From 956b3d89db62a853bc7cdac250123ba8c2fc22f5 Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 00:19:05 +0530
Subject: [PATCH 1/8] New samplers for txt2img: "ddim", "plms", "heun",
 "euler", "euler_a", "dpm2", "dpm2_a", "lms"

---
 ui/index.html                      |  22 +-
 ui/sd_internal/__init__.py         |   3 +
 ui/sd_internal/ddim_callback.patch | 323 ++++++++++++++++++++++-------
 ui/sd_internal/runtime.py          |  37 ++--
 ui/server.py                       |   2 +
 5 files changed, 291 insertions(+), 96 deletions(-)

diff --git a/ui/index.html b/ui/index.html
index e8afeae1..dd1924b0 100644
--- a/ui/index.html
+++ b/ui/index.html
@@ -304,7 +304,7 @@
                     <div id="server-status-color">&nbsp;</div>
                     <span id="server-status-msg">Stable Diffusion is starting..</span>
                 </div>
-                <h1>Stable Diffusion UI <small>v2.14 <span id="updateBranchLabel"></span></small></h1>
+                <h1>Stable Diffusion UI <small>v2.15 <span id="updateBranchLabel"></span></small></h1>
             </div>
             <div id="editor-inputs">
                 <div id="editor-inputs-prompt" class="row">
@@ -353,6 +353,18 @@
                     <br/>
                     <li><label for="seed">Seed:</label> <input id="seed" name="seed" size="10" value="30000"> <input id="random_seed" name="random_seed" type="checkbox" checked> <label for="random_seed">Random Image</label></li>
                     <li><label for="num_outputs_total">Number of images to make:</label> <input id="num_outputs_total" name="num_outputs_total" value="1" size="4"> <label for="num_outputs_parallel">Generate in parallel:</label> <input id="num_outputs_parallel" name="num_outputs_parallel" value="1" size="4"> (images at once)</li>
+                    <li id="samplerSelection"><label for="sampler">Sampler:</label>
+                        <select id="sampler" name="sampler">
+                            <option value="plms" selected>plms</option>
+                            <option value="ddim">ddim</option>
+                            <option value="heun">heun</option>
+                            <option value="euler">euler</option>
+                            <option value="euler_a">euler_a</option>
+                            <option value="dpm2">dpm2</option>
+                            <option value="dpm2_a">dpm2_a</option>
+                            <option value="lms">lms</option>
+                        </select>
+                    </li>
                     <li><label for="width">Width:</label> 
                         <select id="width" name="width" value="512">
                             <option value="128">128 (*)</option>
@@ -488,6 +500,8 @@ let diskPathField = document.querySelector('#diskPath')
 let useBetaChannelField = document.querySelector("#use_beta_channel")
 let promptStrengthSlider = document.querySelector('#prompt_strength_slider')
 let promptStrengthField = document.querySelector('#prompt_strength')
+let samplerField = document.querySelector('#sampler')
+let samplerSelectionContainer = document.querySelector("#samplerSelection")
 let useFaceCorrectionField = document.querySelector("#use_face_correction")
 let useUpscalingField = document.querySelector("#use_upscale")
 let upscaleModelField = document.querySelector("#upscale_model")
@@ -1007,6 +1021,10 @@ async function makeImage() {
         if (maskSetting.checked) {
             reqBody['mask'] = inpaintingEditor.getImg()
         }
+
+        reqBody['sampler'] = 'ddim'
+    } else {
+        reqBody['sampler'] = samplerField.value
     }
 
     if (saveToDiskField.checked && diskPathField.value.trim() !== '') {
@@ -1275,6 +1293,7 @@ function showInitImagePreview() {
         initImagePreviewContainer.style.display = 'block'
         inpaintingEditorContainer.style.display = 'none'
         promptStrengthContainer.style.display = 'block'
+        samplerSelectionContainer.style.display = 'none'
         // maskSetting.checked = false
     })
 
@@ -1306,6 +1325,7 @@ initImageClearBtn.addEventListener('click', function() {
     // maskSetting.style.display = 'none'
 
     promptStrengthContainer.style.display = 'none'
+    samplerSelectionContainer.style.display = 'block'
 })
 
 maskSetting.addEventListener('click', function() {
diff --git a/ui/sd_internal/__init__.py b/ui/sd_internal/__init__.py
index c0b6c6dc..e7630a1a 100644
--- a/ui/sd_internal/__init__.py
+++ b/ui/sd_internal/__init__.py
@@ -12,6 +12,7 @@ class Request:
     height: int = 512
     seed: int = 42
     prompt_strength: float = 0.8
+    sampler: str = None # "ddim", "plms", "heun", "euler", "euler_a", "dpm2", "dpm2_a", "lms"
     # allow_nsfw: bool = False
     precision: str = "autocast" # or "full"
     save_to_disk_path: str = None
@@ -36,6 +37,7 @@ class Request:
             "height": self.height,
             "seed": self.seed,
             "prompt_strength": self.prompt_strength,
+            "sampler": self.sampler,
             "use_face_correction": self.use_face_correction,
             "use_upscale": self.use_upscale,
         }
@@ -46,6 +48,7 @@ class Request:
     prompt: {self.prompt}
     seed: {self.seed}
     num_inference_steps: {self.num_inference_steps}
+    sampler: {self.sampler}
     guidance_scale: {self.guidance_scale}
     w: {self.width}
     h: {self.height}
diff --git a/ui/sd_internal/ddim_callback.patch b/ui/sd_internal/ddim_callback.patch
index 7e40d614..9852aab8 100644
--- a/ui/sd_internal/ddim_callback.patch
+++ b/ui/sd_internal/ddim_callback.patch
@@ -1,5 +1,5 @@
 diff --git a/optimizedSD/ddpm.py b/optimizedSD/ddpm.py
-index b967b55..75ddd8b 100644
+index b967b55..10a7c32 100644
 --- a/optimizedSD/ddpm.py
 +++ b/optimizedSD/ddpm.py
 @@ -22,7 +22,7 @@ from ldm.util import exists, default, instantiate_from_config
@@ -11,122 +11,303 @@ index b967b55..75ddd8b 100644
  
  def disabled_train(self):
      """Overwrite model.train with this function to make sure train/eval mode
-@@ -485,6 +485,7 @@ class UNet(DDPM):
-                log_every_t=100,
-                unconditional_guidance_scale=1.,
-                unconditional_conditioning=None,
-+               streaming_callbacks = False,
-                ):
-         
- 
-@@ -523,12 +524,15 @@ class UNet(DDPM):
-                                         log_every_t=log_every_t,
-                                         unconditional_guidance_scale=unconditional_guidance_scale,
-                                         unconditional_conditioning=unconditional_conditioning,
-+                                        streaming_callbacks=streaming_callbacks
-                                         )
- 
+@@ -528,39 +528,46 @@ class UNet(DDPM):
          elif sampler == "ddim":
              samples = self.ddim_sampling(x_latent, conditioning, S, unconditional_guidance_scale=unconditional_guidance_scale,
                                           unconditional_conditioning=unconditional_conditioning,
 -                                         mask = mask,init_latent=x_T,use_original_steps=False)
 +                                         mask = mask,init_latent=x_T,use_original_steps=False,
-+                                         callback=callback, img_callback=img_callback,
-+                                         streaming_callbacks=streaming_callbacks)
++                                         callback=callback, img_callback=img_callback)
  
          elif sampler == "euler":
              self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
-@@ -555,11 +559,15 @@ class UNet(DDPM):
-             samples = self.lms_sampling(self.alphas_cumprod,x_latent, S, conditioning, unconditional_conditioning=unconditional_conditioning,
-                                         unconditional_guidance_scale=unconditional_guidance_scale)
+             samples = self.euler_sampling(self.alphas_cumprod,x_latent, S, conditioning, unconditional_conditioning=unconditional_conditioning,
+-                                        unconditional_guidance_scale=unconditional_guidance_scale)
++                                        unconditional_guidance_scale=unconditional_guidance_scale,
++                                        img_callback=img_callback)
+         elif sampler == "euler_a":
+             self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
+             samples = self.euler_ancestral_sampling(self.alphas_cumprod,x_latent, S, conditioning, unconditional_conditioning=unconditional_conditioning,
+-                                        unconditional_guidance_scale=unconditional_guidance_scale)
++                                        unconditional_guidance_scale=unconditional_guidance_scale,
++                                        img_callback=img_callback)
  
-+        if streaming_callbacks: # this line needs to be right after the sampling() call
-+            yield from samples
+         elif sampler == "dpm2":
+             samples = self.dpm_2_sampling(self.alphas_cumprod,x_latent, S, conditioning, unconditional_conditioning=unconditional_conditioning,
+-                                        unconditional_guidance_scale=unconditional_guidance_scale)
++                                        unconditional_guidance_scale=unconditional_guidance_scale,
++                                        img_callback=img_callback)
+         elif sampler == "heun":
+             samples = self.heun_sampling(self.alphas_cumprod,x_latent, S, conditioning, unconditional_conditioning=unconditional_conditioning,
+-                                        unconditional_guidance_scale=unconditional_guidance_scale)
++                                        unconditional_guidance_scale=unconditional_guidance_scale,
++                                        img_callback=img_callback)
+ 
+         elif sampler == "dpm2_a":
+             samples = self.dpm_2_ancestral_sampling(self.alphas_cumprod,x_latent, S, conditioning, unconditional_conditioning=unconditional_conditioning,
+-                                        unconditional_guidance_scale=unconditional_guidance_scale)
++                                        unconditional_guidance_scale=unconditional_guidance_scale,
++                                        img_callback=img_callback)
+ 
+ 
+         elif sampler == "lms":
+             samples = self.lms_sampling(self.alphas_cumprod,x_latent, S, conditioning, unconditional_conditioning=unconditional_conditioning,
+-                                        unconditional_guidance_scale=unconditional_guidance_scale)
++                                        unconditional_guidance_scale=unconditional_guidance_scale,
++                                        img_callback=img_callback)
 +
++        yield from samples
+ 
          if(self.turbo):
              self.model1.to("cpu")
              self.model2.to("cpu")
  
 -        return samples
-+        if not streaming_callbacks:
-+            return samples
- 
+-
      @torch.no_grad()
      def plms_sampling(self, cond,b, img,
-@@ -567,7 +575,8 @@ class UNet(DDPM):
-                       callback=None, quantize_denoised=False,
-                       mask=None, x0=None, img_callback=None, log_every_t=100,
-                       temperature=1., noise_dropout=0., score_corrector=None, corrector_kwargs=None,
--                      unconditional_guidance_scale=1., unconditional_conditioning=None,):
-+                      unconditional_guidance_scale=1., unconditional_conditioning=None,
-+                      streaming_callbacks=False):
-         
-         device = self.betas.device
-         timesteps = self.ddim_timesteps
-@@ -599,10 +608,21 @@ class UNet(DDPM):
+                       ddim_use_original_steps=False,
+@@ -599,10 +606,10 @@ class UNet(DDPM):
              old_eps.append(e_t)
              if len(old_eps) >= 4:
                  old_eps.pop(0)
 -            if callback: callback(i)
 -            if img_callback: img_callback(pred_x0, i)
--
++            if callback: yield from callback(i)
++            if img_callback: yield from img_callback(pred_x0, i)
+ 
 -        return img
-+            if callback:
-+                if streaming_callbacks:
-+                    yield from callback(i)
-+                else:
-+                    callback(i)
-+            if img_callback:
-+                if streaming_callbacks:
-+                    yield from img_callback(pred_x0, i)
-+                else:
-+                    img_callback(pred_x0, i)
-+
-+        if streaming_callbacks and img_callback:
-+            yield from img_callback(img, len(iterator)-1)
-+        else:
-+            return img
++        yield from img_callback(img, len(iterator)-1)
  
      @torch.no_grad()
      def p_sample_plms(self, x, c, t, index, repeat_noise=False, use_original_steps=False, quantize_denoised=False,
-@@ -706,7 +726,9 @@ class UNet(DDPM):
+@@ -706,7 +713,8 @@ class UNet(DDPM):
  
      @torch.no_grad()
      def ddim_sampling(self, x_latent, cond, t_start, unconditional_guidance_scale=1.0, unconditional_conditioning=None,
 -               mask = None,init_latent=None,use_original_steps=False):
 +               mask = None,init_latent=None,use_original_steps=False,
-+               callback=None, img_callback=None,
-+               streaming_callbacks=False):
++               callback=None, img_callback=None):
  
          timesteps = self.ddim_timesteps
          timesteps = timesteps[:t_start]
-@@ -730,10 +752,24 @@ class UNet(DDPM):
+@@ -730,10 +738,13 @@ class UNet(DDPM):
                                            unconditional_guidance_scale=unconditional_guidance_scale,
                                            unconditional_conditioning=unconditional_conditioning)
          
-+            if callback:
-+                if streaming_callbacks:
-+                    yield from callback(i)
-+                else:
-+                    callback(i)
-+            if img_callback:
-+                if streaming_callbacks:
-+                    yield from img_callback(x_dec, i)
-+                else:
-+                    img_callback(x_dec, i)
++            if callback: yield from callback(i)
++            if img_callback: yield from img_callback(x_dec, i)
 +
          if mask is not None:
 -            return x0 * mask + (1. - mask) * x_dec
 +            x_dec = x0 * mask + (1. - mask) * x_dec
  
 -        return x_dec
-+        if streaming_callbacks and img_callback:
-+            yield from img_callback(x_dec, len(iterator)-1)
-+        else:
-+            return x_dec
++        yield from img_callback(x_dec, len(iterator)-1)
  
  
      @torch.no_grad()
+@@ -779,13 +790,16 @@ class UNet(DDPM):
+ 
+ 
+     @torch.no_grad()
+-    def euler_sampling(self, ac, x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1,extra_args=None,callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.):
++    def euler_sampling(self, ac, x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1,extra_args=None,callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.,
++                        img_callback=None):
+         """Implements Algorithm 2 (Euler steps) from Karras et al. (2022)."""
+         extra_args = {} if extra_args is None else extra_args
+         cvd = CompVisDenoiser(ac)
+         sigmas = cvd.get_sigmas(S)
+         x = x*sigmas[0]
+ 
++        print(f"Running Euler Sampling with {len(sigmas) - 1} timesteps")
++
+         s_in = x.new_ones([x.shape[0]]).half()
+         for i in trange(len(sigmas) - 1, disable=disable):
+             gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
+@@ -807,13 +821,18 @@ class UNet(DDPM):
+             d = to_d(x, sigma_hat, denoised)
+             if callback is not None:
+                 callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
++
++            if img_callback: yield from img_callback(x, i)
++
+             dt = sigmas[i + 1] - sigma_hat
+             # Euler method
+             x = x + d * dt
+-        return x
++
++        yield from img_callback(x, len(sigmas)-1)
+ 
+     @torch.no_grad()
+-    def euler_ancestral_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1,extra_args=None, callback=None, disable=None):
++    def euler_ancestral_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1,extra_args=None, callback=None, disable=None,
++                        img_callback=None):
+         """Ancestral sampling with Euler method steps."""
+         extra_args = {} if extra_args is None else extra_args
+ 
+@@ -822,6 +841,8 @@ class UNet(DDPM):
+         sigmas = cvd.get_sigmas(S)
+         x = x*sigmas[0]
+ 
++        print(f"Running Euler Ancestral Sampling with {len(sigmas) - 1} timesteps")
++
+         s_in = x.new_ones([x.shape[0]]).half()
+         for i in trange(len(sigmas) - 1, disable=disable):
+ 
+@@ -837,17 +858,22 @@ class UNet(DDPM):
+             sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
+             if callback is not None:
+                 callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
++
++            if img_callback: yield from img_callback(x, i)
++
+             d = to_d(x, sigmas[i], denoised)
+             # Euler method
+             dt = sigma_down - sigmas[i]
+             x = x + d * dt
+             x = x + torch.randn_like(x) * sigma_up
+-        return x
++
++        yield from img_callback(x, len(sigmas)-1)
+ 
+ 
+ 
+     @torch.no_grad()
+-    def heun_sampling(self, ac, x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1, extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.):
++    def heun_sampling(self, ac, x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1, extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.,
++                        img_callback=None):
+         """Implements Algorithm 2 (Heun steps) from Karras et al. (2022)."""
+         extra_args = {} if extra_args is None else extra_args
+ 
+@@ -855,6 +881,8 @@ class UNet(DDPM):
+         sigmas = cvd.get_sigmas(S)
+         x = x*sigmas[0]
+ 
++        print(f"Running Heun Sampling with {len(sigmas) - 1} timesteps")
++
+ 
+         s_in = x.new_ones([x.shape[0]]).half()
+         for i in trange(len(sigmas) - 1, disable=disable):
+@@ -876,6 +904,9 @@ class UNet(DDPM):
+             d = to_d(x, sigma_hat, denoised)
+             if callback is not None:
+                 callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
++
++            if img_callback: yield from img_callback(x, i)
++
+             dt = sigmas[i + 1] - sigma_hat
+             if sigmas[i + 1] == 0:
+                 # Euler method
+@@ -895,11 +926,13 @@ class UNet(DDPM):
+                 d_2 = to_d(x_2, sigmas[i + 1], denoised_2)
+                 d_prime = (d + d_2) / 2
+                 x = x + d_prime * dt
+-        return x
++
++        yield from img_callback(x, len(sigmas)-1)
+ 
+ 
+     @torch.no_grad()
+-    def dpm_2_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1,extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.):
++    def dpm_2_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1,extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.,
++                        img_callback=None):
+         """A sampler inspired by DPM-Solver-2 and Algorithm 2 from Karras et al. (2022)."""
+         extra_args = {} if extra_args is None else extra_args
+ 
+@@ -907,6 +940,8 @@ class UNet(DDPM):
+         sigmas = cvd.get_sigmas(S)
+         x = x*sigmas[0]
+ 
++        print(f"Running DPM2 Sampling with {len(sigmas) - 1} timesteps")
++
+         s_in = x.new_ones([x.shape[0]]).half()
+         for i in trange(len(sigmas) - 1, disable=disable):
+             gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
+@@ -924,7 +959,7 @@ class UNet(DDPM):
+             e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
+             denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
+ 
+-
++            if img_callback: yield from img_callback(x, i)
+             
+             d = to_d(x, sigma_hat, denoised)
+             # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
+@@ -945,11 +980,13 @@ class UNet(DDPM):
+ 
+             d_2 = to_d(x_2, sigma_mid, denoised_2)
+             x = x + d_2 * dt_2
+-        return x
++
++        yield from img_callback(x, len(sigmas)-1)
+ 
+ 
+     @torch.no_grad()
+-    def dpm_2_ancestral_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1, extra_args=None, callback=None, disable=None):
++    def dpm_2_ancestral_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1, extra_args=None, callback=None, disable=None,
++                        img_callback=None):
+         """Ancestral sampling with DPM-Solver inspired second-order steps."""
+         extra_args = {} if extra_args is None else extra_args
+ 
+@@ -957,6 +994,8 @@ class UNet(DDPM):
+         sigmas = cvd.get_sigmas(S)
+         x = x*sigmas[0]
+ 
++        print(f"Running DPM2 Ancestral Sampling with {len(sigmas) - 1} timesteps")
++
+         s_in = x.new_ones([x.shape[0]]).half()
+         for i in trange(len(sigmas) - 1, disable=disable):
+ 
+@@ -973,6 +1012,9 @@ class UNet(DDPM):
+             sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
+             if callback is not None:
+                 callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
++
++            if img_callback: yield from img_callback(x, i)
++
+             d = to_d(x, sigmas[i], denoised)
+             # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
+             sigma_mid = ((sigmas[i] ** (1 / 3) + sigma_down ** (1 / 3)) / 2) ** 3
+@@ -993,11 +1035,13 @@ class UNet(DDPM):
+             d_2 = to_d(x_2, sigma_mid, denoised_2)
+             x = x + d_2 * dt_2
+             x = x + torch.randn_like(x) * sigma_up
+-        return x
++
++        yield from img_callback(x, len(sigmas)-1)
+ 
+ 
+     @torch.no_grad()
+-    def lms_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1, extra_args=None, callback=None, disable=None, order=4):
++    def lms_sampling(self,ac,x, S, cond, unconditional_conditioning = None, unconditional_guidance_scale = 1, extra_args=None, callback=None, disable=None, order=4,
++                        img_callback=None):
+         extra_args = {} if extra_args is None else extra_args
+         s_in = x.new_ones([x.shape[0]])
+ 
+@@ -1005,6 +1049,8 @@ class UNet(DDPM):
+         sigmas = cvd.get_sigmas(S)
+         x = x*sigmas[0]
+ 
++        print(f"Running LMS Sampling with {len(sigmas) - 1} timesteps")
++
+         ds = []
+         for i in trange(len(sigmas) - 1, disable=disable):
+ 
+@@ -1017,6 +1063,7 @@ class UNet(DDPM):
+             e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
+             denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
+ 
++            if img_callback: yield from img_callback(x, i)
+ 
+             d = to_d(x, sigmas[i], denoised)
+             ds.append(d)
+@@ -1027,4 +1074,5 @@ class UNet(DDPM):
+             cur_order = min(i + 1, order)
+             coeffs = [linear_multistep_coeff(cur_order, sigmas.cpu(), i, j) for j in range(cur_order)]
+             x = x + sum(coeff * d for coeff, d in zip(coeffs, reversed(ds)))
+-        return x
++
++        yield from img_callback(x, len(sigmas)-1)
 diff --git a/optimizedSD/openaimodelSplit.py b/optimizedSD/openaimodelSplit.py
 index abc3098..7a32ffe 100644
 --- a/optimizedSD/openaimodelSplit.py
diff --git a/ui/sd_internal/runtime.py b/ui/sd_internal/runtime.py
index 9434414e..9fa88e5b 100644
--- a/ui/sd_internal/runtime.py
+++ b/ui/sd_internal/runtime.py
@@ -275,6 +275,7 @@ def do_mk_img(req: Request):
     opt_use_upscale = req.use_upscale
     opt_show_only_filtered = req.show_only_filtered_image
     opt_format = 'png'
+    opt_sampler_name = req.sampler
 
     print(req.to_string(), '\n    device', device)
 
@@ -399,12 +400,11 @@ def do_mk_img(req: Request):
                     # run the handler
                     try:
                         if handler == _txt2img:
-                            x_samples = _txt2img(opt_W, opt_H, opt_n_samples, opt_ddim_steps, opt_scale, None, opt_C, opt_f, opt_ddim_eta, c, uc, opt_seed, img_callback, req.stream_progress_updates, mask)
+                            x_samples = _txt2img(opt_W, opt_H, opt_n_samples, opt_ddim_steps, opt_scale, None, opt_C, opt_f, opt_ddim_eta, c, uc, opt_seed, img_callback, mask, opt_sampler_name)
                         else:
-                            x_samples = _img2img(init_latent, t_enc, batch_size, opt_scale, c, uc, opt_ddim_steps, opt_ddim_eta, opt_seed, img_callback, req.stream_progress_updates, mask)
+                            x_samples = _img2img(init_latent, t_enc, batch_size, opt_scale, c, uc, opt_ddim_steps, opt_ddim_eta, opt_seed, img_callback, mask)
 
-                        if req.stream_progress_updates:
-                            yield from x_samples
+                        yield from x_samples
 
                         x_samples = partial_x_samples
                     except UserInitiatedStop:
@@ -443,7 +443,7 @@ def do_mk_img(req: Request):
                             if return_orig_img:
                                 save_image(img, img_out_path)
 
-                            save_metadata(meta_out_path, prompts, opt_seed, opt_W, opt_H, opt_ddim_steps, opt_scale, opt_strength, opt_use_face_correction, opt_use_upscale)
+                            save_metadata(meta_out_path, prompts, opt_seed, opt_W, opt_H, opt_ddim_steps, opt_scale, opt_strength, opt_use_face_correction, opt_use_upscale, opt_sampler_name)
 
                         if return_orig_img:
                             img_data = img_to_base64_str(img)
@@ -496,10 +496,7 @@ def do_mk_img(req: Request):
 
     print('Task completed')
 
-    if req.stream_progress_updates:
-        yield json.dumps(res.json())
-    else:
-        return res
+    yield json.dumps(res.json())
 
 def save_image(img, img_out_path):
     try:
@@ -507,8 +504,8 @@ def save_image(img, img_out_path):
     except:
         print('could not save the file', traceback.format_exc())
 
-def save_metadata(meta_out_path, prompts, opt_seed, opt_W, opt_H, opt_ddim_steps, opt_scale, opt_prompt_strength, opt_correct_face, opt_upscale):
-    metadata = f"{prompts[0]}\nWidth: {opt_W}\nHeight: {opt_H}\nSeed: {opt_seed}\nSteps: {opt_ddim_steps}\nGuidance Scale: {opt_scale}\nPrompt Strength: {opt_prompt_strength}\nUse Face Correction: {opt_correct_face}\nUse Upscaling: {opt_upscale}"
+def save_metadata(meta_out_path, prompts, opt_seed, opt_W, opt_H, opt_ddim_steps, opt_scale, opt_prompt_strength, opt_correct_face, opt_upscale, sampler_name):
+    metadata = f"{prompts[0]}\nWidth: {opt_W}\nHeight: {opt_H}\nSeed: {opt_seed}\nSteps: {opt_ddim_steps}\nGuidance Scale: {opt_scale}\nPrompt Strength: {opt_prompt_strength}\nUse Face Correction: {opt_correct_face}\nUse Upscaling: {opt_upscale}\nSampler: {sampler_name}"
 
     try:
         with open(meta_out_path, 'w') as f:
@@ -516,7 +513,7 @@ def save_metadata(meta_out_path, prompts, opt_seed, opt_W, opt_H, opt_ddim_steps
     except:
         print('could not save the file', traceback.format_exc())
 
-def _txt2img(opt_W, opt_H, opt_n_samples, opt_ddim_steps, opt_scale, start_code, opt_C, opt_f, opt_ddim_eta, c, uc, opt_seed, img_callback, streaming_callbacks, mask):
+def _txt2img(opt_W, opt_H, opt_n_samples, opt_ddim_steps, opt_scale, start_code, opt_C, opt_f, opt_ddim_eta, c, uc, opt_seed, img_callback, mask, sampler_name):
     shape = [opt_n_samples, opt_C, opt_H // opt_f, opt_W // opt_f]
 
     if device != "cpu":
@@ -536,17 +533,13 @@ def _txt2img(opt_W, opt_H, opt_n_samples, opt_ddim_steps, opt_scale, start_code,
         eta=opt_ddim_eta,
         x_T=start_code,
         img_callback=img_callback,
-        streaming_callbacks=streaming_callbacks,
         mask=mask,
-        sampler = 'plms',
+        sampler = sampler_name,
     )
 
-    if streaming_callbacks:
-        yield from samples_ddim
-    else:
-        return samples_ddim
+    yield from samples_ddim
 
-def _img2img(init_latent, t_enc, batch_size, opt_scale, c, uc, opt_ddim_steps, opt_ddim_eta, opt_seed, img_callback, streaming_callbacks, mask):
+def _img2img(init_latent, t_enc, batch_size, opt_scale, c, uc, opt_ddim_steps, opt_ddim_eta, opt_seed, img_callback, mask):
     # encode (scaled latent)
     z_enc = model.stochastic_encode(
         init_latent,
@@ -565,16 +558,12 @@ def _img2img(init_latent, t_enc, batch_size, opt_scale, c, uc, opt_ddim_steps, o
         unconditional_guidance_scale=opt_scale,
         unconditional_conditioning=uc,
         img_callback=img_callback,
-        streaming_callbacks=streaming_callbacks,
         mask=mask,
         x_T=x_T,
         sampler = 'ddim'
     )
 
-    if streaming_callbacks:
-        yield from samples_ddim
-    else:
-        return samples_ddim
+    yield from samples_ddim
 
 def move_fs_to_cpu():
     if device != "cpu":
diff --git a/ui/server.py b/ui/server.py
index 2ca0f88d..ad3b1ae0 100644
--- a/ui/server.py
+++ b/ui/server.py
@@ -43,6 +43,7 @@ class ImageRequest(BaseModel):
     height: int = 512
     seed: int = 42
     prompt_strength: float = 0.8
+    sampler: str = None # "ddim", "plms", "heun", "euler", "euler_a", "dpm2", "dpm2_a", "lms"
     # allow_nsfw: bool = False
     save_to_disk_path: str = None
     turbo: bool = True
@@ -105,6 +106,7 @@ def image(req : ImageRequest):
     r.height = req.height
     r.seed = req.seed
     r.prompt_strength = req.prompt_strength
+    r.sampler = req.sampler
     # r.allow_nsfw = req.allow_nsfw
     r.turbo = req.turbo
     r.use_cpu = req.use_cpu

From c675caf3f9021d257609b03dfd6c36e23c74692e Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 00:54:35 +0530
Subject: [PATCH 2/8] Fix a missing ddim_steps error for the DDIM sampler

---
 ui/sd_internal/ddim_callback.patch | 45 ++++++++++++++++--------------
 1 file changed, 24 insertions(+), 21 deletions(-)

diff --git a/ui/sd_internal/ddim_callback.patch b/ui/sd_internal/ddim_callback.patch
index 9852aab8..73dcff22 100644
--- a/ui/sd_internal/ddim_callback.patch
+++ b/ui/sd_internal/ddim_callback.patch
@@ -1,5 +1,5 @@
 diff --git a/optimizedSD/ddpm.py b/optimizedSD/ddpm.py
-index b967b55..10a7c32 100644
+index b967b55..e06e653 100644
 --- a/optimizedSD/ddpm.py
 +++ b/optimizedSD/ddpm.py
 @@ -22,7 +22,7 @@ from ldm.util import exists, default, instantiate_from_config
@@ -11,8 +11,11 @@ index b967b55..10a7c32 100644
  
  def disabled_train(self):
      """Overwrite model.train with this function to make sure train/eval mode
-@@ -528,39 +528,46 @@ class UNet(DDPM):
+@@ -526,41 +526,49 @@ class UNet(DDPM):
+                                         )
+ 
          elif sampler == "ddim":
++            self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
              samples = self.ddim_sampling(x_latent, conditioning, S, unconditional_guidance_scale=unconditional_guidance_scale,
                                           unconditional_conditioning=unconditional_conditioning,
 -                                         mask = mask,init_latent=x_T,use_original_steps=False)
@@ -67,7 +70,7 @@ index b967b55..10a7c32 100644
      @torch.no_grad()
      def plms_sampling(self, cond,b, img,
                        ddim_use_original_steps=False,
-@@ -599,10 +606,10 @@ class UNet(DDPM):
+@@ -599,10 +607,10 @@ class UNet(DDPM):
              old_eps.append(e_t)
              if len(old_eps) >= 4:
                  old_eps.pop(0)
@@ -81,7 +84,7 @@ index b967b55..10a7c32 100644
  
      @torch.no_grad()
      def p_sample_plms(self, x, c, t, index, repeat_noise=False, use_original_steps=False, quantize_denoised=False,
-@@ -706,7 +713,8 @@ class UNet(DDPM):
+@@ -706,7 +714,8 @@ class UNet(DDPM):
  
      @torch.no_grad()
      def ddim_sampling(self, x_latent, cond, t_start, unconditional_guidance_scale=1.0, unconditional_conditioning=None,
@@ -91,7 +94,7 @@ index b967b55..10a7c32 100644
  
          timesteps = self.ddim_timesteps
          timesteps = timesteps[:t_start]
-@@ -730,10 +738,13 @@ class UNet(DDPM):
+@@ -730,10 +739,13 @@ class UNet(DDPM):
                                            unconditional_guidance_scale=unconditional_guidance_scale,
                                            unconditional_conditioning=unconditional_conditioning)
          
@@ -107,7 +110,7 @@ index b967b55..10a7c32 100644
  
  
      @torch.no_grad()
-@@ -779,13 +790,16 @@ class UNet(DDPM):
+@@ -779,13 +791,16 @@ class UNet(DDPM):
  
  
      @torch.no_grad()
@@ -125,7 +128,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -807,13 +821,18 @@ class UNet(DDPM):
+@@ -807,13 +822,18 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -146,7 +149,7 @@ index b967b55..10a7c32 100644
          """Ancestral sampling with Euler method steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -822,6 +841,8 @@ class UNet(DDPM):
+@@ -822,6 +842,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -155,7 +158,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -837,17 +858,22 @@ class UNet(DDPM):
+@@ -837,17 +859,22 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -180,7 +183,7 @@ index b967b55..10a7c32 100644
          """Implements Algorithm 2 (Heun steps) from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -855,6 +881,8 @@ class UNet(DDPM):
+@@ -855,6 +882,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -189,7 +192,7 @@ index b967b55..10a7c32 100644
  
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
-@@ -876,6 +904,9 @@ class UNet(DDPM):
+@@ -876,6 +905,9 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -199,7 +202,7 @@ index b967b55..10a7c32 100644
              dt = sigmas[i + 1] - sigma_hat
              if sigmas[i + 1] == 0:
                  # Euler method
-@@ -895,11 +926,13 @@ class UNet(DDPM):
+@@ -895,11 +927,13 @@ class UNet(DDPM):
                  d_2 = to_d(x_2, sigmas[i + 1], denoised_2)
                  d_prime = (d + d_2) / 2
                  x = x + d_prime * dt
@@ -215,7 +218,7 @@ index b967b55..10a7c32 100644
          """A sampler inspired by DPM-Solver-2 and Algorithm 2 from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -907,6 +940,8 @@ class UNet(DDPM):
+@@ -907,6 +941,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -224,7 +227,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -924,7 +959,7 @@ class UNet(DDPM):
+@@ -924,7 +960,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -233,7 +236,7 @@ index b967b55..10a7c32 100644
              
              d = to_d(x, sigma_hat, denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
-@@ -945,11 +980,13 @@ class UNet(DDPM):
+@@ -945,11 +981,13 @@ class UNet(DDPM):
  
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
@@ -249,7 +252,7 @@ index b967b55..10a7c32 100644
          """Ancestral sampling with DPM-Solver inspired second-order steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -957,6 +994,8 @@ class UNet(DDPM):
+@@ -957,6 +995,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -258,7 +261,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -973,6 +1012,9 @@ class UNet(DDPM):
+@@ -973,6 +1013,9 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -268,7 +271,7 @@ index b967b55..10a7c32 100644
              d = to_d(x, sigmas[i], denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
              sigma_mid = ((sigmas[i] ** (1 / 3) + sigma_down ** (1 / 3)) / 2) ** 3
-@@ -993,11 +1035,13 @@ class UNet(DDPM):
+@@ -993,11 +1036,13 @@ class UNet(DDPM):
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
              x = x + torch.randn_like(x) * sigma_up
@@ -284,7 +287,7 @@ index b967b55..10a7c32 100644
          extra_args = {} if extra_args is None else extra_args
          s_in = x.new_ones([x.shape[0]])
  
-@@ -1005,6 +1049,8 @@ class UNet(DDPM):
+@@ -1005,6 +1050,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -293,7 +296,7 @@ index b967b55..10a7c32 100644
          ds = []
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -1017,6 +1063,7 @@ class UNet(DDPM):
+@@ -1017,6 +1064,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -301,7 +304,7 @@ index b967b55..10a7c32 100644
  
              d = to_d(x, sigmas[i], denoised)
              ds.append(d)
-@@ -1027,4 +1074,5 @@ class UNet(DDPM):
+@@ -1027,4 +1075,5 @@ class UNet(DDPM):
              cur_order = min(i + 1, order)
              coeffs = [linear_multistep_coeff(cur_order, sigmas.cpu(), i, j) for j in range(cur_order)]
              x = x + sum(coeff * d for coeff, d in zip(coeffs, reversed(ds)))

From e7f9db5e5644f6435ea0aec8b2a770b201d6cf94 Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 11:05:35 +0530
Subject: [PATCH 3/8] Fix blurry img2img

---
 ui/sd_internal/ddim_callback.patch | 52 ++++++++++++++++++------------
 1 file changed, 32 insertions(+), 20 deletions(-)

diff --git a/ui/sd_internal/ddim_callback.patch b/ui/sd_internal/ddim_callback.patch
index 73dcff22..d65399a7 100644
--- a/ui/sd_internal/ddim_callback.patch
+++ b/ui/sd_internal/ddim_callback.patch
@@ -1,5 +1,5 @@
 diff --git a/optimizedSD/ddpm.py b/optimizedSD/ddpm.py
-index b967b55..e06e653 100644
+index b967b55..1c5f351 100644
 --- a/optimizedSD/ddpm.py
 +++ b/optimizedSD/ddpm.py
 @@ -22,7 +22,7 @@ from ldm.util import exists, default, instantiate_from_config
@@ -15,7 +15,7 @@ index b967b55..e06e653 100644
                                          )
  
          elif sampler == "ddim":
-+            self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
++            # self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
              samples = self.ddim_sampling(x_latent, conditioning, S, unconditional_guidance_scale=unconditional_guidance_scale,
                                           unconditional_conditioning=unconditional_conditioning,
 -                                         mask = mask,init_latent=x_T,use_original_steps=False)
@@ -84,17 +84,29 @@ index b967b55..e06e653 100644
  
      @torch.no_grad()
      def p_sample_plms(self, x, c, t, index, repeat_noise=False, use_original_steps=False, quantize_denoised=False,
-@@ -706,7 +714,8 @@ class UNet(DDPM):
+@@ -675,7 +683,9 @@ class UNet(DDPM):
+     def stochastic_encode(self, x0, t, seed, ddim_eta,ddim_steps,use_original_steps=False, noise=None):
+         # fast, but does not allow for exact reconstruction
+         # t serves as an index to gather the correct alphas
++        print('making schedule')
+         self.make_schedule(ddim_num_steps=ddim_steps, ddim_eta=ddim_eta, verbose=False)
++        print('made schedule', self.ddim_timesteps)
+         sqrt_alphas_cumprod = torch.sqrt(self.ddim_alphas)
+ 
+         if noise is None:
+@@ -706,7 +716,10 @@ class UNet(DDPM):
  
      @torch.no_grad()
      def ddim_sampling(self, x_latent, cond, t_start, unconditional_guidance_scale=1.0, unconditional_conditioning=None,
 -               mask = None,init_latent=None,use_original_steps=False):
 +               mask = None,init_latent=None,use_original_steps=False,
 +               callback=None, img_callback=None):
++
++        print('ddim steps', self.ddim_timesteps)
  
          timesteps = self.ddim_timesteps
          timesteps = timesteps[:t_start]
-@@ -730,10 +739,13 @@ class UNet(DDPM):
+@@ -730,10 +743,13 @@ class UNet(DDPM):
                                            unconditional_guidance_scale=unconditional_guidance_scale,
                                            unconditional_conditioning=unconditional_conditioning)
          
@@ -110,7 +122,7 @@ index b967b55..e06e653 100644
  
  
      @torch.no_grad()
-@@ -779,13 +791,16 @@ class UNet(DDPM):
+@@ -779,13 +795,16 @@ class UNet(DDPM):
  
  
      @torch.no_grad()
@@ -128,7 +140,7 @@ index b967b55..e06e653 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -807,13 +822,18 @@ class UNet(DDPM):
+@@ -807,13 +826,18 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -149,7 +161,7 @@ index b967b55..e06e653 100644
          """Ancestral sampling with Euler method steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -822,6 +842,8 @@ class UNet(DDPM):
+@@ -822,6 +846,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -158,7 +170,7 @@ index b967b55..e06e653 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -837,17 +859,22 @@ class UNet(DDPM):
+@@ -837,17 +863,22 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -183,7 +195,7 @@ index b967b55..e06e653 100644
          """Implements Algorithm 2 (Heun steps) from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -855,6 +882,8 @@ class UNet(DDPM):
+@@ -855,6 +886,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -192,7 +204,7 @@ index b967b55..e06e653 100644
  
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
-@@ -876,6 +905,9 @@ class UNet(DDPM):
+@@ -876,6 +909,9 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -202,7 +214,7 @@ index b967b55..e06e653 100644
              dt = sigmas[i + 1] - sigma_hat
              if sigmas[i + 1] == 0:
                  # Euler method
-@@ -895,11 +927,13 @@ class UNet(DDPM):
+@@ -895,11 +931,13 @@ class UNet(DDPM):
                  d_2 = to_d(x_2, sigmas[i + 1], denoised_2)
                  d_prime = (d + d_2) / 2
                  x = x + d_prime * dt
@@ -218,7 +230,7 @@ index b967b55..e06e653 100644
          """A sampler inspired by DPM-Solver-2 and Algorithm 2 from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -907,6 +941,8 @@ class UNet(DDPM):
+@@ -907,6 +945,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -227,7 +239,7 @@ index b967b55..e06e653 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -924,7 +960,7 @@ class UNet(DDPM):
+@@ -924,7 +964,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -236,7 +248,7 @@ index b967b55..e06e653 100644
              
              d = to_d(x, sigma_hat, denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
-@@ -945,11 +981,13 @@ class UNet(DDPM):
+@@ -945,11 +985,13 @@ class UNet(DDPM):
  
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
@@ -252,7 +264,7 @@ index b967b55..e06e653 100644
          """Ancestral sampling with DPM-Solver inspired second-order steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -957,6 +995,8 @@ class UNet(DDPM):
+@@ -957,6 +999,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -261,7 +273,7 @@ index b967b55..e06e653 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -973,6 +1013,9 @@ class UNet(DDPM):
+@@ -973,6 +1017,9 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -271,7 +283,7 @@ index b967b55..e06e653 100644
              d = to_d(x, sigmas[i], denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
              sigma_mid = ((sigmas[i] ** (1 / 3) + sigma_down ** (1 / 3)) / 2) ** 3
-@@ -993,11 +1036,13 @@ class UNet(DDPM):
+@@ -993,11 +1040,13 @@ class UNet(DDPM):
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
              x = x + torch.randn_like(x) * sigma_up
@@ -287,7 +299,7 @@ index b967b55..e06e653 100644
          extra_args = {} if extra_args is None else extra_args
          s_in = x.new_ones([x.shape[0]])
  
-@@ -1005,6 +1050,8 @@ class UNet(DDPM):
+@@ -1005,6 +1054,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -296,7 +308,7 @@ index b967b55..e06e653 100644
          ds = []
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -1017,6 +1064,7 @@ class UNet(DDPM):
+@@ -1017,6 +1068,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -304,7 +316,7 @@ index b967b55..e06e653 100644
  
              d = to_d(x, sigmas[i], denoised)
              ds.append(d)
-@@ -1027,4 +1075,5 @@ class UNet(DDPM):
+@@ -1027,4 +1079,5 @@ class UNet(DDPM):
              cur_order = min(i + 1, order)
              coeffs = [linear_multistep_coeff(cur_order, sigmas.cpu(), i, j) for j in range(cur_order)]
              x = x + sum(coeff * d for coeff, d in zip(coeffs, reversed(ds)))

From 83cb473a45de3e1e55a47a2ca2623437177c450f Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 11:14:06 +0530
Subject: [PATCH 4/8] Fix the ddim_timesteps attribute missing error for
 txt2img with the ddim sampler

---
 ui/sd_internal/ddim_callback.patch | 57 +++++++++++-------------------
 ui/sd_internal/runtime.py          |  3 ++
 2 files changed, 24 insertions(+), 36 deletions(-)

diff --git a/ui/sd_internal/ddim_callback.patch b/ui/sd_internal/ddim_callback.patch
index d65399a7..9852aab8 100644
--- a/ui/sd_internal/ddim_callback.patch
+++ b/ui/sd_internal/ddim_callback.patch
@@ -1,5 +1,5 @@
 diff --git a/optimizedSD/ddpm.py b/optimizedSD/ddpm.py
-index b967b55..1c5f351 100644
+index b967b55..10a7c32 100644
 --- a/optimizedSD/ddpm.py
 +++ b/optimizedSD/ddpm.py
 @@ -22,7 +22,7 @@ from ldm.util import exists, default, instantiate_from_config
@@ -11,11 +11,8 @@ index b967b55..1c5f351 100644
  
  def disabled_train(self):
      """Overwrite model.train with this function to make sure train/eval mode
-@@ -526,41 +526,49 @@ class UNet(DDPM):
-                                         )
- 
+@@ -528,39 +528,46 @@ class UNet(DDPM):
          elif sampler == "ddim":
-+            # self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
              samples = self.ddim_sampling(x_latent, conditioning, S, unconditional_guidance_scale=unconditional_guidance_scale,
                                           unconditional_conditioning=unconditional_conditioning,
 -                                         mask = mask,init_latent=x_T,use_original_steps=False)
@@ -70,7 +67,7 @@ index b967b55..1c5f351 100644
      @torch.no_grad()
      def plms_sampling(self, cond,b, img,
                        ddim_use_original_steps=False,
-@@ -599,10 +607,10 @@ class UNet(DDPM):
+@@ -599,10 +606,10 @@ class UNet(DDPM):
              old_eps.append(e_t)
              if len(old_eps) >= 4:
                  old_eps.pop(0)
@@ -84,29 +81,17 @@ index b967b55..1c5f351 100644
  
      @torch.no_grad()
      def p_sample_plms(self, x, c, t, index, repeat_noise=False, use_original_steps=False, quantize_denoised=False,
-@@ -675,7 +683,9 @@ class UNet(DDPM):
-     def stochastic_encode(self, x0, t, seed, ddim_eta,ddim_steps,use_original_steps=False, noise=None):
-         # fast, but does not allow for exact reconstruction
-         # t serves as an index to gather the correct alphas
-+        print('making schedule')
-         self.make_schedule(ddim_num_steps=ddim_steps, ddim_eta=ddim_eta, verbose=False)
-+        print('made schedule', self.ddim_timesteps)
-         sqrt_alphas_cumprod = torch.sqrt(self.ddim_alphas)
- 
-         if noise is None:
-@@ -706,7 +716,10 @@ class UNet(DDPM):
+@@ -706,7 +713,8 @@ class UNet(DDPM):
  
      @torch.no_grad()
      def ddim_sampling(self, x_latent, cond, t_start, unconditional_guidance_scale=1.0, unconditional_conditioning=None,
 -               mask = None,init_latent=None,use_original_steps=False):
 +               mask = None,init_latent=None,use_original_steps=False,
 +               callback=None, img_callback=None):
-+
-+        print('ddim steps', self.ddim_timesteps)
  
          timesteps = self.ddim_timesteps
          timesteps = timesteps[:t_start]
-@@ -730,10 +743,13 @@ class UNet(DDPM):
+@@ -730,10 +738,13 @@ class UNet(DDPM):
                                            unconditional_guidance_scale=unconditional_guidance_scale,
                                            unconditional_conditioning=unconditional_conditioning)
          
@@ -122,7 +107,7 @@ index b967b55..1c5f351 100644
  
  
      @torch.no_grad()
-@@ -779,13 +795,16 @@ class UNet(DDPM):
+@@ -779,13 +790,16 @@ class UNet(DDPM):
  
  
      @torch.no_grad()
@@ -140,7 +125,7 @@ index b967b55..1c5f351 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -807,13 +826,18 @@ class UNet(DDPM):
+@@ -807,13 +821,18 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -161,7 +146,7 @@ index b967b55..1c5f351 100644
          """Ancestral sampling with Euler method steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -822,6 +846,8 @@ class UNet(DDPM):
+@@ -822,6 +841,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -170,7 +155,7 @@ index b967b55..1c5f351 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -837,17 +863,22 @@ class UNet(DDPM):
+@@ -837,17 +858,22 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -195,7 +180,7 @@ index b967b55..1c5f351 100644
          """Implements Algorithm 2 (Heun steps) from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -855,6 +886,8 @@ class UNet(DDPM):
+@@ -855,6 +881,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -204,7 +189,7 @@ index b967b55..1c5f351 100644
  
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
-@@ -876,6 +909,9 @@ class UNet(DDPM):
+@@ -876,6 +904,9 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -214,7 +199,7 @@ index b967b55..1c5f351 100644
              dt = sigmas[i + 1] - sigma_hat
              if sigmas[i + 1] == 0:
                  # Euler method
-@@ -895,11 +931,13 @@ class UNet(DDPM):
+@@ -895,11 +926,13 @@ class UNet(DDPM):
                  d_2 = to_d(x_2, sigmas[i + 1], denoised_2)
                  d_prime = (d + d_2) / 2
                  x = x + d_prime * dt
@@ -230,7 +215,7 @@ index b967b55..1c5f351 100644
          """A sampler inspired by DPM-Solver-2 and Algorithm 2 from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -907,6 +945,8 @@ class UNet(DDPM):
+@@ -907,6 +940,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -239,7 +224,7 @@ index b967b55..1c5f351 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -924,7 +964,7 @@ class UNet(DDPM):
+@@ -924,7 +959,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -248,7 +233,7 @@ index b967b55..1c5f351 100644
              
              d = to_d(x, sigma_hat, denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
-@@ -945,11 +985,13 @@ class UNet(DDPM):
+@@ -945,11 +980,13 @@ class UNet(DDPM):
  
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
@@ -264,7 +249,7 @@ index b967b55..1c5f351 100644
          """Ancestral sampling with DPM-Solver inspired second-order steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -957,6 +999,8 @@ class UNet(DDPM):
+@@ -957,6 +994,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -273,7 +258,7 @@ index b967b55..1c5f351 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -973,6 +1017,9 @@ class UNet(DDPM):
+@@ -973,6 +1012,9 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -283,7 +268,7 @@ index b967b55..1c5f351 100644
              d = to_d(x, sigmas[i], denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
              sigma_mid = ((sigmas[i] ** (1 / 3) + sigma_down ** (1 / 3)) / 2) ** 3
-@@ -993,11 +1040,13 @@ class UNet(DDPM):
+@@ -993,11 +1035,13 @@ class UNet(DDPM):
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
              x = x + torch.randn_like(x) * sigma_up
@@ -299,7 +284,7 @@ index b967b55..1c5f351 100644
          extra_args = {} if extra_args is None else extra_args
          s_in = x.new_ones([x.shape[0]])
  
-@@ -1005,6 +1054,8 @@ class UNet(DDPM):
+@@ -1005,6 +1049,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -308,7 +293,7 @@ index b967b55..1c5f351 100644
          ds = []
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -1017,6 +1068,7 @@ class UNet(DDPM):
+@@ -1017,6 +1063,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -316,7 +301,7 @@ index b967b55..1c5f351 100644
  
              d = to_d(x, sigmas[i], denoised)
              ds.append(d)
-@@ -1027,4 +1079,5 @@ class UNet(DDPM):
+@@ -1027,4 +1074,5 @@ class UNet(DDPM):
              cur_order = min(i + 1, order)
              coeffs = [linear_multistep_coeff(cur_order, sigmas.cpu(), i, j) for j in range(cur_order)]
              x = x + sum(coeff * d for coeff, d in zip(coeffs, reversed(ds)))
diff --git a/ui/sd_internal/runtime.py b/ui/sd_internal/runtime.py
index 9fa88e5b..c0382f82 100644
--- a/ui/sd_internal/runtime.py
+++ b/ui/sd_internal/runtime.py
@@ -522,6 +522,9 @@ def _txt2img(opt_W, opt_H, opt_n_samples, opt_ddim_steps, opt_scale, start_code,
         while torch.cuda.memory_allocated() / 1e6 >= mem:
             time.sleep(1)
 
+    if sampler_name == 'ddim' and not hasattr(model, 'ddim_timesteps'):
+        model.make_schedule(ddim_num_steps=opt_ddim_steps, ddim_eta=opt_ddim_eta, verbose=False)
+
     samples_ddim = model.sample(
         S=opt_ddim_steps,
         conditioning=c,

From a3de0820b33f9261fc4184d3c0711e1e56cdefa0 Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 11:44:50 +0530
Subject: [PATCH 5/8] Fix the 'Expected all tensors to be on the same device'
 error

---
 ui/sd_internal/ddim_callback.patch | 51 ++++++++++++++++++------------
 ui/sd_internal/runtime.py          |  3 --
 2 files changed, 30 insertions(+), 24 deletions(-)

diff --git a/ui/sd_internal/ddim_callback.patch b/ui/sd_internal/ddim_callback.patch
index 9852aab8..36335abe 100644
--- a/ui/sd_internal/ddim_callback.patch
+++ b/ui/sd_internal/ddim_callback.patch
@@ -1,5 +1,5 @@
 diff --git a/optimizedSD/ddpm.py b/optimizedSD/ddpm.py
-index b967b55..10a7c32 100644
+index b967b55..35ef520 100644
 --- a/optimizedSD/ddpm.py
 +++ b/optimizedSD/ddpm.py
 @@ -22,7 +22,7 @@ from ldm.util import exists, default, instantiate_from_config
@@ -11,7 +11,16 @@ index b967b55..10a7c32 100644
  
  def disabled_train(self):
      """Overwrite model.train with this function to make sure train/eval mode
-@@ -528,39 +528,46 @@ class UNet(DDPM):
+@@ -506,6 +506,8 @@ class UNet(DDPM):
+ 
+         x_latent = noise if x0 is None else x0
+         # sampling
++        if sampler in ('ddim', 'dpm2', 'heun', 'dpm2_a', 'lms') and not hasattr(self, 'ddim_timesteps'):
++            self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
+         
+         if sampler == "plms":
+             self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=False)
+@@ -528,39 +530,46 @@ class UNet(DDPM):
          elif sampler == "ddim":
              samples = self.ddim_sampling(x_latent, conditioning, S, unconditional_guidance_scale=unconditional_guidance_scale,
                                           unconditional_conditioning=unconditional_conditioning,
@@ -67,7 +76,7 @@ index b967b55..10a7c32 100644
      @torch.no_grad()
      def plms_sampling(self, cond,b, img,
                        ddim_use_original_steps=False,
-@@ -599,10 +606,10 @@ class UNet(DDPM):
+@@ -599,10 +608,10 @@ class UNet(DDPM):
              old_eps.append(e_t)
              if len(old_eps) >= 4:
                  old_eps.pop(0)
@@ -81,7 +90,7 @@ index b967b55..10a7c32 100644
  
      @torch.no_grad()
      def p_sample_plms(self, x, c, t, index, repeat_noise=False, use_original_steps=False, quantize_denoised=False,
-@@ -706,7 +713,8 @@ class UNet(DDPM):
+@@ -706,7 +715,8 @@ class UNet(DDPM):
  
      @torch.no_grad()
      def ddim_sampling(self, x_latent, cond, t_start, unconditional_guidance_scale=1.0, unconditional_conditioning=None,
@@ -91,7 +100,7 @@ index b967b55..10a7c32 100644
  
          timesteps = self.ddim_timesteps
          timesteps = timesteps[:t_start]
-@@ -730,10 +738,13 @@ class UNet(DDPM):
+@@ -730,10 +740,13 @@ class UNet(DDPM):
                                            unconditional_guidance_scale=unconditional_guidance_scale,
                                            unconditional_conditioning=unconditional_conditioning)
          
@@ -107,7 +116,7 @@ index b967b55..10a7c32 100644
  
  
      @torch.no_grad()
-@@ -779,13 +790,16 @@ class UNet(DDPM):
+@@ -779,13 +792,16 @@ class UNet(DDPM):
  
  
      @torch.no_grad()
@@ -125,7 +134,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -807,13 +821,18 @@ class UNet(DDPM):
+@@ -807,13 +823,18 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -146,7 +155,7 @@ index b967b55..10a7c32 100644
          """Ancestral sampling with Euler method steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -822,6 +841,8 @@ class UNet(DDPM):
+@@ -822,6 +843,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -155,7 +164,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -837,17 +858,22 @@ class UNet(DDPM):
+@@ -837,17 +860,22 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -180,7 +189,7 @@ index b967b55..10a7c32 100644
          """Implements Algorithm 2 (Heun steps) from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -855,6 +881,8 @@ class UNet(DDPM):
+@@ -855,6 +883,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -189,7 +198,7 @@ index b967b55..10a7c32 100644
  
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
-@@ -876,6 +904,9 @@ class UNet(DDPM):
+@@ -876,6 +906,9 @@ class UNet(DDPM):
              d = to_d(x, sigma_hat, denoised)
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigma_hat, 'denoised': denoised})
@@ -199,7 +208,7 @@ index b967b55..10a7c32 100644
              dt = sigmas[i + 1] - sigma_hat
              if sigmas[i + 1] == 0:
                  # Euler method
-@@ -895,11 +926,13 @@ class UNet(DDPM):
+@@ -895,11 +928,13 @@ class UNet(DDPM):
                  d_2 = to_d(x_2, sigmas[i + 1], denoised_2)
                  d_prime = (d + d_2) / 2
                  x = x + d_prime * dt
@@ -215,7 +224,7 @@ index b967b55..10a7c32 100644
          """A sampler inspired by DPM-Solver-2 and Algorithm 2 from Karras et al. (2022)."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -907,6 +940,8 @@ class UNet(DDPM):
+@@ -907,6 +942,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -224,7 +233,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
              gamma = min(s_churn / (len(sigmas) - 1), 2 ** 0.5 - 1) if s_tmin <= sigmas[i] <= s_tmax else 0.
-@@ -924,7 +959,7 @@ class UNet(DDPM):
+@@ -924,7 +961,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -233,7 +242,7 @@ index b967b55..10a7c32 100644
              
              d = to_d(x, sigma_hat, denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
-@@ -945,11 +980,13 @@ class UNet(DDPM):
+@@ -945,11 +982,13 @@ class UNet(DDPM):
  
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
@@ -249,7 +258,7 @@ index b967b55..10a7c32 100644
          """Ancestral sampling with DPM-Solver inspired second-order steps."""
          extra_args = {} if extra_args is None else extra_args
  
-@@ -957,6 +994,8 @@ class UNet(DDPM):
+@@ -957,6 +996,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -258,7 +267,7 @@ index b967b55..10a7c32 100644
          s_in = x.new_ones([x.shape[0]]).half()
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -973,6 +1012,9 @@ class UNet(DDPM):
+@@ -973,6 +1014,9 @@ class UNet(DDPM):
              sigma_down, sigma_up = get_ancestral_step(sigmas[i], sigmas[i + 1])
              if callback is not None:
                  callback({'x': x, 'i': i, 'sigma': sigmas[i], 'sigma_hat': sigmas[i], 'denoised': denoised})
@@ -268,7 +277,7 @@ index b967b55..10a7c32 100644
              d = to_d(x, sigmas[i], denoised)
              # Midpoint method, where the midpoint is chosen according to a rho=3 Karras schedule
              sigma_mid = ((sigmas[i] ** (1 / 3) + sigma_down ** (1 / 3)) / 2) ** 3
-@@ -993,11 +1035,13 @@ class UNet(DDPM):
+@@ -993,11 +1037,13 @@ class UNet(DDPM):
              d_2 = to_d(x_2, sigma_mid, denoised_2)
              x = x + d_2 * dt_2
              x = x + torch.randn_like(x) * sigma_up
@@ -284,7 +293,7 @@ index b967b55..10a7c32 100644
          extra_args = {} if extra_args is None else extra_args
          s_in = x.new_ones([x.shape[0]])
  
-@@ -1005,6 +1049,8 @@ class UNet(DDPM):
+@@ -1005,6 +1051,8 @@ class UNet(DDPM):
          sigmas = cvd.get_sigmas(S)
          x = x*sigmas[0]
  
@@ -293,7 +302,7 @@ index b967b55..10a7c32 100644
          ds = []
          for i in trange(len(sigmas) - 1, disable=disable):
  
-@@ -1017,6 +1063,7 @@ class UNet(DDPM):
+@@ -1017,6 +1065,7 @@ class UNet(DDPM):
              e_t_uncond, e_t = (x_in  + eps * c_out).chunk(2)
              denoised = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
  
@@ -301,7 +310,7 @@ index b967b55..10a7c32 100644
  
              d = to_d(x, sigmas[i], denoised)
              ds.append(d)
-@@ -1027,4 +1074,5 @@ class UNet(DDPM):
+@@ -1027,4 +1076,5 @@ class UNet(DDPM):
              cur_order = min(i + 1, order)
              coeffs = [linear_multistep_coeff(cur_order, sigmas.cpu(), i, j) for j in range(cur_order)]
              x = x + sum(coeff * d for coeff, d in zip(coeffs, reversed(ds)))
diff --git a/ui/sd_internal/runtime.py b/ui/sd_internal/runtime.py
index c0382f82..9fa88e5b 100644
--- a/ui/sd_internal/runtime.py
+++ b/ui/sd_internal/runtime.py
@@ -522,9 +522,6 @@ def _txt2img(opt_W, opt_H, opt_n_samples, opt_ddim_steps, opt_scale, start_code,
         while torch.cuda.memory_allocated() / 1e6 >= mem:
             time.sleep(1)
 
-    if sampler_name == 'ddim' and not hasattr(model, 'ddim_timesteps'):
-        model.make_schedule(ddim_num_steps=opt_ddim_steps, ddim_eta=opt_ddim_eta, verbose=False)
-
     samples_ddim = model.sample(
         S=opt_ddim_steps,
         conditioning=c,

From 5c827703a12b9bea21a61e42d5f596fe4d46e027 Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 13:38:33 +0530
Subject: [PATCH 6/8] Report the steps for img2img correctly

---
 ui/index.html             | 2 +-
 ui/sd_internal/runtime.py | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/ui/index.html b/ui/index.html
index dd1924b0..db393213 100644
--- a/ui/index.html
+++ b/ui/index.html
@@ -771,7 +771,7 @@ async function doMakeImage(reqBody, batchCount) {
                     if (stepUpdate.step === undefined) {
                         finalJSON += jsonStr
                     } else {
-                        let batchSize = parseInt(reqBody['num_inference_steps'])
+                        let batchSize = stepUpdate.total_steps
                         let overallStepCount = stepUpdate.step + batchesDone * batchSize
                         let totalSteps = batchCount * batchSize
                         let percent = 100 * (overallStepCount / totalSteps)
diff --git a/ui/sd_internal/runtime.py b/ui/sd_internal/runtime.py
index 9fa88e5b..f03a0d34 100644
--- a/ui/sd_internal/runtime.py
+++ b/ui/sd_internal/runtime.py
@@ -369,7 +369,8 @@ def do_mk_img(req: Request):
                         partial_x_samples = x_samples
 
                         if req.stream_progress_updates:
-                            progress = {"step": i, "total_steps": opt_ddim_steps}
+                            n_steps = opt_ddim_steps if req.init_image is None else t_enc
+                            progress = {"step": i, "total_steps": n_steps}
 
                             if req.stream_image_progress and i % 5 == 0:
                                 partial_images = []

From ca4229c7324362c79760b98bcde8dc92f3152b45 Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 17:07:54 +0530
Subject: [PATCH 7/8] Rearrange the UI, to use the bigger screen better, move
 the advanced settings to a different setting, links to help & community
 resources; cleaner image settings

---
 ui/index.html | 240 ++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 185 insertions(+), 55 deletions(-)

diff --git a/ui/index.html b/ui/index.html
index db393213..3ef2cedb 100644
--- a/ui/index.html
+++ b/ui/index.html
@@ -21,7 +21,8 @@
     }
     #prompt {
         width: 100%;
-        height: 50pt;
+        height: 65pt;
+        box-sizing: border-box;
     }
     @media screen and (max-width: 600px) {
         #prompt {
@@ -47,14 +48,14 @@
         font-family: Verdana;
         font-size: 8pt;
     }
-    #editor-settings-entries {
+    .settings-box ul {
         font-size: 9pt;
         margin-bottom: 5px;
         padding-left: 10px;
         list-style-type: none;
     }
-    #editor-settings-entries li {
-        padding-bottom: 3pt;
+    .settings-box li {
+        padding-bottom: 4pt;
     }
     .editor-slider {
         transform: translateY(30%);
@@ -107,23 +108,26 @@
     }
 
     #container {
-        width: 75%;
+        width: 90%;
         margin-left: auto;
         margin-right: auto;
     }
-    @media screen and (max-width: 1400px) {
+    @media screen and (max-width: 1800px) {
         #container {
             width: 100%;
         }
     }
-    #meta small {
+    #logo small {
         font-size: 11pt;
     }
     #editor {
         padding: 5px;
     }
     #editor label {
-        font-weight: bold;
+        font-weight: normal;
+    }
+    .settings-box label small {
+        color: rgb(153, 153, 153);
     }
     #preview {
         padding: 5px;
@@ -174,6 +178,9 @@
     .col-50 {
         flex: 50%;
     }
+    .col-fixed-10 {
+        flex: 0 0 400pt;
+    }
     .col-free {
         flex: 1;
     }
@@ -252,16 +259,19 @@
         display: none;
     }
     #server-status {
+        display: inline;
         float: right;
+        transform: translateY(-5pt);
     }
     #server-status-color {
-        width: 8pt;
+        /* width: 8pt;
         height: 8pt;
-        border-radius: 4pt;
-        background-color: rgb(128, 87, 0);
+        border-radius: 4pt; */
+        font-size: 14pt;
+        color: rgb(128, 87, 0);
         /* background-color: rgb(197, 1, 1); */
-        float: left;
-        transform: translateY(15%);
+        /* transform: translateY(15%); */
+        display: inline;
     }
     #server-status-msg {
         color: rgb(128, 87, 0);
@@ -290,21 +300,141 @@
     #enable_mask {
         margin-top: 8pt;
     }
+
+    #top-nav {
+        padding-top: 3pt;
+        padding-bottom: 15pt;
+    }
+    #top-nav .icon {
+        padding-right: 4pt;
+        font-size: 14pt;
+        transform: translateY(1pt);
+    }
+    #logo {
+        display: inline;
+    }
+    #logo h1 {
+        display: inline;
+    }
+    #top-nav-items {
+        list-style-type: none;
+        display: inline;
+        float: right;
+    }
+    #top-nav-items > li {
+        float: left;
+        display: inline;
+        padding-left: 20pt;
+        cursor: default;
+    }
+    #initial-text {
+        padding-top: 15pt;
+        padding-left: 4pt;
+    }
+    .settings-subheader {
+        font-size: 10pt;
+        font-weight: bold;
+    }
+    .pl-5 {
+        padding-left: 5pt;
+    }
+    #system-settings {
+        width: 360pt;
+        transform: translateX(-100%) translateX(70pt);
+
+        padding-top: 10pt;
+        padding-bottom: 10pt;
+    }
+    #system-settings ul {
+        margin: 0;
+        padding: 0;
+    }
+    #system-settings li {
+        padding-left: 5pt;
+    }
+    #community-links {
+        list-style-type: none;
+        margin: 0;
+        padding: 12pt;
+        padding-bottom: 0pt;
+        transform: translateX(-15%);
+    }
+    #community-links li {
+        padding-bottom: 12pt;
+        display: block;
+        font-size: 10pt;
+    }
+    #community-links li .fa-fw {
+        padding-right: 2pt;
+    }
+    #community-links li a {
+        color: white;
+        text-decoration: none;
+    }
+    .dropdown {
+        overflow: hidden;
+    }
+    .dropdown-content {
+        display: none;
+        position: absolute;
+        z-index: 2;
+
+        background: rgb(18, 18, 19);
+        border: 2px solid rgb(37, 38, 41);
+        border-radius: 7px;
+        padding: 5px;
+        margin-bottom: 15px;
+        box-shadow: 0 20px 28px 0 rgba(0, 0, 0, 0.15), 0 6px 20px 0 rgba(0, 0, 0, 0.15);
+    }
+    .dropdown:hover .dropdown-content {
+        display: block;
+    }
     </style>
+    <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.2.0/css/all.min.css">
     <link rel="stylesheet" href="/media/drawingboard.min.css">
     <script src="/media/jquery-3.6.1.min.js"></script>
     <script src="/media/drawingboard.min.js"></script>
 </html>
 <body>
 <div id="container">
-    <div class="flex-container">
-        <div id="editor" class="col-50">
-            <div id="meta">
-                <div id="server-status">
-                    <div id="server-status-color">&nbsp;</div>
-                    <span id="server-status-msg">Stable Diffusion is starting..</span>
+    <div id="top-nav">
+        <div id="logo">
+            <h1>Stable Diffusion UI <small>v2.16 <span id="updateBranchLabel"></span></small></h1>
+        </div>
+        <ul id="top-nav-items">
+            <li class="dropdown">
+                <span><i class="fa fa-comments icon"></i> Help & Community</span>
+                <ul id="community-links" class="dropdown-content">
+                    <li><a href="https://github.com/cmdr2/stable-diffusion-ui/blob/main/Troubleshooting.md" target="_blank"><i class="fa-solid fa-circle-question fa-fw"></i> Usual problems and solutions</a></li>
+                    <li><a href="https://discord.com/invite/u9yhsFmEkB" target="_blank"><i class="fa-brands fa-discord fa-fw"></i> Discord user community</a></li>
+                    <li><a href="https://github.com/cmdr2/stable-diffusion-ui" target="_blank"><i class="fa-brands fa-github fa-fw"></i> Source code on GitHub</a></li>
+                </ul>
+            </li>
+            <li class="dropdown">
+                <span><i class="fa fa-gear icon"></i> Settings</span>
+                <div id="system-settings" class="panel-box settings-box dropdown-content">
+                    <ul id="system-settings-entries">
+                        <li><b class="settings-subheader">System Settings</b></li>
+                        <br/>
+                        <li><input id="save_to_disk" name="save_to_disk" type="checkbox"> <label for="save_to_disk">Automatically save to <input id="diskPath" name="diskPath" size="40" disabled></label></li>
+                        <li><input id="sound_toggle" name="sound_toggle" type="checkbox" checked> <label for="sound_toggle">Play sound on task completion</label></li>
+                        <li><input id="turbo" name="turbo" type="checkbox" checked> <label for="turbo">Turbo mode <small>(generates images faster, but uses an additional 1 GB of GPU memory)</small></label></li>
+                        <li><input id="use_cpu" name="use_cpu" type="checkbox"> <label for="use_cpu">Use CPU instead of GPU <small>(warning: this will be *very* slow)</small></label></li>
+                        <li><input id="use_full_precision" name="use_full_precision" type="checkbox"> <label for="use_full_precision">Use full precision <small>(for GPU-only. warning: this will consume more VRAM)</small></label></li>
+                        <!-- <li><input id="allow_nsfw" name="allow_nsfw" type="checkbox"> <label for="allow_nsfw">Allow NSFW Content (You confirm you are above 18 years of age)</label></li> -->
+                        <br/>
+                        <li><input id="use_beta_channel" name="use_beta_channel" type="checkbox"> <label for="use_beta_channel">🔥Beta channel. Get the latest features immediately (but could be less stable). Please restart the program after changing this.</label></li>
+                    </ul>
                 </div>
-                <h1>Stable Diffusion UI <small>v2.15 <span id="updateBranchLabel"></span></small></h1>
+            </li>
+        </ul>
+    </div>
+
+    <div class="flex-container">
+        <div id="editor" class="col-fixed-10">
+            <div id="server-status">
+                <div id="server-status-color">●</div>
+                <span id="server-status-msg">Stable Diffusion is starting..</span>
             </div>
             <div id="editor-inputs">
                 <div id="editor-inputs-prompt" class="row">
@@ -337,23 +467,13 @@
 
             <div class="line-separator">&nbsp;</div>
 
-            <div id="editor-settings" class="panel-box">
-                <h4 class="collapsible">Advanced Settings</h4>
+            <div id="editor-settings" class="panel-box settings-box">
+                <h4 class="collapsible">Image Settings</h4>
                 <ul id="editor-settings-entries" class="collapsible-content">
-                    <li><input id="stream_image_progress" name="stream_image_progress" type="checkbox"> <label for="stream_image_progress">Show a live preview of the image (consumes more VRAM, slightly slower image generation)</label></li>
-                    <li><input id="use_face_correction" name="use_face_correction" type="checkbox" checked> <label for="use_face_correction">Fix incorrect faces and eyes (uses GFPGAN)</label></li>
-                    <li>
-                        <input id="use_upscale" name="use_upscale" type="checkbox"> <label for="use_upscale">Upscale the image to 4x resolution using </label>
-                        <select id="upscale_model" name="upscale_model">
-                            <option value="RealESRGAN_x4plus" selected>RealESRGAN_x4plus</option>
-                            <option value="RealESRGAN_x4plus_anime_6B">RealESRGAN_x4plus_anime_6B</option>
-                        </select>
-                    </li>
-                    <li><input id="show_only_filtered_image" name="show_only_filtered_image" type="checkbox" checked> <label for="show_only_filtered_image">Show only the corrected/upscaled image</label></li>
-                    <br/>
-                    <li><label for="seed">Seed:</label> <input id="seed" name="seed" size="10" value="30000"> <input id="random_seed" name="random_seed" type="checkbox" checked> <label for="random_seed">Random Image</label></li>
-                    <li><label for="num_outputs_total">Number of images to make:</label> <input id="num_outputs_total" name="num_outputs_total" value="1" size="4"> <label for="num_outputs_parallel">Generate in parallel:</label> <input id="num_outputs_parallel" name="num_outputs_parallel" value="1" size="4"> (images at once)</li>
-                    <li id="samplerSelection"><label for="sampler">Sampler:</label>
+                    <li><b class="settings-subheader">Image Settings</b></li>
+                    <li class="pl-5"><label for="seed">Seed:</label> <input id="seed" name="seed" size="10" value="30000"> <input id="random_seed" name="random_seed" type="checkbox" checked> <label for="random_seed">Random Image</label></li>
+                    <li class="pl-5"><label for="num_outputs_total">Number of images to make:</label> <input id="num_outputs_total" name="num_outputs_total" value="1" size="1"> <label for="num_outputs_parallel">Generate in parallel:</label> <input id="num_outputs_parallel" name="num_outputs_parallel" value="1" size="1"> (images at once)</li>
+                    <li id="samplerSelection" class="pl-5"><label for="sampler">Sampler:</label>
                         <select id="sampler" name="sampler">
                             <option value="plms" selected>plms</option>
                             <option value="ddim">ddim</option>
@@ -365,7 +485,7 @@
                             <option value="lms">lms</option>
                         </select>
                     </li>
-                    <li><label for="width">Width:</label> 
+                    <li class="pl-5"><label>Image Size: </label>
                         <select id="width" name="width" value="512">
                             <option value="128">128 (*)</option>
                             <option value="192">192</option>
@@ -386,9 +506,7 @@
                             <option value="1536">1536</option>
                             <option value="1792">1792</option>
                             <option value="2048">2048</option>
-                        </select>
-                    </li>
-                    <li><label for="height">Height:</label> 
+                        </select> <label for="width"><small>(width)</small></label> 
                         <select id="height" name="height" value="512">
                             <option value="128">128 (*)</option>
                             <option value="192">192</option>
@@ -410,19 +528,27 @@
                             <option value="1792">1792</option>
                             <option value="2048">2048</option>
                         </select>
+                        <label for="height"><small>(height)</small></label>
                     </li>
-                    <li><label for="num_inference_steps">Number of inference steps:</label> <input id="num_inference_steps" name="num_inference_steps" size="4" value="50"></li>
-                    <li><label for="guidance_scale_slider">Guidance Scale:</label> <input id="guidance_scale_slider" name="guidance_scale_slider" class="editor-slider" value="75" type="range" min="10" max="200"> <input id="guidance_scale" name="guidance_scale" size="4"></li>
-                    <li><span id="prompt_strength_container"><label for="prompt_strength_slider">Prompt Strength:</label> <input id="prompt_strength_slider" name="prompt_strength_slider" class="editor-slider" value="80" type="range" min="0" max="99"> <input id="prompt_strength" name="prompt_strength" size="4"><br/></span></li>
-                    <li>&nbsp;</li>
-                    <li><input id="save_to_disk" name="save_to_disk" type="checkbox"> <label for="save_to_disk">Automatically save to <input id="diskPath" name="diskPath" size="40" disabled></label></li>
-                    <li><input id="sound_toggle" name="sound_toggle" type="checkbox" checked> <label for="sound_toggle">Play sound on task completion</label></li>
-                    <li><input id="turbo" name="turbo" type="checkbox" checked> <label for="turbo">Turbo mode (generates images faster, but uses an additional 1 GB of GPU memory)</label></li>
-                    <li><input id="use_cpu" name="use_cpu" type="checkbox"> <label for="use_cpu">Use CPU instead of GPU (warning: this will be *very* slow)</label></li>
-                    <li><input id="use_full_precision" name="use_full_precision" type="checkbox"> <label for="use_full_precision">Use full precision (for GPU-only. warning: this will consume more VRAM)</label></li>
-                    <!-- <li><input id="allow_nsfw" name="allow_nsfw" type="checkbox"> <label for="allow_nsfw">Allow NSFW Content (You confirm you are above 18 years of age)</label></li> -->
+                    <li class="pl-5"><label for="num_inference_steps">Number of inference steps:</label> <input id="num_inference_steps" name="num_inference_steps" size="4" value="50"></li>
+                    <li class="pl-5"><label for="guidance_scale_slider">Guidance Scale:</label> <input id="guidance_scale_slider" name="guidance_scale_slider" class="editor-slider" value="75" type="range" min="10" max="200"> <input id="guidance_scale" name="guidance_scale" size="4"></li>
+                    <li class="pl-5"><span id="prompt_strength_container"><label for="prompt_strength_slider">Prompt Strength:</label> <input id="prompt_strength_slider" name="prompt_strength_slider" class="editor-slider" value="80" type="range" min="0" max="99"> <input id="prompt_strength" name="prompt_strength" size="4"><br/></span></li>
+
                     <br/>
-                    <li><input id="use_beta_channel" name="use_beta_channel" type="checkbox"> <label for="use_beta_channel">🔥Beta channel. Get the latest features immediately (but could be less stable). Please restart the program after changing this.</label></li>
+
+                    <li><b class="settings-subheader">Render Settings</b></li>
+                    <li class="pl-5"><input id="stream_image_progress" name="stream_image_progress" type="checkbox"> <label for="stream_image_progress">Show a live preview of the image <small>(consumes more VRAM, slightly slower image generation)</small></label></li>
+                    <li class="pl-5"><input id="use_face_correction" name="use_face_correction" type="checkbox" checked> <label for="use_face_correction">Fix incorrect faces and eyes <small>(uses GFPGAN)</small></label></li>
+                    <li class="pl-5">
+                        <input id="use_upscale" name="use_upscale" type="checkbox"> <label for="use_upscale">Upscale the image to 4x resolution using </label>
+                        <select id="upscale_model" name="upscale_model">
+                            <option value="RealESRGAN_x4plus" selected>RealESRGAN_x4plus</option>
+                            <option value="RealESRGAN_x4plus_anime_6B">RealESRGAN_x4plus_anime_6B</option>
+                        </select>
+                    </li>
+                    <li class="pl-5"><input id="show_only_filtered_image" name="show_only_filtered_image" type="checkbox" checked> <label for="show_only_filtered_image">Show only the corrected/upscaled image</label></li>
+                    <br/>
+                    <li><small>The system-related settings have been moved to the top-right corner.</small></li>
                 </ul>
             </div>
 
@@ -433,8 +559,12 @@
             </div>
         </div>
 
-        <div id="preview" class="col-50">
-            <div id="preview-prompt">Type a prompt and press the "Make Image" button.<br/><br/>You can set an "Initial Image" if you want to guide the AI.<br/><br/>You can also add modifiers like "Realistic", "Pencil Sketch", "ArtStation" etc by browsing through the "Image Modifiers" section and selecting the desired modifiers.<br/><br/>Click "Advanced Settings" for additional settings like seed, image size, number of images to generate etc.<br/><br/>Enjoy! :)</div>
+        <div id="preview" class="col-free">
+            <div id="preview-prompt">
+                <div id="initial-text">
+                    Type a prompt and press the "Make Image" button.<br/><br/>You can set an "Initial Image" if you want to guide the AI.<br/><br/>You can also add modifiers like "Realistic", "Pencil Sketch", "ArtStation" etc by browsing through the "Image Modifiers" section and selecting the desired modifiers.<br/><br/>Click "Advanced Settings" for additional settings like seed, image size, number of images to generate etc.<br/><br/>Enjoy! :)
+                </div>
+            </div>
 
             <div id="outputMsg"></div>
             <div id="progressBar"></div>
@@ -655,12 +785,12 @@ function setStatus(statusType, msg, msgType) {
 
     if (msgType == 'error') {
         // msg = '<span style="color: red">' + msg + '<span>'
-        serverStatusColor.style.backgroundColor = 'red'
+        serverStatusColor.style.color = 'red'
         serverStatusMsg.style.color = 'red'
         serverStatusMsg.innerHTML = 'Stable Diffusion has stopped'
     } else if (msgType == 'success') {
         // msg = '<span style="color: green">' + msg + '<span>'
-        serverStatusColor.style.backgroundColor = 'green'
+        serverStatusColor.style.color = 'green'
         serverStatusMsg.style.color = 'green'
         serverStatusMsg.innerHTML = 'Stable Diffusion is ready'
         serverStatus = 'online'

From ae108bb603a882997c54e733e53052f3c58b0933 Mon Sep 17 00:00:00 2001
From: cmdr2 <secondary.cmdr2@gmail.com>
Date: Fri, 23 Sep 2022 17:19:11 +0530
Subject: [PATCH 8/8] Increase the max guidance scale

---
 ui/index.html | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/ui/index.html b/ui/index.html
index 3ef2cedb..5b2675f5 100644
--- a/ui/index.html
+++ b/ui/index.html
@@ -531,7 +531,7 @@
                         <label for="height"><small>(height)</small></label>
                     </li>
                     <li class="pl-5"><label for="num_inference_steps">Number of inference steps:</label> <input id="num_inference_steps" name="num_inference_steps" size="4" value="50"></li>
-                    <li class="pl-5"><label for="guidance_scale_slider">Guidance Scale:</label> <input id="guidance_scale_slider" name="guidance_scale_slider" class="editor-slider" value="75" type="range" min="10" max="200"> <input id="guidance_scale" name="guidance_scale" size="4"></li>
+                    <li class="pl-5"><label for="guidance_scale_slider">Guidance Scale:</label> <input id="guidance_scale_slider" name="guidance_scale_slider" class="editor-slider" value="75" type="range" min="10" max="500"> <input id="guidance_scale" name="guidance_scale" size="4"></li>
                     <li class="pl-5"><span id="prompt_strength_container"><label for="prompt_strength_slider">Prompt Strength:</label> <input id="prompt_strength_slider" name="prompt_strength_slider" class="editor-slider" value="80" type="range" min="0" max="99"> <input id="prompt_strength" name="prompt_strength" size="4"><br/></span></li>
 
                     <br/>
@@ -1321,8 +1321,8 @@ function updateGuidanceScale() {
 function updateGuidanceScaleSlider() {
     if (guidanceScaleField.value < 0) {
         guidanceScaleField.value = 0
-    } else if (guidanceScaleField.value > 20) {
-        guidanceScaleField.value = 20
+    } else if (guidanceScaleField.value > 50) {
+        guidanceScaleField.value = 50
     }
 
     guidanceScaleSlider.value = guidanceScaleField.value * 10