| Commit message (Collapse) | Author | Age | Files | Lines | ||
|---|---|---|---|---|---|---|
| ... | ||||||
| * | add edit_image_conditioning from my earlier edits in case there's an attempt ↵ | AUTOMATIC | 2023-01-25 | 1 | -1/+9 | |
| | | | | | | | to inegrate pix2pix properly this allows to use pix2pix model in img2img though it won't work well this way | |||||
| * | Merge pull request #6510 from brkirch/unet16-upcast-precision | AUTOMATIC1111 | 2023-01-25 | 1 | -7/+8 | |
| |\ | | | | | Add upcast options, full precision sampling from float16 UNet and upcasting attention for inference using SD 2.1 models without --no-half | |||||
| | * | Add UI setting for upcasting attention to float32 | brkirch | 2023-01-25 | 1 | -1/+1 | |
| | | | | | | | | | | | | | Adds "Upcast cross attention layer to float32" option in Stable Diffusion settings. This allows for generating images using SD 2.1 models without --no-half or xFormers. In order to make upcasting cross attention layer optimizations possible it is necessary to indent several sections of code in sd_hijack_optimizations.py so that a context manager can be used to disable autocast. Also, even though Stable Diffusion (and Diffusers) only upcast q and k, unfortunately my findings were that most of the cross attention layer optimizations could not function unless v is upcast also. | |||||
| | * | Add option for float32 sampling with float16 UNet | brkirch | 2023-01-25 | 1 | -7/+8 | |
| | | | | | | | | | This also handles type casting so that ROCm and MPS torch devices work correctly without --no-half. One cast is required for deepbooru in deepbooru_model.py, some explicit casting is required for img2img and inpainting. depth_model can't be converted to float16 or it won't work correctly on some systems (it's known to have issues on MPS) so in sd_models.py model.depth_model is removed for model.half(). | |||||
| * | | change to code for live preview fix on OSX to be bit more obvious | AUTOMATIC | 2023-01-25 | 1 | -2/+2 | |
| | | | ||||||
| * | | Fix different first gen with Approx NN previews | brkirch | 2023-01-24 | 1 | -1/+5 | |
| |/ | | | | The loading of the model for approx nn live previews can change the internal state of PyTorch, resulting in a different image. This can be avoided by preloading the approx nn model in advance. | |||||
| * | enable compact view for train tab | AUTOMATIC | 2023-01-21 | 1 | -2/+6 | |
| | | | | | prevent previews from ruining hypernetwork training | |||||
| * | extract extra network data from prompt earlier | AUTOMATIC | 2023-01-21 | 1 | -2/+2 | |
| | | ||||||
| * | make it so that extra networks are not removed from infotext | AUTOMATIC | 2023-01-21 | 1 | -1/+3 | |
| | | ||||||
| * | extra networks UI | AUTOMATIC | 2023-01-21 | 1 | -11/+13 | |
| | | | | | rework of hypernets: rather than via settings, hypernets are added directly to prompt as <hypernet:name:weight> | |||||
| * | Merge pull request #6854 from EllangoK/master | AUTOMATIC1111 | 2023-01-18 | 1 | -4/+4 | |
| |\ | | | | | Saves Extra Generation Parameters to params.txt | |||||
| | * | Changed params.txt save to after manual init call | EllangoK | 2023-01-17 | 1 | -4/+4 | |
| | | | ||||||
| * | | use DDIM in hires fix is the sampler is PLMS | AUTOMATIC | 2023-01-18 | 1 | -1/+2 | |
| |/ | ||||||
| * | make StableDiffusionProcessing class not hold a reference to shared.sd_model ↵ | AUTOMATIC | 2023-01-16 | 1 | -4/+5 | |
| | | | | | object | |||||
| * | Add a check and explanation for tensor with all NaNs. | AUTOMATIC | 2023-01-16 | 1 | -0/+3 | |
| | | ||||||
| * | change hypernets to use sha256 hashes | AUTOMATIC | 2023-01-14 | 1 | -1/+1 | |
| | | ||||||
| * | Fix extension parameters not being saved to last used parameters | space-nuko | 2023-01-12 | 1 | -4/+4 | |
| | | ||||||
| * | add an option to use old hiresfix width/height behavior | AUTOMATIC | 2023-01-09 | 1 | -2/+24 | |
| | | | | | add a visual effect to inactive hires fix elements | |||||
| * | Merge branch 'AUTOMATIC1111:master' into img2img-api-scripts | noodleanon | 2023-01-07 | 1 | -1/+0 | |
| |\ | ||||||
| | * | rework hires fix preview for #6437: movie it to where it takes less place, ↵ | AUTOMATIC | 2023-01-07 | 1 | -1/+0 | |
| | | | | | | | | | make it actually account for all relevant sliders and calculate dimensions correctly | |||||
| * | | allow img2img api to run scripts | noodleanon | 2023-01-05 | 1 | -2/+2 | |
| |/ | ||||||
| * | experimental optimization | AUTOMATIC | 2023-01-05 | 1 | -3/+25 | |
| | | ||||||
| * | move sd_model assignment to the place where we change the sd_model | AUTOMATIC | 2023-01-05 | 1 | -5/+9 | |
| | | ||||||
| * | Merge branch 'AUTOMATIC1111:master' into fix-sd-arch-switch-in-override-settings | Philpax | 2023-01-05 | 1 | -67/+113 | |
| |\ | ||||||
| | * | make hires fix not do anything if the user chooses the second pass ↵ | AUTOMATIC | 2023-01-04 | 1 | -8/+17 | |
| | | | | | | | | | resolution to be the same as first pass resolution | |||||
| | * | fix incorrect display/calculation for number of steps for hires fix in ↵ | AUTOMATIC | 2023-01-04 | 1 | -3/+6 | |
| | | | | | | | | | progress bars | |||||
| | * | added the option to specify target resolution with possibility of truncating ↵ | AUTOMATIC | 2023-01-04 | 1 | -5/+46 | |
| | | | | | | | | | for hires fix; also sampling steps | |||||
| | * | add XY plot parameters to grid image and do not add them to individual images | AUTOMATIC | 2023-01-04 | 1 | -1/+1 | |
| | | | ||||||
| | * | use shared function from processing for creating dummy mask when training ↵ | AUTOMATIC | 2023-01-04 | 1 | -19/+20 | |
| | | | | | | | | | inpainting model | |||||
| | * | add infotext to "-before-highres-fix" images | AUTOMATIC | 2023-01-04 | 1 | -1/+5 | |
| | | | ||||||
| | * | Merge pull request #6299 from stysmmaker/feat/latent-upscale-modes | AUTOMATIC1111 | 2023-01-04 | 1 | -1/+1 | |
| | |\ | | | | | | | Add more latent upscale modes | |||||
| | | * | Update processing.py | MMaker | 2023-01-04 | 1 | -1/+1 | |
| | | | | ||||||
| | * | | fix: Save full res of intermediate step | MMaker | 2023-01-04 | 1 | -1/+1 | |
| | |/ | ||||||
| | * | fix hires fix not working in API when user does not specify upscaler | AUTOMATIC | 2023-01-03 | 1 | -1/+1 | |
| | | | ||||||
| | * | Hires fix rework | AUTOMATIC | 2023-01-02 | 1 | -41/+27 | |
| | | | ||||||
| | * | make it so that memory/embeddings info is displayed in a separate UI element ↵ | AUTOMATIC | 2022-12-31 | 1 | -2/+3 | |
| | | | | | | | | | from generation parameters, and is preserved when you change the displayed infotext by clicking on gallery images | |||||
| | * | make it so that blank ENSD does not break image generation | AUTOMATIC | 2022-12-26 | 1 | -3/+4 | |
| | | | ||||||
| * | | fix(api): assign sd_model after settings change | Philpax | 2022-12-25 | 1 | -2/+4 | |
| |/ | ||||||
| * | Merge pull request #5718 from space-nuko/feature/save-hypernetwork-hash | AUTOMATIC1111 | 2022-12-24 | 1 | -2/+3 | |
| |\ | | | | | Save hypernetwork hash and fix hypernetwork parameter restoring | |||||
| | * | Save hypernetwork hash in infotext | space-nuko | 2022-12-13 | 1 | -2/+3 | |
| | | | ||||||
| * | | rename the option for img2img latent upscale | AUTOMATIC | 2022-12-24 | 1 | -1/+1 | |
| | | | ||||||
| * | | Merge pull request #5521 from AndrewRyanChama/ryan/img2imglatentscale | AUTOMATIC1111 | 2022-12-24 | 1 | -1/+5 | |
| |\ \ | | | | | | | Add latent upscale option to img2img | |||||
| | * | | Add latent upscale option to img2img | Andrew Ryan | 2022-12-08 | 1 | -1/+5 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Recently, the option to do latent upscale was added to txt2img highres fix. This feature runs by scaling the latent sample of the image, and then running a second pass of img2img. But, in this edition of highres fix, the image and parameters cannot be changed between the first pass and second pass. We might want to do a fixup in img2img before doing the second pass, or might want to run the second pass at a different resolution. This change adds the option for img2img to perform its upscale in latent space, rather than image space, giving very similar results to highres fix with latent upscale. The result is not exactly the same because there is an additional latent -> decoder -> image -> encoder -> latent that won't happen in highres fix, but this conversion has relatively small losses | |||||
| * | | | Merge pull request #5873 from philpax/override-settings-restore-afterwards | AUTOMATIC1111 | 2022-12-24 | 1 | -7/+10 | |
| |\ \ \ | | | | | | | | | feat(api): add override_settings_restore_afterwards | |||||
| | * | | | feat(api): add override_settings_restore_afterwards | Philpax | 2022-12-20 | 1 | -13/+16 | |
| | | |/ | |/| | ||||||
| * | | | Merge pull request #5753 from calvinballing/master | AUTOMATIC1111 | 2022-12-24 | 1 | -7/+7 | |
| |\ \ \ | | | | | | | | | Fix various typos | |||||
| | * | | | Fix various typos | Jim Hays | 2022-12-15 | 1 | -7/+7 | |
| | |/ / | ||||||
| * | | | add an option for img2img background color | AUTOMATIC | 2022-12-24 | 1 | -1/+1 | |
| | | | | ||||||
| * | | | Improve color correction with luminosity blend | ThereforeGames | 2022-12-11 | 1 | -3/+6 | |
| |/ / | ||||||
| * | | Merge pull request #5191 from ↵ | AUTOMATIC1111 | 2022-12-10 | 1 | -2/+8 | |
| |\ \ | | | | | | | | | | | | | aliencaocao/enable_checkpoint_switching_in_override_settings Support changing checkpoint and vae through override_settings | |||||
 |
index : stable-diffusion-webui-gfx803.git | |
| stable-diffusion-webui by AUTOMATIC1111 with patches for gfx803 GPU and Dockerfile |
| aboutsummaryrefslogtreecommitdiffstats |