Merge branch 'release_candidate'

10 files changed, 22 insertions, 16 deletions

diff --git a/modules/cmd_args.py b/modules/cmd_args.py
index 3eeb84d5..71a21b1a 100644
--- a/modules/cmd_args.py
+++ b/modules/cmd_args.py
@@ -62,7 +62,7 @@ parser.add_argument("--opt-split-attention-invokeai", action='store_true', help=
 parser.add_argument("--opt-split-attention-v1", action='store_true', help="prefer older version of split attention optimization for automatic choice of optimization")

 parser.add_argument("--opt-sdp-attention", action='store_true', help="prefer scaled dot product cross-attention layer optimization for automatic choice of optimization; requires PyTorch 2.*")

 parser.add_argument("--opt-sdp-no-mem-attention", action='store_true', help="prefer scaled dot product cross-attention layer optimization without memory efficient attention for automatic choice of optimization, makes image generation deterministic; requires PyTorch 2.*")

-parser.add_argument("--disable-opt-split-attention", action='store_true', help="does not do anything")

+parser.add_argument("--disable-opt-split-attention", action='store_true', help="prefer no cross-attention layer optimization for automatic choice of optimization")

 parser.add_argument("--disable-nan-check", action='store_true', help="do not check if produced images/latent spaces have nans; useful for running without a checkpoint in CI")

 parser.add_argument("--use-cpu", nargs='+', help="use CPU as torch device for specified modules", default=[], type=str.lower)

 parser.add_argument("--listen", action='store_true', help="launch gradio with 0.0.0.0 as server name, allowing to respond to network requests")

diff --git a/modules/extra_networks.py b/modules/extra_networks.py
index 34a3ba63..f4743928 100644
--- a/modules/extra_networks.py
+++ b/modules/extra_networks.py
@@ -26,7 +26,7 @@ class ExtraNetworkParams:
         self.named = {}

 

         for item in self.items:

-            parts = item.split('=', 2)

+            parts = item.split('=', 2) if isinstance(item, str) else [item]

             if len(parts) == 2:

                 self.named[parts[0]] = parts[1]

             else:

diff --git a/modules/generation_parameters_copypaste.py b/modules/generation_parameters_copypaste.py
index d5f0a49b..cfc04114 100644
--- a/modules/generation_parameters_copypaste.py
+++ b/modules/generation_parameters_copypaste.py
@@ -35,7 +35,7 @@ def reset():
 

 

 def quote(text):

-    if ',' not in str(text) and '\n' not in str(text):

+    if ',' not in str(text) and '\n' not in str(text) and ':' not in str(text):

         return text

 

     return json.dumps(text, ensure_ascii=False)

diff --git a/modules/images.py b/modules/images.py
index 4e8cd993..3f60e944 100644
--- a/modules/images.py
+++ b/modules/images.py
@@ -493,9 +493,12 @@ def save_image_with_geninfo(image, geninfo, filename, extension=None, existing_p
         existing_pnginfo['parameters'] = geninfo

 

     if extension.lower() == '.png':

-        pnginfo_data = PngImagePlugin.PngInfo()

-        for k, v in (existing_pnginfo or {}).items():

-            pnginfo_data.add_text(k, str(v))

+        if opts.enable_pnginfo:

+            pnginfo_data = PngImagePlugin.PngInfo()

+            for k, v in (existing_pnginfo or {}).items():

+                pnginfo_data.add_text(k, str(v))

+        else:

+            pnginfo_data = None

 

         image.save(filename, format=image_format, quality=opts.jpeg_quality, pnginfo=pnginfo_data)

 

diff --git a/modules/processing.py b/modules/processing.py
index 29a3743f..d22b353f 100644
--- a/modules/processing.py
+++ b/modules/processing.py
@@ -321,14 +321,13 @@ class StableDiffusionProcessing:
         have been used before. The second element is where the previously

         computed result is stored.

         """

-

-        if cache[0] is not None and (required_prompts, steps) == cache[0]:

+        if cache[0] is not None and (required_prompts, steps, opts.CLIP_stop_at_last_layers, shared.sd_model.sd_checkpoint_info) == cache[0]:

             return cache[1]

 

         with devices.autocast():

             cache[1] = function(shared.sd_model, required_prompts, steps)

 

-        cache[0] = (required_prompts, steps)

+        cache[0] = (required_prompts, steps, opts.CLIP_stop_at_last_layers, shared.sd_model.sd_checkpoint_info)

         return cache[1]

 

     def setup_conds(self):

diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py
index f93df0a6..221771da 100644
--- a/modules/sd_hijack.py
+++ b/modules/sd_hijack.py
@@ -68,6 +68,8 @@ def apply_optimizations():
 

     if selection == "None":

         matching_optimizer = None

+    elif selection == "Automatic" and shared.cmd_opts.disable_opt_split_attention:

+        matching_optimizer = None

     elif matching_optimizer is None:

         matching_optimizer = optimizers[0]

 

diff --git a/modules/sd_hijack_optimizations.py b/modules/sd_hijack_optimizations.py
index 2ec0b049..80e48a42 100644
--- a/modules/sd_hijack_optimizations.py
+++ b/modules/sd_hijack_optimizations.py
@@ -59,7 +59,7 @@ class SdOptimizationSdpNoMem(SdOptimization):
     name = "sdp-no-mem"

     label = "scaled dot product without memory efficient attention"

     cmd_opt = "opt_sdp_no_mem_attention"

-    priority = 90

+    priority = 80

 

     def is_available(self):

         return hasattr(torch.nn.functional, "scaled_dot_product_attention") and callable(torch.nn.functional.scaled_dot_product_attention)

@@ -73,7 +73,7 @@ class SdOptimizationSdp(SdOptimizationSdpNoMem):
     name = "sdp"

     label = "scaled dot product"

     cmd_opt = "opt_sdp_attention"

-    priority = 80

+    priority = 70

 

     def apply(self):

         ldm.modules.attention.CrossAttention.forward = scaled_dot_product_attention_forward

@@ -116,7 +116,7 @@ class SdOptimizationInvokeAI(SdOptimization):
 class SdOptimizationDoggettx(SdOptimization):

     name = "Doggettx"

     cmd_opt = "opt_split_attention"

-    priority = 20

+    priority = 90

 

     def apply(self):

         ldm.modules.attention.CrossAttention.forward = split_cross_attention_forward

diff --git a/modules/sd_models.py b/modules/sd_models.py
index b1afbaa7..7742998a 100644
--- a/modules/sd_models.py
+++ b/modules/sd_models.py
@@ -313,8 +313,6 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer
 

         timer.record("apply half()")

 

-    devices.dtype = torch.float32 if shared.cmd_opts.no_half else torch.float16

-    devices.dtype_vae = torch.float32 if shared.cmd_opts.no_half or shared.cmd_opts.no_half_vae else torch.float16

     devices.dtype_unet = model.model.diffusion_model.dtype

     devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16

 

diff --git a/modules/shared.py b/modules/shared.py
index 3099d1d2..271a062d 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -6,6 +6,7 @@ import threading
 import time

 

 import gradio as gr

+import torch

 import tqdm

 

 import modules.interrogate

@@ -76,6 +77,9 @@ cmd_opts.disable_extension_access = (cmd_opts.share or cmd_opts.listen or cmd_op
 devices.device, devices.device_interrogate, devices.device_gfpgan, devices.device_esrgan, devices.device_codeformer = \

     (devices.cpu if any(y in cmd_opts.use_cpu for y in [x, 'all']) else devices.get_optimal_device() for x in ['sd', 'interrogate', 'gfpgan', 'esrgan', 'codeformer'])

 

+devices.dtype = torch.float32 if cmd_opts.no_half else torch.float16

+devices.dtype_vae = torch.float32 if cmd_opts.no_half or cmd_opts.no_half_vae else torch.float16

+

 device = devices.device

 weight_load_location = None if cmd_opts.lowram else "cpu"

 

diff --git a/modules/ui.py b/modules/ui.py
index e62182da..361f596e 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -505,10 +505,10 @@ def create_ui():
                             with FormRow(elem_id="txt2img_hires_fix_row4", variant="compact", visible=opts.hires_fix_show_prompts) as hr_prompts_container:

                                 with gr.Column(scale=80):

                                     with gr.Row():

-                                        hr_prompt = gr.Textbox(label="Prompt", elem_id="hires_prompt", show_label=False, lines=3, placeholder="Prompt for hires fix pass.\nLeave empty to use the same prompt as in first pass.", elem_classes=["prompt"])

+                                        hr_prompt = gr.Textbox(label="Hires prompt", elem_id="hires_prompt", show_label=False, lines=3, placeholder="Prompt for hires fix pass.\nLeave empty to use the same prompt as in first pass.", elem_classes=["prompt"])

                                 with gr.Column(scale=80):

                                     with gr.Row():

-                                        hr_negative_prompt = gr.Textbox(label="Negative prompt", elem_id="hires_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt for hires fix pass.\nLeave empty to use the same negative prompt as in first pass.", elem_classes=["prompt"])

+                                        hr_negative_prompt = gr.Textbox(label="Hires negative prompt", elem_id="hires_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt for hires fix pass.\nLeave empty to use the same negative prompt as in first pass.", elem_classes=["prompt"])

 

                     elif category == "batch":

                         if not opts.dimensions_and_batch_together: