aboutsummaryrefslogtreecommitdiffstats
path: root/modules/sd_samplers_cfg_denoiser.py
Commit message (Collapse)AuthorAgeFilesLines
* Update comment for Pad prompt/negative prompt v0 to add a warning about ↵AUTOMATIC11112024-02-171-3/+3
| | | | truncation, make it override the v1 implementation
* fix the typo -- thanks CyberbeingAUTOMATIC11112024-01-291-1/+1
|
* Add Pad conds v0 optionAUTOMATIC11112024-01-271-9/+61
|
* add self insteadcontinue-revolution2024-01-071-1/+1
|
* add p to cfgdenoiserparamscontinue-revolution2024-01-071-1/+1
|
* Removed soft inpainting, added hooks for softpainting to work instead.CodeHatchling2023-12-071-13/+10
|
* A custom blending function can be provided by p, replacing the use of ↵CodeHatchling2023-12-071-17/+17
| | | | soft_inpainting.
* Cleaned up code, moved main code contributions into soft_inpainting.pyCodeHatchling2023-12-041-74/+10
|
* Organized the settings and UI of soft inpainting to allow for toggling the ↵CodeHatchling2023-12-041-11/+24
| | | | feature, and centralizes default values to reduce the amount of copy-pasta.
* Rewrote latent_blend() to use in-place operations and to aggressively "del" ↵CodeHatchling2023-12-031-13/+28
| | | | references with the intention of minimizing allocations and easing garbage collection.
* Tweaked default values.CodeHatchling2023-11-291-2/+2
|
* Added slider for detail preservation strength, removed largely needless ↵CodeHatchling2023-11-291-6/+5
| | | | offset parameter, changed labels in UI and for saving to/pasting data from PNG files.
* Implements "scheduling" for blending of the original latents and a latent ↵CodeHatchling2023-11-281-2/+59
| | | | blending formula that preserves details in blend transition areas.
* ditch --always-batch-cond-uncond in favor of an UI settingAUTOMATIC11112023-08-211-2/+2
|
* revert to applying mask before denoising for k-diffusion, like it was beforeAUTOMATIC11112023-08-141-1/+5
|
* put refiner into main UI, into the new accordions sectionAUTOMATIC11112023-08-121-1/+5
| | | | | | add VAE from main model into infotext, not from refiner model option to make scripts UI without gr.Group fix inconsistencies with refiner when usings samplers that do more denoising than steps
* merge errorsAUTOMATIC11112023-08-081-2/+21
|
* REMOVEAUTOMATIC11112023-08-081-1/+0
|
* add CFG denoiser implementation for DDIM, PLMS and UniPC (this is the commit ↵AUTOMATIC11112023-08-081-32/+18
| | | | when you can run both old and new implementations to compare them)
* split sd_samplers_kdiffusion into twoAUTOMATIC11112023-08-081-294/+1
|
* Split history: mv modules/sd_samplers_kdiffusion.py ↵AUTOMATIC11112023-08-081-0/+511
modules/sd_samplers_cfg_denoiser.py