You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* add backend support for changing the VAE tiling threshold
* trigger VAE tiling by image area instead of dimensions
I've tested with GGML_VULKAN_MEMORY_DEBUG all resolutions with
the same 768x768 area (even extremes like 64x9216), and many
below that: all consistently allocate 6656 bytes per image pixel.
As tiling is primarily useful to avoid excessive memory usage, it
seems reasonable to enable VAE tiling based on area rather than
maximum image side.
However, as there is currently no user interface option to change
it back to a lower value, it's best to maintain the default
behavior for now.
* replace the notile option with a configurable threshold
This allows selecting a lower threshold value, reducing the
peak memory usage.
The legacy sdnotile parameter gets automatically converted to
the new parameter, if it's the only one supplied.
* simplify tiling checks, 768 default visible in launcher
---------
Co-authored-by: Concedo <[email protected]>
makecheckbox(images_tab, "Use TAE SD (AutoFix Broken VAE)", sd_vaeauto_var, 42,command=toggletaesd,tooltiptxt="Replace VAE with TAESD. May fix bad VAE.")
5036
-
makecheckbox(images_tab, "No VAE Tiling", sd_notile_var, 44,tooltiptxt="Disables VAE tiling, may not work for large images.")
5037
+
makelabelentry(images_tab, "VAE Tiling Threshold:", sd_tiled_vae_var, 44,50, padx=144,singleline=True,tooltip="Enable VAE Tiling for images above this size, to save memory.\nSet to 0 to disable VAE tiling.")
sdparsergrouplora.add_argument("--sdquant", help="If specified, loads the model quantized to save memory.", action='store_true')
7270
7274
sdparsergrouplora.add_argument("--sdlora", metavar=('[filename]'), help="Specify an image generation LORA safetensors model to be applied.", default="")
7271
7275
sdparsergroup.add_argument("--sdloramult", metavar=('[amount]'), help="Multiplier for the image LORA model to be applied.", type=float, default=1.0)
7272
-
sdparsergroup.add_argument("--sdnotile", help="Disables VAE tiling, may not work for large images.", action='store_true')
7273
-
7276
+
sdparsergroup.add_argument("--sdtiledvae", metavar=('[maxres]'), help="Adjust the automatic VAE tiling trigger for images above this size. 0 disables vae tiling.", type=int, default=default_vae_tile_threshold)
whisperparsergroup.add_argument("--whispermodel", metavar=('[filename]'), help="Specify a Whisper .bin model to enable Speech-To-Text transcription.", default="")
0 commit comments