added highres fix feature

javascript/hints.js

@@ -72,6 +72,10 @@ titles = {
     "Checkpoint name": "Loads weights from checkpoint before making images. You can either use hash or a part of filename (as seen in settings) for checkpoint name. Recommended to use with Y axis for less switching.",
 
     "vram": "Torch active: Peak amount of VRAM used by Torch during generation, excluding cached data.\nTorch reserved: Peak amount of VRAM allocated by Torch, including all active and cached data.\nSys VRAM: Peak amount of VRAM allocation across all applications / total GPU VRAM (peak utilization%).",
+
+    "Highres. fix": "Use a two step process to partially create an image at smaller resolution, upscale, and then improve details in it without changing composition",
+    "Scale latent": "Uscale the image in latent space. Alternative is to produce the full image from latent representation, upscale that, and then move it back to latent space.",
+
 }
 
 

modules/processing.py

@@ -74,11 +74,12 @@ class StableDiffusionProcessing:
         self.overlay_images = overlay_images
         self.paste_to = None
         self.color_corrections = None
+        self.denoising_strength: float = 0
 
-    def init(self, seed):
+    def init(self, all_prompts, all_seeds, all_subseeds):
         pass
 
-    def sample(self, x, conditioning, unconditional_conditioning):
+    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength):
         raise NotImplementedError()
 
 
@@ -303,7 +304,7 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
     precision_scope = torch.autocast if cmd_opts.precision == "autocast" else contextlib.nullcontext
     ema_scope = (contextlib.nullcontext if cmd_opts.lowvram else p.sd_model.ema_scope)
     with torch.no_grad(), precision_scope("cuda"), ema_scope():
-        p.init(seed=all_seeds[0])
+        p.init(all_prompts, all_seeds, all_subseeds)
 
         if state.job_count == -1:
             state.job_count = p.n_iter
@@ -328,13 +329,10 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
                 for comment in model_hijack.comments:
                     comments[comment] = 1
 
-            # we manually generate all input noises because each one should have a specific seed
-            x = create_random_tensors([opt_C, p.height // opt_f, p.width // opt_f], seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength, seed_resize_from_h=p.seed_resize_from_h, seed_resize_from_w=p.seed_resize_from_w, p=p)
-
             if p.n_iter > 1:
                 shared.state.job = f"Batch {n+1} out of {p.n_iter}"
 
-            samples_ddim = p.sample(x=x, conditioning=c, unconditional_conditioning=uc)
+            samples_ddim = p.sample(conditioning=c, unconditional_conditioning=uc, seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength)
             if state.interrupted:
 
                 # if we are interruped, sample returns just noise
@@ -406,13 +404,64 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
 
 class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
     sampler = None
+    firstphase_width = 0
+    firstphase_height = 0
+    firstphase_width_truncated = 0
+    firstphase_height_truncated = 0
 
-    def init(self, seed):
+    def __init__(self, enable_hr=False, scale_latent=True, denoising_strength=0.75, **kwargs):
+        super().__init__(**kwargs)
+        self.enable_hr = enable_hr
+        self.scale_latent = scale_latent
+        self.denoising_strength = denoising_strength
+
+    def init(self, all_prompts, all_seeds, all_subseeds):
+        if self.enable_hr:
+            if state.job_count == -1:
+                state.job_count = self.n_iter * 2
+            else:
+                state.job_count = state.job_count * 2
+
+            desired_pixel_count = 512 * 512
+            actual_pixel_count = self.width * self.height
+            scale = math.sqrt(desired_pixel_count / actual_pixel_count)
+
+            self.firstphase_width = math.ceil(scale * self.width / 64) * 64
+            self.firstphase_height = math.ceil(scale * self.height / 64) * 64
+            self.firstphase_width_truncated = int(scale * self.width)
+            self.firstphase_height_truncated = int(scale * self.height)
+
+    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength):
         self.sampler = samplers[self.sampler_index].constructor(self.sd_model)
 
-    def sample(self, x, conditioning, unconditional_conditioning):
+        if not self.enable_hr:
-        samples_ddim = self.sampler.sample(self, x, conditioning, unconditional_conditioning)
+            x = create_random_tensors([opt_C, self.height // opt_f, self.width // opt_f], seeds=seeds, subseeds=subseeds, subseed_strength=self.subseed_strength, seed_resize_from_h=self.seed_resize_from_h, seed_resize_from_w=self.seed_resize_from_w, p=self)
-        return samples_ddim
+            samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning)
+            return samples
+
+        x = create_random_tensors([opt_C, self.firstphase_height // opt_f, self.firstphase_width // opt_f], seeds=seeds, subseeds=subseeds, subseed_strength=self.subseed_strength, seed_resize_from_h=self.seed_resize_from_h, seed_resize_from_w=self.seed_resize_from_w, p=self)
+        samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning)
+
+        truncate_x = (self.firstphase_width - self.firstphase_width_truncated) // opt_f
+        truncate_y = (self.firstphase_height - self.firstphase_height_truncated) // opt_f
+
+        samples = samples[:, :, truncate_y//2:samples.shape[2]-truncate_y//2, truncate_x//2:samples.shape[3]-truncate_x//2]
+
+        if self.scale_latent:
+            samples = torch.nn.functional.interpolate(samples, size=(self.height // opt_f, self.width // opt_f), mode="bilinear")
+        else:
+            decoded_samples = self.sd_model.decode_first_stage(samples)
+            decoded_samples = torch.nn.functional.interpolate(decoded_samples, size=(self.height, self.width), mode="bilinear")
+            samples = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(decoded_samples))
+
+        shared.state.nextjob()
+
+        self.sampler = samplers[self.sampler_index].constructor(self.sd_model)
+        noise = create_random_tensors(samples.shape[1:], seeds=seeds, subseeds=subseeds, subseed_strength=subseed_strength, seed_resize_from_h=self.seed_resize_from_h, seed_resize_from_w=self.seed_resize_from_w, p=self)
+        samples = self.sampler.sample_img2img(self, samples, noise, conditioning, unconditional_conditioning, steps=self.steps)
+
+        return samples
+
 
 class StableDiffusionProcessingImg2Img(StableDiffusionProcessing):
     sampler = None
@@ -435,7 +484,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing):
         self.mask = None
         self.nmask = None
 
-    def init(self, seed):
+    def init(self, all_prompts, all_seeds, all_subseeds):
         self.sampler = samplers_for_img2img[self.sampler_index].constructor(self.sd_model)
         crop_region = None
 
@@ -529,12 +578,15 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing):
             self.mask = torch.asarray(1.0 - latmask).to(shared.device).type(self.sd_model.dtype)
             self.nmask = torch.asarray(latmask).to(shared.device).type(self.sd_model.dtype)
 
+            # this needs to be fixed to be done in sample() using actual seeds for batches
             if self.inpainting_fill == 2:
-                self.init_latent = self.init_latent * self.mask + create_random_tensors(self.init_latent.shape[1:], [seed + x + 1 for x in range(self.init_latent.shape[0])]) * self.nmask
+                self.init_latent = self.init_latent * self.mask + create_random_tensors(self.init_latent.shape[1:], all_seeds[0:self.init_latent.shape[0]]) * self.nmask
             elif self.inpainting_fill == 3:
                 self.init_latent = self.init_latent * self.mask
 
-    def sample(self, x, conditioning, unconditional_conditioning):
+    def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength):
+        x = create_random_tensors([opt_C, self.height // opt_f, self.width // opt_f], seeds=seeds, subseeds=subseeds, subseed_strength=self.subseed_strength, seed_resize_from_h=self.seed_resize_from_h, seed_resize_from_w=self.seed_resize_from_w, p=self)
+
         samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning)
 
         if self.mask is not None:

modules/sd_samplers.py

@@ -38,9 +38,9 @@ samplers = [
 samplers_for_img2img = [x for x in samplers if x.name != 'PLMS']
 
 
-def setup_img2img_steps(p):
+def setup_img2img_steps(p, steps=None):
-    if opts.img2img_fix_steps:
+    if opts.img2img_fix_steps or steps is not None:
-        steps = int(p.steps / min(p.denoising_strength, 0.999))
+        steps = int((steps or p.steps) / min(p.denoising_strength, 0.999)) if p.denoising_strength > 0 else 0
         t_enc = p.steps - 1
     else:
         steps = p.steps
@@ -115,8 +115,8 @@ class VanillaStableDiffusionSampler:
         self.step += 1
         return res
 
-    def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning):
+    def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning, steps=None):
-        steps, t_enc = setup_img2img_steps(p)
+        steps, t_enc = setup_img2img_steps(p, steps)
 
         # existing code fails with cetain step counts, like 9
         try:
@@ -127,16 +127,16 @@ class VanillaStableDiffusionSampler:
         x1 = self.sampler.stochastic_encode(x, torch.tensor([t_enc] * int(x.shape[0])).to(shared.device), noise=noise)
 
         self.sampler.p_sample_ddim = self.p_sample_ddim_hook
-        self.mask = p.mask
+        self.mask = p.mask if hasattr(p, 'mask') else None
-        self.nmask = p.nmask
+        self.nmask = p.nmask if hasattr(p, 'nmask') else None
-        self.init_latent = p.init_latent
+        self.init_latent = x
         self.step = 0
 
         samples = self.sampler.decode(x1, conditioning, t_enc, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning)
 
         return samples
 
-    def sample(self, p, x, conditioning, unconditional_conditioning):
+    def sample(self, p, x, conditioning, unconditional_conditioning, steps=None):
         for fieldname in ['p_sample_ddim', 'p_sample_plms']:
             if hasattr(self.sampler, fieldname):
                 setattr(self.sampler, fieldname, self.p_sample_ddim_hook)
@@ -145,11 +145,13 @@ class VanillaStableDiffusionSampler:
         self.init_latent = None
         self.step = 0
 
+        steps = steps or p.steps
+
         # existing code fails with cetin step counts, like 9
         try:
-            samples_ddim, _ = self.sampler.sample(S=p.steps, conditioning=conditioning, batch_size=int(x.shape[0]), shape=x[0].shape, verbose=False, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning, x_T=x)
+            samples_ddim, _ = self.sampler.sample(S=steps, conditioning=conditioning, batch_size=int(x.shape[0]), shape=x[0].shape, verbose=False, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning, x_T=x)
         except Exception:
-            samples_ddim, _ = self.sampler.sample(S=p.steps+1, conditioning=conditioning, batch_size=int(x.shape[0]), shape=x[0].shape, verbose=False, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning, x_T=x)
+            samples_ddim, _ = self.sampler.sample(S=steps+1, conditioning=conditioning, batch_size=int(x.shape[0]), shape=x[0].shape, verbose=False, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning, x_T=x)
 
         return samples_ddim
 
@@ -186,7 +188,7 @@ class CFGDenoiser(torch.nn.Module):
         return denoised
 
 
-def extended_trange(count, *args, **kwargs):
+def extended_trange(sampler, count, *args, **kwargs):
     state.sampling_steps = count
     state.sampling_step = 0
 
@@ -194,6 +196,9 @@ def extended_trange(count, *args, **kwargs):
         if state.interrupted:
             break
 
+        if sampler.stop_at is not None and x > sampler.stop_at:
+            break
+
         yield x
 
         state.sampling_step += 1
@@ -222,6 +227,7 @@ class KDiffusionSampler:
         self.model_wrap_cfg = CFGDenoiser(self.model_wrap)
         self.sampler_noises = None
         self.sampler_noise_index = 0
+        self.stop_at = None
 
     def callback_state(self, d):
         store_latent(d["denoised"])
@@ -240,8 +246,8 @@ class KDiffusionSampler:
         self.sampler_noise_index += 1
         return res
 
-    def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning):
+    def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning, steps=None):
-        steps, t_enc = setup_img2img_steps(p)
+        steps, t_enc = setup_img2img_steps(p, steps)
 
         sigmas = self.model_wrap.get_sigmas(steps)
 
@@ -251,33 +257,36 @@ class KDiffusionSampler:
 
         sigma_sched = sigmas[steps - t_enc - 1:]
 
-        self.model_wrap_cfg.mask = p.mask
+        self.model_wrap_cfg.mask = p.mask if hasattr(p, 'mask') else None
-        self.model_wrap_cfg.nmask = p.nmask
+        self.model_wrap_cfg.nmask = p.nmask if hasattr(p, 'nmask') else None
-        self.model_wrap_cfg.init_latent = p.init_latent
+        self.model_wrap_cfg.init_latent = x
         self.model_wrap.step = 0
         self.sampler_noise_index = 0
 
         if hasattr(k_diffusion.sampling, 'trange'):
-            k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(*args, **kwargs)
+            k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(self, *args, **kwargs)
 
         if self.sampler_noises is not None:
             k_diffusion.sampling.torch = TorchHijack(self)
 
         return self.func(self.model_wrap_cfg, xi, sigma_sched, extra_args={'cond': conditioning, 'uncond': unconditional_conditioning, 'cond_scale': p.cfg_scale}, disable=False, callback=self.callback_state)
 
-    def sample(self, p, x, conditioning, unconditional_conditioning):
+    def sample(self, p, x, conditioning, unconditional_conditioning, steps=None):
-        sigmas = self.model_wrap.get_sigmas(p.steps)
+        steps = steps or p.steps
+
+        sigmas = self.model_wrap.get_sigmas(steps)
         x = x * sigmas[0]
 
         self.model_wrap_cfg.step = 0
         self.sampler_noise_index = 0
 
         if hasattr(k_diffusion.sampling, 'trange'):
-            k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(*args, **kwargs)
+            k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(self, *args, **kwargs)
 
         if self.sampler_noises is not None:
             k_diffusion.sampling.torch = TorchHijack(self)
 
-        samples_ddim = self.func(self.model_wrap_cfg, x, sigmas, extra_args={'cond': conditioning, 'uncond': unconditional_conditioning, 'cond_scale': p.cfg_scale}, disable=False, callback=self.callback_state)
+        samples = self.func(self.model_wrap_cfg, x, sigmas, extra_args={'cond': conditioning, 'uncond': unconditional_conditioning, 'cond_scale': p.cfg_scale}, disable=False, callback=self.callback_state)
-        return samples_ddim
+
+        return samples
 

modules/txt2img.py

@@ -6,7 +6,7 @@ import modules.processing as processing
 from modules.ui import plaintext_to_html
 
 
-def txt2img(prompt: str, negative_prompt: str, prompt_style: str, prompt_style2: str, steps: int, sampler_index: int, restore_faces: bool, tiling: bool, n_iter: int, batch_size: int, cfg_scale: float, seed: int, subseed: int, subseed_strength: float, seed_resize_from_h: int, seed_resize_from_w: int, height: int, width: int, *args):
+def txt2img(prompt: str, negative_prompt: str, prompt_style: str, prompt_style2: str, steps: int, sampler_index: int, restore_faces: bool, tiling: bool, n_iter: int, batch_size: int, cfg_scale: float, seed: int, subseed: int, subseed_strength: float, seed_resize_from_h: int, seed_resize_from_w: int, height: int, width: int, enable_hr: bool, scale_latent: bool, denoising_strength: float, *args):
     p = StableDiffusionProcessingTxt2Img(
         sd_model=shared.sd_model,
         outpath_samples=opts.outdir_samples or opts.outdir_txt2img_samples,
@@ -28,6 +28,9 @@ def txt2img(prompt: str, negative_prompt: str, prompt_style: str, prompt_style2:
         height=height,
         restore_faces=restore_faces,
         tiling=tiling,
+        enable_hr=enable_hr,
+        scale_latent=scale_latent,
+        denoising_strength=denoising_strength,
     )
 
     print(f"\ntxt2img: {prompt}", file=shared.progress_print_out)

modules/ui.py

@@ -327,6 +327,7 @@ def connect_reuse_seed(seed: gr.Number, reuse_seed: gr.Button, generation_info:
         outputs=[seed, dummy_component]
     )
 
+
 def create_toprow(is_img2img):
     with gr.Row(elem_id="toprow"):
         with gr.Column(scale=4):
@@ -392,6 +393,11 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
                 with gr.Row():
                     restore_faces = gr.Checkbox(label='Restore faces', value=False, visible=len(shared.face_restorers) > 1)
                     tiling = gr.Checkbox(label='Tiling', value=False)
+                    enable_hr = gr.Checkbox(label='Highres. fix', value=False)
+
+                with gr.Row(visible=False) as hr_options:
+                    scale_latent = gr.Checkbox(label='Scale latent', value=True)
+                    denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.7)
 
                 with gr.Row():
                     batch_count = gr.Slider(minimum=1, maximum=cmd_opts.max_batch_count, step=1, label='Batch count', value=1)
@@ -451,6 +457,9 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
                     subseed, subseed_strength, seed_resize_from_h, seed_resize_from_w,
                     height,
                     width,
+                    enable_hr,
+                    scale_latent,
+                    denoising_strength,
                 ] + custom_inputs,
                 outputs=[
                     txt2img_gallery,
@@ -463,6 +472,12 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
             txt2img_prompt.submit(**txt2img_args)
             submit.click(**txt2img_args)
 
+            enable_hr.change(
+                fn=lambda x: gr_show(x),
+                inputs=[enable_hr],
+                outputs=[hr_options],
+            )
+
             interrupt.click(
                 fn=lambda: shared.state.interrupt(),
                 inputs=[],