2022-09-07 22:50:55 -04:00
Johan Roxendal
56f155c590
added support for parsing run log and displaying images in the frontend init state ( #410 )
...
Co-authored-by: Johan Roxendal <johan.roxendal@litteraturbanken.se>
Co-authored-by: Lincoln Stein <lincoln.stein@gmail.com>
2022-09-07 22:50:06 -04:00
Peter Baylies
bd4fc64156
Merge branch 'development' into development
2022-09-07 21:30:39 -04:00
Lincoln Stein
41687746be
added missing initialization of latent_noise to None
2022-09-07 20:24:35 -04:00
Lincoln Stein
d1d044aa87
actual image seed now written into web log rather than -1 ( #428 )
2022-09-07 17:56:59 -04:00
Peter Baylies
8b0d1e59fe
Merge branch 'development' into development
2022-09-07 15:23:58 -04:00
