aboutsummaryrefslogtreecommitdiffstats
path: root/modules/processing.py
AgeCommit message (Collapse)AuthorLines
2023-01-25Add UI setting for upcasting attention to float32brkirch-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.
2023-01-25Add option for float32 sampling with float16 UNetbrkirch-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().
2023-01-24Merge branch 'AUTOMATIC1111:master' into masterInvincibleDude-2/+6
2023-01-23Fix different first gen with Approx NN previewsbrkirch-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.
2023-01-22Gen params paste improvementinvincibledude-2/+2
2023-01-22Gen params paste improvementinvincibledude-2/+2
2023-01-22UI and PNG info improvementsinvincibledude-2/+2
2023-01-22UI and PNG info improvementsinvincibledude-0/+3
2023-01-22hr conditioninginvincibledude-1/+1
2023-01-22hr conditioninginvincibledude-2/+2
2023-01-22hr conditioninginvincibledude-7/+12
2023-01-22hr conditioninginvincibledude-4/+5
2023-01-22hr conditioninginvincibledude-21/+13
2023-01-22hr conditioninginvincibledude-26/+46
2023-01-22Hr-fix separate prompt experimentationinvincibledude-21/+22
2023-01-22Logging for debugginginvincibledude-0/+3
2023-01-22Fixinvincibledude-1/+1
2023-01-22Hr separate prompt testinvincibledude-1/+22
2023-01-22PLMS edge-case handling fix 5invincibledude-2/+0
2023-01-22PLMS edge-case handling fix 3invincibledude-2/+2
2023-01-22PLMS edge-case handling fix 2invincibledude-2/+6
2023-01-22PLMS edge-case handling fixinvincibledude-1/+1
2023-01-22enable compact view for train tabAUTOMATIC-2/+6
prevent previews from ruining hypernetwork training
2023-01-21Type mismatch fixinvincibledude-2/+2
2023-01-21First test of different sampler for hi-res fixinvincibledude-1/+6
2023-01-21extract extra network data from prompt earlierAUTOMATIC-2/+2
2023-01-21make it so that extra networks are not removed from infotextAUTOMATIC-1/+3
2023-01-21extra networks UIAUTOMATIC-11/+13
rework of hypernets: rather than via settings, hypernets are added directly to prompt as <hypernet:name:weight>
2023-01-18Merge pull request #6854 from EllangoK/masterAUTOMATIC1111-4/+4
Saves Extra Generation Parameters to params.txt
2023-01-18use DDIM in hires fix is the sampler is PLMSAUTOMATIC-1/+2
2023-01-17Changed params.txt save to after manual init callEllangoK-4/+4
2023-01-16make StableDiffusionProcessing class not hold a reference to shared.sd_model ↵AUTOMATIC-4/+5
object
2023-01-16Add a check and explanation for tensor with all NaNs.AUTOMATIC-0/+3
2023-01-14change hypernets to use sha256 hashesAUTOMATIC-1/+1
2023-01-12Fix extension parameters not being saved to last used parametersspace-nuko-4/+4
2023-01-09add an option to use old hiresfix width/height behaviorAUTOMATIC-2/+24
add a visual effect to inactive hires fix elements
2023-01-07Merge branch 'AUTOMATIC1111:master' into img2img-api-scriptsnoodleanon-1/+0
2023-01-07rework hires fix preview for #6437: movie it to where it takes less place, ↵AUTOMATIC-1/+0
make it actually account for all relevant sliders and calculate dimensions correctly
2023-01-05allow img2img api to run scriptsnoodleanon-2/+2
2023-01-05experimental optimizationAUTOMATIC-3/+25
2023-01-05move sd_model assignment to the place where we change the sd_modelAUTOMATIC-5/+9
2023-01-05Merge branch 'AUTOMATIC1111:master' into fix-sd-arch-switch-in-override-settingsPhilpax-67/+113
2023-01-05make hires fix not do anything if the user chooses the second pass ↵AUTOMATIC-8/+17
resolution to be the same as first pass resolution
2023-01-04fix incorrect display/calculation for number of steps for hires fix in ↵AUTOMATIC-3/+6
progress bars
2023-01-04added the option to specify target resolution with possibility of truncating ↵AUTOMATIC-5/+46
for hires fix; also sampling steps
2023-01-04add XY plot parameters to grid image and do not add them to individual imagesAUTOMATIC-1/+1
2023-01-04use shared function from processing for creating dummy mask when training ↵AUTOMATIC-19/+20
inpainting model
2023-01-04add infotext to "-before-highres-fix" imagesAUTOMATIC-1/+5
2023-01-04Merge pull request #6299 from stysmmaker/feat/latent-upscale-modesAUTOMATIC1111-1/+1
Add more latent upscale modes
2023-01-04Update processing.pyMMaker-1/+1
generated by cgit v1.2.3 (git 2.39.1) at 2025-10-08 02:05:06 +0000