From f64af77adcd20fabe00e1e642512db9c6742ed23 Mon Sep 17 00:00:00 2001 From: brkirch Date: Mon, 23 Jan 2023 22:49:20 -0500 Subject: Fix different first gen with Approx NN previews The loading of the model for approx nn live previews can change the internal state of PyTorch, resulting in a different image. This can be avoided by preloading the approx nn model in advance. --- modules/processing.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'modules/processing.py') diff --git a/modules/processing.py b/modules/processing.py index bc541e2f..3bd590ba 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -13,7 +13,7 @@ from skimage import exposure from typing import Any, Dict, List, Optional import modules.sd_hijack -from modules import devices, prompt_parser, masking, sd_samplers, lowvram, generation_parameters_copypaste, script_callbacks, extra_networks +from modules import devices, prompt_parser, masking, sd_samplers, lowvram, generation_parameters_copypaste, script_callbacks, extra_networks, sd_vae_approx from modules.sd_hijack import model_hijack from modules.shared import opts, cmd_opts, state import modules.shared as shared @@ -568,6 +568,10 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: with devices.autocast(): p.init(p.all_prompts, p.all_seeds, p.all_subseeds) + if shared.opts.live_previews_enable and sd_samplers.approximation_indexes.get(shared.opts.show_progress_type, 0) == 1: + # preload approx nn model before sampling for a more deterministic result + sd_vae_approx.model() + if not p.disable_extra_networks: extra_networks.activate(p, extra_network_data) -- cgit v1.2.3 From 84d9ce30cb427759547bc7876ed80ab91787d175 Mon Sep 17 00:00:00 2001 From: brkirch Date: Tue, 24 Jan 2023 23:51:45 -0500 Subject: Add option for float32 sampling with float16 UNet This also handles type casting so that ROCm and MPS torch devices work correctly without --no-half. One cast is required for deepbooru in deepbooru_model.py, some explicit casting is required for img2img and inpainting. depth_model can't be converted to float16 or it won't work correctly on some systems (it's known to have issues on MPS) so in sd_models.py model.depth_model is removed for model.half(). --- README.md | 1 + modules/deepbooru_model.py | 4 +++- modules/devices.py | 2 ++ modules/processing.py | 15 ++++++++------- modules/sd_hijack_unet.py | 29 +++++++++++++++++++++++++++++ modules/sd_hijack_utils.py | 28 ++++++++++++++++++++++++++++ modules/sd_models.py | 10 ++++++++++ modules/shared.py | 1 + 8 files changed, 82 insertions(+), 8 deletions(-) create mode 100644 modules/sd_hijack_utils.py (limited to 'modules/processing.py') diff --git a/README.md b/README.md index 9c0cd1ef..a5611671 100644 --- a/README.md +++ b/README.md @@ -157,4 +157,5 @@ Licenses for borrowed code can be found in `Settings -> Licenses` screen, and al - DeepDanbooru - interrogator for anime diffusers https://github.com/KichangKim/DeepDanbooru - Security advice - RyotaK - Initial Gradio script - posted on 4chan by an Anonymous user. Thank you Anonymous user. +- Sampling in float32 precision from a float16 UNet - marunine for the idea, Birch-san for the example Diffusers implementation (https://github.com/Birch-san/diffusers-play/tree/92feee6) - (You) diff --git a/modules/deepbooru_model.py b/modules/deepbooru_model.py index edd40c81..83d2ff09 100644 --- a/modules/deepbooru_model.py +++ b/modules/deepbooru_model.py @@ -2,6 +2,8 @@ import torch import torch.nn as nn import torch.nn.functional as F +from modules import devices + # see https://github.com/AUTOMATIC1111/TorchDeepDanbooru for more @@ -196,7 +198,7 @@ class DeepDanbooruModel(nn.Module): t_358, = inputs t_359 = t_358.permute(*[0, 3, 1, 2]) t_359_padded = F.pad(t_359, [2, 3, 2, 3], value=0) - t_360 = self.n_Conv_0(t_359_padded) + t_360 = self.n_Conv_0(t_359_padded.to(self.n_Conv_0.bias.dtype) if devices.unet_needs_upcast else t_359_padded) t_361 = F.relu(t_360) t_361 = F.pad(t_361, [0, 1, 0, 1], value=float('-inf')) t_362 = self.n_MaxPool_0(t_361) diff --git a/modules/devices.py b/modules/devices.py index 524ec7af..0981ef80 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -79,6 +79,8 @@ cpu = torch.device("cpu") device = device_interrogate = device_gfpgan = device_esrgan = device_codeformer = None dtype = torch.float16 dtype_vae = torch.float16 +dtype_unet = torch.float16 +unet_needs_upcast = False def randn(seed, shape): diff --git a/modules/processing.py b/modules/processing.py index bc541e2f..2d186ba0 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -172,7 +172,8 @@ class StableDiffusionProcessing: midas_in = torch.from_numpy(transformed["midas_in"][None, ...]).to(device=shared.device) midas_in = repeat(midas_in, "1 ... -> n ...", n=self.batch_size) - conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(source_image)) + conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(source_image.to(devices.dtype_unet) if devices.unet_needs_upcast else source_image)) + conditioning_image = conditioning_image.float() if devices.unet_needs_upcast else conditioning_image conditioning = torch.nn.functional.interpolate( self.sd_model.depth_model(midas_in), size=conditioning_image.shape[2:], @@ -203,7 +204,7 @@ class StableDiffusionProcessing: # Create another latent image, this time with a masked version of the original input. # Smoothly interpolate between the masked and unmasked latent conditioning image using a parameter. - conditioning_mask = conditioning_mask.to(source_image.device).to(source_image.dtype) + conditioning_mask = conditioning_mask.to(device=source_image.device, dtype=source_image.dtype) conditioning_image = torch.lerp( source_image, source_image * (1.0 - conditioning_mask), @@ -211,7 +212,7 @@ class StableDiffusionProcessing: ) # Encode the new masked image using first stage of network. - conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(conditioning_image)) + conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(conditioning_image.to(devices.dtype_unet) if devices.unet_needs_upcast else conditioning_image)) # Create the concatenated conditioning tensor to be fed to `c_concat` conditioning_mask = torch.nn.functional.interpolate(conditioning_mask, size=latent_image.shape[-2:]) @@ -225,10 +226,10 @@ class StableDiffusionProcessing: # HACK: Using introspection as the Depth2Image model doesn't appear to uniquely # identify itself with a field common to all models. The conditioning_key is also hybrid. if isinstance(self.sd_model, LatentDepth2ImageDiffusion): - return self.depth2img_image_conditioning(source_image) + return self.depth2img_image_conditioning(source_image.float() if devices.unet_needs_upcast else source_image) if self.sampler.conditioning_key in {'hybrid', 'concat'}: - return self.inpainting_image_conditioning(source_image, latent_image, image_mask=image_mask) + return self.inpainting_image_conditioning(source_image.float() if devices.unet_needs_upcast else source_image, latent_image, image_mask=image_mask) # Dummy zero conditioning if we're not using inpainting or depth model. return latent_image.new_zeros(latent_image.shape[0], 5, 1, 1) @@ -610,7 +611,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" - with devices.autocast(): + with devices.autocast(disable=devices.unet_needs_upcast): samples_ddim = p.sample(conditioning=c, unconditional_conditioning=uc, seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength, prompts=prompts) x_samples_ddim = [decode_first_stage(p.sd_model, samples_ddim[i:i+1].to(dtype=devices.dtype_vae))[0].cpu() for i in range(samples_ddim.size(0))] @@ -988,7 +989,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image = torch.from_numpy(batch_images) image = 2. * image - 1. - image = image.to(shared.device) + image = image.to(device=shared.device, dtype=devices.dtype_unet if devices.unet_needs_upcast else None) self.init_latent = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(image)) diff --git a/modules/sd_hijack_unet.py b/modules/sd_hijack_unet.py index 18daf8c1..88c94e54 100644 --- a/modules/sd_hijack_unet.py +++ b/modules/sd_hijack_unet.py @@ -1,4 +1,8 @@ import torch +from packaging import version + +from modules import devices +from modules.sd_hijack_utils import CondFunc class TorchHijackForUnet: @@ -28,3 +32,28 @@ class TorchHijackForUnet: th = TorchHijackForUnet() + + +# Below are monkey patches to enable upcasting a float16 UNet for float32 sampling +def apply_model(orig_func, self, x_noisy, t, cond, **kwargs): + for y in cond.keys(): + cond[y] = [x.to(devices.dtype_unet) if isinstance(x, torch.Tensor) else x for x in cond[y]] + with devices.autocast(): + return orig_func(self, x_noisy.to(devices.dtype_unet), t.to(devices.dtype_unet), cond, **kwargs).float() + +class GELUHijack(torch.nn.GELU, torch.nn.Module): + def __init__(self, *args, **kwargs): + torch.nn.GELU.__init__(self, *args, **kwargs) + def forward(self, x): + if devices.unet_needs_upcast: + return torch.nn.GELU.forward(self.float(), x.float()).to(devices.dtype_unet) + else: + return torch.nn.GELU.forward(self, x) + +unet_needs_upcast = lambda *args, **kwargs: devices.unet_needs_upcast +CondFunc('ldm.models.diffusion.ddpm.LatentDiffusion.apply_model', apply_model, unet_needs_upcast) +CondFunc('ldm.modules.diffusionmodules.openaimodel.timestep_embedding', lambda orig_func, *args, **kwargs: orig_func(*args, **kwargs).to(devices.dtype_unet), unet_needs_upcast) +if version.parse(torch.__version__) <= version.parse("1.13.1"): + CondFunc('ldm.modules.diffusionmodules.util.GroupNorm32.forward', lambda orig_func, self, *args, **kwargs: orig_func(self.float(), *args, **kwargs), unet_needs_upcast) + CondFunc('ldm.modules.attention.GEGLU.forward', lambda orig_func, self, x: orig_func(self.float(), x.float()).to(devices.dtype_unet), unet_needs_upcast) + CondFunc('open_clip.transformer.ResidualAttentionBlock.__init__', lambda orig_func, *args, **kwargs: kwargs.update({'act_layer': GELUHijack}) and False or orig_func(*args, **kwargs), lambda _, *args, **kwargs: kwargs.get('act_layer') is None or kwargs['act_layer'] == torch.nn.GELU) diff --git a/modules/sd_hijack_utils.py b/modules/sd_hijack_utils.py new file mode 100644 index 00000000..f81b169a --- /dev/null +++ b/modules/sd_hijack_utils.py @@ -0,0 +1,28 @@ +import importlib + +class CondFunc: + def __new__(cls, orig_func, sub_func, cond_func): + self = super(CondFunc, cls).__new__(cls) + if isinstance(orig_func, str): + func_path = orig_func.split('.') + for i in range(len(func_path)-2, -1, -1): + try: + resolved_obj = importlib.import_module('.'.join(func_path[:i])) + break + except ImportError: + pass + for attr_name in func_path[i:-1]: + resolved_obj = getattr(resolved_obj, attr_name) + orig_func = getattr(resolved_obj, func_path[-1]) + setattr(resolved_obj, func_path[-1], lambda *args, **kwargs: self(*args, **kwargs)) + self.__init__(orig_func, sub_func, cond_func) + return lambda *args, **kwargs: self(*args, **kwargs) + def __init__(self, orig_func, sub_func, cond_func): + self.__orig_func = orig_func + self.__sub_func = sub_func + self.__cond_func = cond_func + def __call__(self, *args, **kwargs): + if not self.__cond_func or self.__cond_func(self.__orig_func, *args, **kwargs): + return self.__sub_func(self.__orig_func, *args, **kwargs) + else: + return self.__orig_func(*args, **kwargs) diff --git a/modules/sd_models.py b/modules/sd_models.py index 12083848..7c98991a 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -257,16 +257,24 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo): if not shared.cmd_opts.no_half: vae = model.first_stage_model + depth_model = getattr(model, 'depth_model', None) # with --no-half-vae, remove VAE from model when doing half() to prevent its weights from being converted to float16 if shared.cmd_opts.no_half_vae: model.first_stage_model = None + # with --upcast-sampling, don't convert the depth model weights to float16 + if shared.cmd_opts.upcast_sampling and depth_model: + model.depth_model = None model.half() model.first_stage_model = vae + if depth_model: + model.depth_model = depth_model devices.dtype = torch.float32 if shared.cmd_opts.no_half else torch.float16 devices.dtype_vae = torch.float32 if shared.cmd_opts.no_half or shared.cmd_opts.no_half_vae else torch.float16 + devices.dtype_unet = model.model.diffusion_model.dtype + devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16 model.first_stage_model.to(devices.dtype_vae) @@ -372,6 +380,8 @@ def load_model(checkpoint_info=None): if shared.cmd_opts.no_half: sd_config.model.params.unet_config.params.use_fp16 = False + elif shared.cmd_opts.upcast_sampling: + sd_config.model.params.unet_config.params.use_fp16 = True timer = Timer() diff --git a/modules/shared.py b/modules/shared.py index 5f713bee..4ce1209b 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -45,6 +45,7 @@ parser.add_argument("--lowram", action='store_true', help="load stable diffusion parser.add_argument("--always-batch-cond-uncond", action='store_true', help="disables cond/uncond batching that is enabled to save memory with --medvram or --lowvram") parser.add_argument("--unload-gfpgan", action='store_true', help="does not do anything.") parser.add_argument("--precision", type=str, help="evaluate at this precision", choices=["full", "autocast"], default="autocast") +parser.add_argument("--upcast-sampling", action='store_true', help="upcast sampling. No effect with --no-half. Usually produces similar results to --no-half with better performance while using less memory.") parser.add_argument("--share", action='store_true', help="use share=True for gradio and make the UI accessible through their site") parser.add_argument("--ngrok", type=str, help="ngrok authtoken, alternative to gradio --share", default=None) parser.add_argument("--ngrok-region", type=str, help="The region in which ngrok should start.", default="us") -- cgit v1.2.3 From e3b53fd295aca784253dfc8668ec87b537a72f43 Mon Sep 17 00:00:00 2001 From: brkirch Date: Wed, 25 Jan 2023 00:23:10 -0500 Subject: Add UI setting for upcasting attention to float32 Adds "Upcast cross attention layer to float32" option in Stable Diffusion settings. This allows for generating images using SD 2.1 models without --no-half or xFormers. In order to make upcasting cross attention layer optimizations possible it is necessary to indent several sections of code in sd_hijack_optimizations.py so that a context manager can be used to disable autocast. Also, even though Stable Diffusion (and Diffusers) only upcast q and k, unfortunately my findings were that most of the cross attention layer optimizations could not function unless v is upcast also. --- modules/devices.py | 6 +- modules/processing.py | 2 +- modules/sd_hijack_optimizations.py | 159 +++++++++++++++++++++++-------------- modules/shared.py | 1 + modules/sub_quadratic_attention.py | 4 +- 5 files changed, 108 insertions(+), 64 deletions(-) (limited to 'modules/processing.py') diff --git a/modules/devices.py b/modules/devices.py index 0981ef80..6b36622c 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -108,6 +108,10 @@ def autocast(disable=False): return torch.autocast("cuda") +def without_autocast(disable=False): + return torch.autocast("cuda", enabled=False) if torch.is_autocast_enabled() and not disable else contextlib.nullcontext() + + class NansException(Exception): pass @@ -125,7 +129,7 @@ def test_for_nans(x, where): message = "A tensor with all NaNs was produced in Unet." if not shared.cmd_opts.no_half: - message += " This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try using --no-half commandline argument to fix this." + message += " This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the \"Upcast cross attention layer to float32\" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this." elif where == "vae": message = "A tensor with all NaNs was produced in VAE." diff --git a/modules/processing.py b/modules/processing.py index 2d186ba0..a850082d 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -611,7 +611,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" - with devices.autocast(disable=devices.unet_needs_upcast): + with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): samples_ddim = p.sample(conditioning=c, unconditional_conditioning=uc, seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength, prompts=prompts) x_samples_ddim = [decode_first_stage(p.sd_model, samples_ddim[i:i+1].to(dtype=devices.dtype_vae))[0].cpu() for i in range(samples_ddim.size(0))] diff --git a/modules/sd_hijack_optimizations.py b/modules/sd_hijack_optimizations.py index 74452709..c02d954c 100644 --- a/modules/sd_hijack_optimizations.py +++ b/modules/sd_hijack_optimizations.py @@ -9,7 +9,7 @@ from torch import einsum from ldm.util import default from einops import rearrange -from modules import shared, errors +from modules import shared, errors, devices from modules.hypernetworks import hypernetwork from .sub_quadratic_attention import efficient_dot_product_attention @@ -52,18 +52,25 @@ def split_cross_attention_forward_v1(self, x, context=None, mask=None): q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q_in, k_in, v_in)) del q_in, k_in, v_in - r1 = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device) - for i in range(0, q.shape[0], 2): - end = i + 2 - s1 = einsum('b i d, b j d -> b i j', q[i:end], k[i:end]) - s1 *= self.scale + dtype = q.dtype + if shared.opts.upcast_attn: + q, k, v = q.float(), k.float(), v.float() - s2 = s1.softmax(dim=-1) - del s1 + with devices.without_autocast(disable=not shared.opts.upcast_attn): + r1 = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device, dtype=q.dtype) + for i in range(0, q.shape[0], 2): + end = i + 2 + s1 = einsum('b i d, b j d -> b i j', q[i:end], k[i:end]) + s1 *= self.scale + + s2 = s1.softmax(dim=-1) + del s1 + + r1[i:end] = einsum('b i j, b j d -> b i d', s2, v[i:end]) + del s2 + del q, k, v - r1[i:end] = einsum('b i j, b j d -> b i d', s2, v[i:end]) - del s2 - del q, k, v + r1 = r1.to(dtype) r2 = rearrange(r1, '(b h) n d -> b n (h d)', h=h) del r1 @@ -82,45 +89,52 @@ def split_cross_attention_forward(self, x, context=None, mask=None): k_in = self.to_k(context_k) v_in = self.to_v(context_v) - k_in *= self.scale - - del context, x - - q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q_in, k_in, v_in)) - del q_in, k_in, v_in - - r1 = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device, dtype=q.dtype) - - mem_free_total = get_available_vram() - - gb = 1024 ** 3 - tensor_size = q.shape[0] * q.shape[1] * k.shape[1] * q.element_size() - modifier = 3 if q.element_size() == 2 else 2.5 - mem_required = tensor_size * modifier - steps = 1 - - if mem_required > mem_free_total: - steps = 2 ** (math.ceil(math.log(mem_required / mem_free_total, 2))) - # print(f"Expected tensor size:{tensor_size/gb:0.1f}GB, cuda free:{mem_free_cuda/gb:0.1f}GB " - # f"torch free:{mem_free_torch/gb:0.1f} total:{mem_free_total/gb:0.1f} steps:{steps}") + dtype = q_in.dtype + if shared.opts.upcast_attn: + q_in, k_in, v_in = q_in.float(), k_in.float(), v_in if v_in.device.type == 'mps' else v_in.float() - if steps > 64: - max_res = math.floor(math.sqrt(math.sqrt(mem_free_total / 2.5)) / 8) * 64 - raise RuntimeError(f'Not enough memory, use lower resolution (max approx. {max_res}x{max_res}). ' - f'Need: {mem_required / 64 / gb:0.1f}GB free, Have:{mem_free_total / gb:0.1f}GB free') - - slice_size = q.shape[1] // steps if (q.shape[1] % steps) == 0 else q.shape[1] - for i in range(0, q.shape[1], slice_size): - end = i + slice_size - s1 = einsum('b i d, b j d -> b i j', q[:, i:end], k) - - s2 = s1.softmax(dim=-1, dtype=q.dtype) - del s1 - - r1[:, i:end] = einsum('b i j, b j d -> b i d', s2, v) - del s2 + with devices.without_autocast(disable=not shared.opts.upcast_attn): + k_in = k_in * self.scale + + del context, x + + q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q_in, k_in, v_in)) + del q_in, k_in, v_in + + r1 = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device, dtype=q.dtype) + + mem_free_total = get_available_vram() + + gb = 1024 ** 3 + tensor_size = q.shape[0] * q.shape[1] * k.shape[1] * q.element_size() + modifier = 3 if q.element_size() == 2 else 2.5 + mem_required = tensor_size * modifier + steps = 1 + + if mem_required > mem_free_total: + steps = 2 ** (math.ceil(math.log(mem_required / mem_free_total, 2))) + # print(f"Expected tensor size:{tensor_size/gb:0.1f}GB, cuda free:{mem_free_cuda/gb:0.1f}GB " + # f"torch free:{mem_free_torch/gb:0.1f} total:{mem_free_total/gb:0.1f} steps:{steps}") + + if steps > 64: + max_res = math.floor(math.sqrt(math.sqrt(mem_free_total / 2.5)) / 8) * 64 + raise RuntimeError(f'Not enough memory, use lower resolution (max approx. {max_res}x{max_res}). ' + f'Need: {mem_required / 64 / gb:0.1f}GB free, Have:{mem_free_total / gb:0.1f}GB free') + + slice_size = q.shape[1] // steps if (q.shape[1] % steps) == 0 else q.shape[1] + for i in range(0, q.shape[1], slice_size): + end = i + slice_size + s1 = einsum('b i d, b j d -> b i j', q[:, i:end], k) + + s2 = s1.softmax(dim=-1, dtype=q.dtype) + del s1 + + r1[:, i:end] = einsum('b i j, b j d -> b i d', s2, v) + del s2 + + del q, k, v - del q, k, v + r1 = r1.to(dtype) r2 = rearrange(r1, '(b h) n d -> b n (h d)', h=h) del r1 @@ -204,12 +218,20 @@ def split_cross_attention_forward_invokeAI(self, x, context=None, mask=None): context = default(context, x) context_k, context_v = hypernetwork.apply_hypernetworks(shared.loaded_hypernetworks, context) - k = self.to_k(context_k) * self.scale + k = self.to_k(context_k) v = self.to_v(context_v) del context, context_k, context_v, x - q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q, k, v)) - r = einsum_op(q, k, v) + dtype = q.dtype + if shared.opts.upcast_attn: + q, k, v = q.float(), k.float(), v if v.device.type == 'mps' else v.float() + + with devices.without_autocast(disable=not shared.opts.upcast_attn): + k = k * self.scale + + q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q, k, v)) + r = einsum_op(q, k, v) + r = r.to(dtype) return self.to_out(rearrange(r, '(b h) n d -> b n (h d)', h=h)) # -- End of code from https://github.com/invoke-ai/InvokeAI -- @@ -234,8 +256,14 @@ def sub_quad_attention_forward(self, x, context=None, mask=None): k = k.unflatten(-1, (h, -1)).transpose(1,2).flatten(end_dim=1) v = v.unflatten(-1, (h, -1)).transpose(1,2).flatten(end_dim=1) + dtype = q.dtype + if shared.opts.upcast_attn: + q, k = q.float(), k.float() + x = sub_quad_attention(q, k, v, q_chunk_size=shared.cmd_opts.sub_quad_q_chunk_size, kv_chunk_size=shared.cmd_opts.sub_quad_kv_chunk_size, chunk_threshold=shared.cmd_opts.sub_quad_chunk_threshold, use_checkpoint=self.training) + x = x.to(dtype) + x = x.unflatten(0, (-1, h)).transpose(1,2).flatten(start_dim=2) out_proj, dropout = self.to_out @@ -268,15 +296,16 @@ def sub_quad_attention(q, k, v, q_chunk_size=1024, kv_chunk_size=None, kv_chunk_ query_chunk_size = q_tokens kv_chunk_size = k_tokens - return efficient_dot_product_attention( - q, - k, - v, - query_chunk_size=q_chunk_size, - kv_chunk_size=kv_chunk_size, - kv_chunk_size_min = kv_chunk_size_min, - use_checkpoint=use_checkpoint, - ) + with devices.without_autocast(disable=q.dtype == v.dtype): + return efficient_dot_product_attention( + q, + k, + v, + query_chunk_size=q_chunk_size, + kv_chunk_size=kv_chunk_size, + kv_chunk_size_min = kv_chunk_size_min, + use_checkpoint=use_checkpoint, + ) def get_xformers_flash_attention_op(q, k, v): @@ -306,8 +335,14 @@ def xformers_attention_forward(self, x, context=None, mask=None): q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> b n h d', h=h), (q_in, k_in, v_in)) del q_in, k_in, v_in + dtype = q.dtype + if shared.opts.upcast_attn: + q, k = q.float(), k.float() + out = xformers.ops.memory_efficient_attention(q, k, v, attn_bias=None, op=get_xformers_flash_attention_op(q, k, v)) + out = out.to(dtype) + out = rearrange(out, 'b n h d -> b n (h d)', h=h) return self.to_out(out) @@ -378,10 +413,14 @@ def xformers_attnblock_forward(self, x): v = self.v(h_) b, c, h, w = q.shape q, k, v = map(lambda t: rearrange(t, 'b c h w -> b (h w) c'), (q, k, v)) + dtype = q.dtype + if shared.opts.upcast_attn: + q, k = q.float(), k.float() q = q.contiguous() k = k.contiguous() v = v.contiguous() out = xformers.ops.memory_efficient_attention(q, k, v, op=get_xformers_flash_attention_op(q, k, v)) + out = out.to(dtype) out = rearrange(out, 'b (h w) c -> b c h w', h=h) out = self.proj_out(out) return x + out diff --git a/modules/shared.py b/modules/shared.py index 4ce1209b..6a0b96cb 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -410,6 +410,7 @@ options_templates.update(options_section(('sd', "Stable Diffusion"), { "comma_padding_backtrack": OptionInfo(20, "Increase coherency by padding from the last comma within n tokens when using more than 75 tokens", gr.Slider, {"minimum": 0, "maximum": 74, "step": 1 }), "CLIP_stop_at_last_layers": OptionInfo(1, "Clip skip", gr.Slider, {"minimum": 1, "maximum": 12, "step": 1}), "extra_networks_default_multiplier": OptionInfo(1.0, "Multiplier for extra networks", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}), + "upcast_attn": OptionInfo(False, "Upcast cross attention layer to float32"), })) options_templates.update(options_section(('compatibility', "Compatibility"), { diff --git a/modules/sub_quadratic_attention.py b/modules/sub_quadratic_attention.py index 55052815..05595323 100644 --- a/modules/sub_quadratic_attention.py +++ b/modules/sub_quadratic_attention.py @@ -67,7 +67,7 @@ def _summarize_chunk( max_score, _ = torch.max(attn_weights, -1, keepdim=True) max_score = max_score.detach() exp_weights = torch.exp(attn_weights - max_score) - exp_values = torch.bmm(exp_weights, value) + exp_values = torch.bmm(exp_weights, value) if query.device.type == 'mps' else torch.bmm(exp_weights, value.to(exp_weights.dtype)).to(value.dtype) max_score = max_score.squeeze(-1) return AttnChunk(exp_values, exp_weights.sum(dim=-1), max_score) @@ -129,7 +129,7 @@ def _get_attention_scores_no_kv_chunking( ) attn_probs = attn_scores.softmax(dim=-1) del attn_scores - hidden_states_slice = torch.bmm(attn_probs, value) + hidden_states_slice = torch.bmm(attn_probs, value) if query.device.type == 'mps' else torch.bmm(attn_probs, value.to(attn_probs.dtype)).to(value.dtype) return hidden_states_slice -- cgit v1.2.3 From 57c1baa774d07060af0abbd2974c5f36c8cb63ac Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Wed, 25 Jan 2023 18:56:23 +0300 Subject: change to code for live preview fix on OSX to be bit more obvious --- modules/processing.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'modules/processing.py') diff --git a/modules/processing.py b/modules/processing.py index 3bd590ba..57c3db1b 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -568,8 +568,8 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: with devices.autocast(): p.init(p.all_prompts, p.all_seeds, p.all_subseeds) - if shared.opts.live_previews_enable and sd_samplers.approximation_indexes.get(shared.opts.show_progress_type, 0) == 1: - # preload approx nn model before sampling for a more deterministic result + # for OSX, loading the model during sampling changes the generated picture, so it is loaded here + if shared.opts.live_previews_enable and opts.show_progress_type == "Approx NN": sd_vae_approx.model() if not p.disable_extra_networks: -- cgit v1.2.3 From d1d6ce29831d1b067801c3206f314258de88f683 Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Wed, 25 Jan 2023 23:25:25 +0300 Subject: add edit_image_conditioning from my earlier edits in case there's an attempt to inegrate pix2pix properly this allows to use pix2pix model in img2img though it won't work well this way --- modules/processing.py | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'modules/processing.py') diff --git a/modules/processing.py b/modules/processing.py index 9e5a2f38..cb41288a 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -185,7 +185,12 @@ class StableDiffusionProcessing: conditioning = 2. * (conditioning - depth_min) / (depth_max - depth_min) - 1. return conditioning - def inpainting_image_conditioning(self, source_image, latent_image, image_mask = None): + def edit_image_conditioning(self, source_image): + conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(source_image)) + + return conditioning_image + + def inpainting_image_conditioning(self, source_image, latent_image, image_mask=None): self.is_using_inpainting_conditioning = True # Handle the different mask inputs @@ -228,6 +233,9 @@ class StableDiffusionProcessing: if isinstance(self.sd_model, LatentDepth2ImageDiffusion): return self.depth2img_image_conditioning(source_image.float() if devices.unet_needs_upcast else source_image) + if self.sd_model.cond_stage_key == "edit": + return self.edit_image_conditioning(source_image) + if self.sampler.conditioning_key in {'hybrid', 'concat'}: return self.inpainting_image_conditioning(source_image.float() if devices.unet_needs_upcast else source_image, latent_image, image_mask=image_mask) -- cgit v1.2.3 From 10421f93c3f7f7ce88cb40391b46d4e6664eff74 Mon Sep 17 00:00:00 2001 From: brkirch Date: Thu, 26 Jan 2023 00:34:38 -0500 Subject: Fix full previews, --no-half-vae --- modules/processing.py | 8 ++++---- modules/sd_hijack_utils.py | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'modules/processing.py') diff --git a/modules/processing.py b/modules/processing.py index cb41288a..92894d67 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -172,7 +172,7 @@ class StableDiffusionProcessing: midas_in = torch.from_numpy(transformed["midas_in"][None, ...]).to(device=shared.device) midas_in = repeat(midas_in, "1 ... -> n ...", n=self.batch_size) - conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(source_image.to(devices.dtype_unet) if devices.unet_needs_upcast else source_image)) + conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(source_image.to(devices.dtype_vae) if devices.unet_needs_upcast else source_image)) conditioning_image = conditioning_image.float() if devices.unet_needs_upcast else conditioning_image conditioning = torch.nn.functional.interpolate( self.sd_model.depth_model(midas_in), @@ -217,7 +217,7 @@ class StableDiffusionProcessing: ) # Encode the new masked image using first stage of network. - conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(conditioning_image.to(devices.dtype_unet) if devices.unet_needs_upcast else conditioning_image)) + conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(conditioning_image.to(devices.dtype_vae) if devices.unet_needs_upcast else conditioning_image)) # Create the concatenated conditioning tensor to be fed to `c_concat` conditioning_mask = torch.nn.functional.interpolate(conditioning_mask, size=latent_image.shape[-2:]) @@ -417,7 +417,7 @@ def create_random_tensors(shape, seeds, subseeds=None, subseed_strength=0.0, see def decode_first_stage(model, x): with devices.autocast(disable=x.dtype == devices.dtype_vae): - x = model.decode_first_stage(x) + x = model.decode_first_stage(x.to(devices.dtype_vae) if devices.unet_needs_upcast else x) return x @@ -1001,7 +1001,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image = torch.from_numpy(batch_images) image = 2. * image - 1. - image = image.to(device=shared.device, dtype=devices.dtype_unet if devices.unet_needs_upcast else None) + image = image.to(device=shared.device, dtype=devices.dtype_vae if devices.unet_needs_upcast else None) self.init_latent = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(image)) diff --git a/modules/sd_hijack_utils.py b/modules/sd_hijack_utils.py index f81b169a..f8684475 100644 --- a/modules/sd_hijack_utils.py +++ b/modules/sd_hijack_utils.py @@ -5,7 +5,7 @@ class CondFunc: self = super(CondFunc, cls).__new__(cls) if isinstance(orig_func, str): func_path = orig_func.split('.') - for i in range(len(func_path)-2, -1, -1): + for i in range(len(func_path)-1, -1, -1): try: resolved_obj = importlib.import_module('.'.join(func_path[:i])) break -- cgit v1.2.3 From 7a14c8ab45da8a681792a6331d48a88dd684a0a9 Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Thu, 26 Jan 2023 23:29:27 +0300 Subject: add an option to enable sections from extras tab in txt2img/img2img fix some style inconsistenices --- modules/processing.py | 7 +++++- modules/scripts.py | 32 ++++++++++++++++++++++---- modules/scripts_auto_postprocessing.py | 42 ++++++++++++++++++++++++++++++++++ modules/scripts_postprocessing.py | 11 ++++++--- modules/shared.py | 15 ++++-------- modules/shared_items.py | 10 ++++++++ modules/ui_components.py | 8 +++++++ scripts/postprocessing_upscale.py | 25 ++++++++++++++++++++ style.css | 6 +---- 9 files changed, 133 insertions(+), 23 deletions(-) create mode 100644 modules/scripts_auto_postprocessing.py create mode 100644 modules/shared_items.py (limited to 'modules/processing.py') diff --git a/modules/processing.py b/modules/processing.py index 92894d67..262806a1 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -13,7 +13,7 @@ from skimage import exposure from typing import Any, Dict, List, Optional import modules.sd_hijack -from modules import devices, prompt_parser, masking, sd_samplers, lowvram, generation_parameters_copypaste, script_callbacks, extra_networks, sd_vae_approx +from modules import devices, prompt_parser, masking, sd_samplers, lowvram, generation_parameters_copypaste, script_callbacks, extra_networks, sd_vae_approx, scripts from modules.sd_hijack import model_hijack from modules.shared import opts, cmd_opts, state import modules.shared as shared @@ -658,6 +658,11 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: image = Image.fromarray(x_sample) + if p.scripts is not None: + pp = scripts.PostprocessImageArgs(image) + p.scripts.postprocess_image(p, pp) + image = pp.image + if p.color_corrections is not None and i < len(p.color_corrections): if opts.save and not p.do_not_save_samples and opts.save_images_before_color_correction: image_without_cc = apply_overlay(image, p.paste_to, i, p.overlay_images) diff --git a/modules/scripts.py b/modules/scripts.py index 03907a63..6e9dc0c0 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -6,12 +6,16 @@ from collections import namedtuple import gradio as gr -from modules.processing import StableDiffusionProcessing from modules import shared, paths, script_callbacks, extensions, script_loading, scripts_postprocessing AlwaysVisible = object() +class PostprocessImageArgs: + def __init__(self, image): + self.image = image + + class Script: filename = None args_from = None @@ -65,7 +69,7 @@ class Script: args contains all values returned by components from ui() """ - raise NotImplementedError() + pass def process(self, p, *args): """ @@ -100,6 +104,13 @@ class Script: pass + def postprocess_image(self, p, pp: PostprocessImageArgs, *args): + """ + Called for every image after it has been generated. + """ + + pass + def postprocess(self, p, processed, *args): """ This function is called after processing ends for AlwaysVisible scripts. @@ -247,11 +258,15 @@ class ScriptRunner: self.infotext_fields = [] def initialize_scripts(self, is_img2img): + from modules import scripts_auto_postprocessing + self.scripts.clear() self.alwayson_scripts.clear() self.selectable_scripts.clear() - for script_class, path, basedir, script_module in scripts_data: + auto_processing_scripts = scripts_auto_postprocessing.create_auto_preprocessing_script_data() + + for script_class, path, basedir, script_module in auto_processing_scripts + scripts_data: script = script_class() script.filename = path script.is_txt2img = not is_img2img @@ -332,7 +347,7 @@ class ScriptRunner: return inputs - def run(self, p: StableDiffusionProcessing, *args): + def run(self, p, *args): script_index = args[0] if script_index == 0: @@ -386,6 +401,15 @@ class ScriptRunner: print(f"Error running postprocess_batch: {script.filename}", file=sys.stderr) print(traceback.format_exc(), file=sys.stderr) + def postprocess_image(self, p, pp: PostprocessImageArgs): + for script in self.alwayson_scripts: + try: + script_args = p.script_args[script.args_from:script.args_to] + script.postprocess_image(p, pp, *script_args) + except Exception: + print(f"Error running postprocess_batch: {script.filename}", file=sys.stderr) + print(traceback.format_exc(), file=sys.stderr) + def before_component(self, component, **kwargs): for script in self.scripts: try: diff --git a/modules/scripts_auto_postprocessing.py b/modules/scripts_auto_postprocessing.py new file mode 100644 index 00000000..30d6d658 --- /dev/null +++ b/modules/scripts_auto_postprocessing.py @@ -0,0 +1,42 @@ +from modules import scripts, scripts_postprocessing, shared + + +class ScriptPostprocessingForMainUI(scripts.Script): + def __init__(self, script_postproc): + self.script: scripts_postprocessing.ScriptPostprocessing = script_postproc + self.postprocessing_controls = None + + def title(self): + return self.script.name + + def show(self, is_img2img): + return scripts.AlwaysVisible + + def ui(self, is_img2img): + self.postprocessing_controls = self.script.ui() + return self.postprocessing_controls.values() + + def postprocess_image(self, p, script_pp, *args): + args_dict = {k: v for k, v in zip(self.postprocessing_controls, args)} + + pp = scripts_postprocessing.PostprocessedImage(script_pp.image) + pp.info = {} + self.script.process(pp, **args_dict) + p.extra_generation_params.update(pp.info) + script_pp.image = pp.image + + +def create_auto_preprocessing_script_data(): + from modules import scripts + + res = [] + + for name in shared.opts.postprocessing_enable_in_main_ui: + script = next(iter([x for x in scripts.postprocessing_scripts_data if x.script_class.name == name]), None) + if script is None: + continue + + constructor = lambda s=script: ScriptPostprocessingForMainUI(s.script_class()) + res.append(scripts.ScriptClassData(script_class=constructor, path=script.path, basedir=script.basedir, module=script.module)) + + return res diff --git a/modules/scripts_postprocessing.py b/modules/scripts_postprocessing.py index 25de02d0..ce0ebb61 100644 --- a/modules/scripts_postprocessing.py +++ b/modules/scripts_postprocessing.py @@ -46,6 +46,8 @@ class ScriptPostprocessing: pass + + def wrap_call(func, filename, funcname, *args, default=None, **kwargs): try: res = func(*args, **kwargs) @@ -68,6 +70,9 @@ class ScriptPostprocessingRunner: script: ScriptPostprocessing = script_class() script.filename = path + if script.name == "Simple Upscale": + continue + self.scripts.append(script) def create_script_ui(self, script, inputs): @@ -87,12 +92,11 @@ class ScriptPostprocessingRunner: import modules.scripts self.initialize_scripts(modules.scripts.postprocessing_scripts_data) - scripts_order = [x.lower().strip() for x in shared.opts.postprocessing_scipts_order.split(",")] + scripts_order = shared.opts.postprocessing_operation_order def script_score(name): - name = name.lower() for i, possible_match in enumerate(scripts_order): - if possible_match in name: + if possible_match == name: return i return len(self.scripts) @@ -145,3 +149,4 @@ class ScriptPostprocessingRunner: def image_changed(self): for script in self.scripts_in_preferred_order(): script.image_changed() + diff --git a/modules/shared.py b/modules/shared.py index 6a0b96cb..cdeed55d 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -13,8 +13,8 @@ import modules.interrogate import modules.memmon import modules.styles import modules.devices as devices -from modules import localization, sd_vae, extensions, script_loading, errors, ui_components -from modules.paths import models_path, script_path, sd_path +from modules import localization, sd_vae, extensions, script_loading, errors, ui_components, shared_items +from modules.paths import models_path, script_path demo = None @@ -264,12 +264,6 @@ interrogator = modules.interrogate.InterrogateModels("interrogate") face_restorers = [] - -def realesrgan_models_names(): - import modules.realesrgan_model - return [x.name for x in modules.realesrgan_model.get_realesrgan_models(None)] - - class OptionInfo: def __init__(self, default=None, label="", component=None, component_args=None, onchange=None, section=None, refresh=None): self.default = default @@ -360,7 +354,7 @@ options_templates.update(options_section(('saving-to-dirs', "Saving to a directo options_templates.update(options_section(('upscaling', "Upscaling"), { "ESRGAN_tile": OptionInfo(192, "Tile size for ESRGAN upscalers. 0 = no tiling.", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), "ESRGAN_tile_overlap": OptionInfo(8, "Tile overlap, in pixels for ESRGAN upscalers. Low values = visible seam.", gr.Slider, {"minimum": 0, "maximum": 48, "step": 1}), - "realesrgan_enabled_models": OptionInfo(["R-ESRGAN 4x+", "R-ESRGAN 4x+ Anime6B"], "Select which Real-ESRGAN models to show in the web UI. (Requires restart)", gr.CheckboxGroup, lambda: {"choices": realesrgan_models_names()}), + "realesrgan_enabled_models": OptionInfo(["R-ESRGAN 4x+", "R-ESRGAN 4x+ Anime6B"], "Select which Real-ESRGAN models to show in the web UI. (Requires restart)", gr.CheckboxGroup, lambda: {"choices": shared_items.realesrgan_models_names()}), "upscaler_for_img2img": OptionInfo(None, "Upscaler for img2img", gr.Dropdown, lambda: {"choices": [x.name for x in sd_upscalers]}), })) @@ -483,7 +477,8 @@ options_templates.update(options_section(('sampler-params', "Sampler parameters" })) options_templates.update(options_section(('postprocessing', "Postprocessing"), { - 'postprocessing_scipts_order': OptionInfo("upscale, gfpgan, codeformer", "Postprocessing operation order"), + 'postprocessing_enable_in_main_ui': OptionInfo([], "Enable postprocessing operations in txt2img and img2img tabs", ui_components.DropdownMulti, lambda: {"choices": [x.name for x in shared_items.postprocessing_scripts()]}), + 'postprocessing_operation_order': OptionInfo([], "Postprocessing operation order", ui_components.DropdownMulti, lambda: {"choices": [x.name for x in shared_items.postprocessing_scripts()]}), 'upscaling_max_images_in_cache': OptionInfo(5, "Maximum number of images in upscaling cache", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}), })) diff --git a/modules/shared_items.py b/modules/shared_items.py new file mode 100644 index 00000000..b5d480c9 --- /dev/null +++ b/modules/shared_items.py @@ -0,0 +1,10 @@ + + +def realesrgan_models_names(): + import modules.realesrgan_model + return [x.name for x in modules.realesrgan_model.get_realesrgan_models(None)] + +def postprocessing_scripts(): + import modules.scripts + + return modules.scripts.scripts_postproc.scripts \ No newline at end of file diff --git a/modules/ui_components.py b/modules/ui_components.py index 9aec3097..284ca0cf 100644 --- a/modules/ui_components.py +++ b/modules/ui_components.py @@ -48,3 +48,11 @@ class FormColorPicker(gr.ColorPicker, gr.components.FormComponent): def get_block_name(self): return "colorpicker" + +class DropdownMulti(gr.Dropdown): + """Same as gr.Dropdown but always multiselect""" + def __init__(self, **kwargs): + super().__init__(multiselect=True, **kwargs) + + def get_block_name(self): + return "dropdown" diff --git a/scripts/postprocessing_upscale.py b/scripts/postprocessing_upscale.py index 095d29b2..8842bd91 100644 --- a/scripts/postprocessing_upscale.py +++ b/scripts/postprocessing_upscale.py @@ -104,3 +104,28 @@ class ScriptPostprocessingUpscale(scripts_postprocessing.ScriptPostprocessing): def image_changed(self): upscale_cache.clear() + + +class ScriptPostprocessingUpscaleSimple(ScriptPostprocessingUpscale): + name = "Simple Upscale" + order = 900 + + def ui(self): + with FormRow(): + upscaler_name = gr.Dropdown(label='Upscaler', choices=[x.name for x in shared.sd_upscalers], value=shared.sd_upscalers[0].name) + upscale_by = gr.Slider(minimum=0.05, maximum=8.0, step=0.05, label="Upscale by", value=2) + + return { + "upscale_by": upscale_by, + "upscaler_name": upscaler_name, + } + + def process(self, pp: scripts_postprocessing.PostprocessedImage, upscale_by=2.0, upscaler_name=None): + if upscaler_name is None or upscaler_name == "None": + return + + upscaler1 = next(iter([x for x in shared.sd_upscalers if x.name == upscaler_name]), None) + assert upscaler1, f'could not find upscaler named {upscaler_name}' + + pp.image = self.upscale(pp.image, pp.info, upscaler1, 0, upscale_by, 0, 0, False) + pp.info[f"Postprocess upscaler"] = upscaler1.name diff --git a/style.css b/style.css index ec046f78..dd914104 100644 --- a/style.css +++ b/style.css @@ -164,7 +164,7 @@ min-height: 3.2em; } -#txt2img_styles ul, #img2img_styles ul{ +ul.list-none{ max-height: 35em; z-index: 2000; } @@ -714,9 +714,6 @@ footer { white-space: nowrap; min-width: auto; } -#txt2img_hires_fix{ - margin-left: -0.8em; -} #img2img_copy_to_img2img, #img2img_copy_to_sketch, #img2img_copy_to_inpaint, #img2img_copy_to_inpaint_sketch{ margin-left: 0em; @@ -744,7 +741,6 @@ footer { .dark .gr-compact{ background-color: rgb(31 41 55 / var(--tw-bg-opacity)); - margin-left: 0.8em; } .gr-compact{ -- cgit v1.2.3 From 5eee2ac39863f9e44591b50d0710dd2615416a13 Mon Sep 17 00:00:00 2001 From: Max Audron Date: Wed, 25 Jan 2023 17:15:42 +0100 Subject: add data-dir flag and set all user data directories based on it --- modules/extensions.py | 2 +- modules/generation_parameters_copypaste.py | 4 ++-- modules/gfpgan_model.py | 5 ++--- modules/hashes.py | 4 +++- modules/interrogate.py | 2 +- modules/paths.py | 10 +++++++++- modules/processing.py | 3 ++- modules/sd_models.py | 6 +++--- modules/sd_vae.py | 5 ++--- modules/shared.py | 11 ++++++----- modules/textual_inversion/preprocess.py | 5 ++--- modules/ui.py | 6 +++--- modules/ui_extensions.py | 2 +- modules/upscaler.py | 5 ++--- 14 files changed, 39 insertions(+), 31 deletions(-) (limited to 'modules/processing.py') diff --git a/modules/extensions.py b/modules/extensions.py index b522125c..92ee8144 100644 --- a/modules/extensions.py +++ b/modules/extensions.py @@ -7,7 +7,7 @@ import git from modules import paths, shared extensions = [] -extensions_dir = os.path.join(paths.script_path, "extensions") +extensions_dir = os.path.join(paths.data_path, "extensions") extensions_builtin_dir = os.path.join(paths.script_path, "extensions-builtin") diff --git a/modules/generation_parameters_copypaste.py b/modules/generation_parameters_copypaste.py index 46e12dc6..35f72808 100644 --- a/modules/generation_parameters_copypaste.py +++ b/modules/generation_parameters_copypaste.py @@ -6,7 +6,7 @@ import re from pathlib import Path import gradio as gr -from modules.shared import script_path +from modules.paths import data_path, script_path from modules import shared, ui_tempdir, script_callbacks import tempfile from PIL import Image @@ -289,7 +289,7 @@ Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 965400086, Size: 512x512, Model def connect_paste(button, paste_fields, input_comp, jsfunc=None): def paste_func(prompt): if not prompt and not shared.cmd_opts.hide_ui_dir_config: - filename = os.path.join(script_path, "params.txt") + filename = os.path.join(data_path, "params.txt") if os.path.exists(filename): with open(filename, "r", encoding="utf8") as file: prompt = file.read() diff --git a/modules/gfpgan_model.py b/modules/gfpgan_model.py index 1e2dbc32..fbe6215a 100644 --- a/modules/gfpgan_model.py +++ b/modules/gfpgan_model.py @@ -6,12 +6,11 @@ import facexlib import gfpgan import modules.face_restoration -from modules import shared, devices, modelloader -from modules.paths import models_path +from modules import paths, shared, devices, modelloader model_dir = "GFPGAN" user_path = None -model_path = os.path.join(models_path, model_dir) +model_path = os.path.join(paths.models_path, model_dir) model_url = "https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth" have_gfpgan = False loaded_gfpgan_model = None diff --git a/modules/hashes.py b/modules/hashes.py index b85a7580..819362a3 100644 --- a/modules/hashes.py +++ b/modules/hashes.py @@ -4,8 +4,10 @@ import os.path import filelock +from modules.paths import data_path -cache_filename = "cache.json" + +cache_filename = os.path.join(data_path, "cache.json") cache_data = None diff --git a/modules/interrogate.py b/modules/interrogate.py index c72ff694..cbb80683 100644 --- a/modules/interrogate.py +++ b/modules/interrogate.py @@ -12,7 +12,7 @@ from torchvision import transforms from torchvision.transforms.functional import InterpolationMode import modules.shared as shared -from modules import devices, paths, lowvram, modelloader, errors +from modules import devices, paths, shared, lowvram, modelloader, errors blip_image_eval_size = 384 clip_model_name = 'ViT-L/14' diff --git a/modules/paths.py b/modules/paths.py index 20b3e4d8..08e6f9b9 100644 --- a/modules/paths.py +++ b/modules/paths.py @@ -4,7 +4,15 @@ import sys import modules.safe script_path = os.path.dirname(os.path.dirname(os.path.realpath(__file__))) -models_path = os.path.join(script_path, "models") + +# Parse the --data-dir flag first so we can use it as a base for our other argument default values +parser = argparse.ArgumentParser() +parser.add_argument("--data-dir", type=str, default=os.path.dirname(os.path.dirname(os.path.realpath(__file__))), help="base path where all user data is stored",) +cmd_opts_pre = parser.parse_known_args()[0] +data_path = cmd_opts_pre.data_dir +models_path = os.path.join(data_path, "models") + +# data_path = cmd_opts_pre.data sys.path.insert(0, script_path) # search for directory of stable diffusion in following places diff --git a/modules/processing.py b/modules/processing.py index 262806a1..5072fc40 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -17,6 +17,7 @@ from modules import devices, prompt_parser, masking, sd_samplers, lowvram, gener from modules.sd_hijack import model_hijack from modules.shared import opts, cmd_opts, state import modules.shared as shared +import modules.paths as paths import modules.face_restoration import modules.images as images import modules.styles @@ -584,7 +585,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if not p.disable_extra_networks: extra_networks.activate(p, extra_network_data) - with open(os.path.join(shared.script_path, "params.txt"), "w", encoding="utf8") as file: + with open(os.path.join(paths.data_path, "params.txt"), "w", encoding="utf8") as file: processed = Processed(p, [], p.seed, "") file.write(processed.infotext(p, 0)) diff --git a/modules/sd_models.py b/modules/sd_models.py index 37dad18d..b2d48a51 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -12,13 +12,13 @@ import ldm.modules.midas as midas from ldm.util import instantiate_from_config -from modules import shared, modelloader, devices, script_callbacks, sd_vae, sd_disable_initialization, errors, hashes, sd_models_config +from modules import paths, shared, modelloader, devices, script_callbacks, sd_vae, sd_disable_initialization, errors, hashes, sd_models_config from modules.paths import models_path from modules.sd_hijack_inpainting import do_inpainting_hijack from modules.timer import Timer model_dir = "Stable-diffusion" -model_path = os.path.abspath(os.path.join(models_path, model_dir)) +model_path = os.path.abspath(os.path.join(paths.models_path, model_dir)) checkpoints_list = {} checkpoint_alisases = {} @@ -307,7 +307,7 @@ def enable_midas_autodownload(): location automatically. """ - midas_path = os.path.join(models_path, 'midas') + midas_path = os.path.join(paths.models_path, 'midas') # stable-diffusion-stability-ai hard-codes the midas model path to # a location that differs from where other scripts using this model look. diff --git a/modules/sd_vae.py b/modules/sd_vae.py index 4ce238b8..9b00f76e 100644 --- a/modules/sd_vae.py +++ b/modules/sd_vae.py @@ -3,13 +3,12 @@ import safetensors.torch import os import collections from collections import namedtuple -from modules import shared, devices, script_callbacks, sd_models -from modules.paths import models_path +from modules import paths, shared, devices, script_callbacks, sd_models import glob from copy import deepcopy -vae_path = os.path.abspath(os.path.join(models_path, "VAE")) +vae_path = os.path.abspath(os.path.join(paths.models_path, "VAE")) vae_ignore_keys = {"model_ema.decay", "model_ema.num_updates"} vae_dict = {} diff --git a/modules/shared.py b/modules/shared.py index 14be993d..474fcc42 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -14,7 +14,7 @@ import modules.memmon import modules.styles import modules.devices as devices from modules import localization, extensions, script_loading, errors, ui_components, shared_items -from modules.paths import models_path, script_path +from modules.paths import models_path, script_path, data_path demo = None @@ -25,6 +25,7 @@ sd_model_file = os.path.join(script_path, 'model.ckpt') default_sd_model_file = sd_model_file parser = argparse.ArgumentParser() +parser.add_argument("--data-dir", type=str, default=os.path.dirname(os.path.dirname(os.path.realpath(__file__))), help="base path where all user data is stored",) parser.add_argument("--config", type=str, default=sd_default_config, help="path to config which constructs model",) parser.add_argument("--ckpt", type=str, default=sd_model_file, help="path to checkpoint of stable diffusion model; if specified, this checkpoint will be added to the list of checkpoints and loaded",) parser.add_argument("--ckpt-dir", type=str, default=None, help="Path to directory with stable diffusion checkpoints") @@ -35,7 +36,7 @@ parser.add_argument("--no-half", action='store_true', help="do not switch the mo parser.add_argument("--no-half-vae", action='store_true', help="do not switch the VAE model to 16-bit floats") parser.add_argument("--no-progressbar-hiding", action='store_true', help="do not hide progressbar in gradio UI (we hide it because it slows down ML if you have hardware acceleration in browser)") parser.add_argument("--max-batch-count", type=int, default=16, help="maximum batch count value for the UI") -parser.add_argument("--embeddings-dir", type=str, default=os.path.join(script_path, 'embeddings'), help="embeddings directory for textual inversion (default: embeddings)") +parser.add_argument("--embeddings-dir", type=str, default=os.path.join(data_path, 'embeddings'), help="embeddings directory for textual inversion (default: embeddings)") parser.add_argument("--textual-inversion-templates-dir", type=str, default=os.path.join(script_path, 'textual_inversion_templates'), help="directory with textual inversion templates") parser.add_argument("--hypernetwork-dir", type=str, default=os.path.join(models_path, 'hypernetworks'), help="hypernetwork directory") parser.add_argument("--localizations-dir", type=str, default=os.path.join(script_path, 'localizations'), help="localizations directory") @@ -74,16 +75,16 @@ parser.add_argument("--use-cpu", nargs='+', help="use CPU as torch device for sp parser.add_argument("--listen", action='store_true', help="launch gradio with 0.0.0.0 as server name, allowing to respond to network requests") parser.add_argument("--port", type=int, help="launch gradio with given server port, you need root/admin rights for ports < 1024, defaults to 7860 if available", default=None) parser.add_argument("--show-negative-prompt", action='store_true', help="does not do anything", default=False) -parser.add_argument("--ui-config-file", type=str, help="filename to use for ui configuration", default=os.path.join(script_path, 'ui-config.json')) +parser.add_argument("--ui-config-file", type=str, help="filename to use for ui configuration", default=os.path.join(data_path, 'ui-config.json')) parser.add_argument("--hide-ui-dir-config", action='store_true', help="hide directory configuration from webui", default=False) parser.add_argument("--freeze-settings", action='store_true', help="disable editing settings", default=False) -parser.add_argument("--ui-settings-file", type=str, help="filename to use for ui settings", default=os.path.join(script_path, 'config.json')) +parser.add_argument("--ui-settings-file", type=str, help="filename to use for ui settings", default=os.path.join(data_path, 'config.json')) parser.add_argument("--gradio-debug", action='store_true', help="launch gradio with --debug option") parser.add_argument("--gradio-auth", type=str, help='set gradio authentication like "username:password"; or comma-delimit multiple like "u1:p1,u2:p2,u3:p3"', default=None) parser.add_argument("--gradio-img2img-tool", type=str, help='does not do anything') parser.add_argument("--gradio-inpaint-tool", type=str, help="does not do anything") parser.add_argument("--opt-channelslast", action='store_true', help="change memory type for stable diffusion to channels last") -parser.add_argument("--styles-file", type=str, help="filename to use for styles", default=os.path.join(script_path, 'styles.csv')) +parser.add_argument("--styles-file", type=str, help="filename to use for styles", default=os.path.join(data_path, 'styles.csv')) parser.add_argument("--autolaunch", action='store_true', help="open the webui URL in the system's default browser upon launch", default=False) parser.add_argument("--theme", type=str, help="launches the UI with light or dark theme", default=None) parser.add_argument("--use-textbox-seed", action='store_true', help="use textbox for seeds in UI (no up/down, but possible to input long seeds)", default=False) diff --git a/modules/textual_inversion/preprocess.py b/modules/textual_inversion/preprocess.py index c0ac11d3..2239cb84 100644 --- a/modules/textual_inversion/preprocess.py +++ b/modules/textual_inversion/preprocess.py @@ -6,8 +6,7 @@ import sys import tqdm import time -from modules import shared, images, deepbooru -from modules.paths import models_path +from modules import paths, shared, images, deepbooru from modules.shared import opts, cmd_opts from modules.textual_inversion import autocrop @@ -199,7 +198,7 @@ def preprocess_work(process_src, process_dst, process_width, process_height, pre dnn_model_path = None try: - dnn_model_path = autocrop.download_and_cache_models(os.path.join(models_path, "opencv")) + dnn_model_path = autocrop.download_and_cache_models(os.path.join(paths.models_path, "opencv")) except Exception as e: print("Unable to load face detection model for auto crop selection. Falling back to lower quality haar method.", e) diff --git a/modules/ui.py b/modules/ui.py index 85ae62c7..0117df3e 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -21,7 +21,7 @@ from modules.call_queue import wrap_gradio_gpu_call, wrap_queued_call, wrap_grad from modules import sd_hijack, sd_models, localization, script_callbacks, ui_extensions, deepbooru, sd_vae, extra_networks, postprocessing, ui_components, ui_common, ui_postprocessing from modules.ui_components import FormRow, FormGroup, ToolButton, FormHTML -from modules.paths import script_path +from modules.paths import script_path, data_path from modules.shared import opts, cmd_opts, restricted_opts @@ -1497,8 +1497,8 @@ def create_ui(): with open(cssfile, "r", encoding="utf8") as file: css += file.read() + "\n" - if os.path.exists(os.path.join(script_path, "user.css")): - with open(os.path.join(script_path, "user.css"), "r", encoding="utf8") as file: + if os.path.exists(os.path.join(data_path, "user.css")): + with open(os.path.join(data_path, "user.css"), "r", encoding="utf8") as file: css += file.read() + "\n" if not cmd_opts.no_progressbar_hiding: diff --git a/modules/ui_extensions.py b/modules/ui_extensions.py index 742e745e..66a41865 100644 --- a/modules/ui_extensions.py +++ b/modules/ui_extensions.py @@ -132,7 +132,7 @@ def install_extension_from_url(dirname, url): normalized_url = normalize_git_url(url) assert len([x for x in extensions.extensions if normalize_git_url(x.remote) == normalized_url]) == 0, 'Extension with this URL is already installed' - tmpdir = os.path.join(paths.script_path, "tmp", dirname) + tmpdir = os.path.join(paths.data_path, "tmp", dirname) try: shutil.rmtree(tmpdir, True) diff --git a/modules/upscaler.py b/modules/upscaler.py index a5bf5acb..e2eaa730 100644 --- a/modules/upscaler.py +++ b/modules/upscaler.py @@ -11,7 +11,6 @@ from modules import modelloader, shared LANCZOS = (Image.Resampling.LANCZOS if hasattr(Image, 'Resampling') else Image.LANCZOS) NEAREST = (Image.Resampling.NEAREST if hasattr(Image, 'Resampling') else Image.NEAREST) -from modules.paths import models_path class Upscaler: @@ -39,7 +38,7 @@ class Upscaler: self.mod_scale = None if self.model_path is None and self.name: - self.model_path = os.path.join(models_path, self.name) + self.model_path = os.path.join(shared.models_path, self.name) if self.model_path and create_dirs: os.makedirs(self.model_path, exist_ok=True) @@ -143,4 +142,4 @@ class UpscalerNearest(Upscaler): def __init__(self, dirname=None): super().__init__(False) self.name = "Nearest" - self.scalers = [UpscalerData("Nearest", None, self)] \ No newline at end of file + self.scalers = [UpscalerData("Nearest", None, self)] -- cgit v1.2.3 From ada17dbd7c4c68a4e559848d2e6f2a7799722806 Mon Sep 17 00:00:00 2001 From: brkirch Date: Fri, 27 Jan 2023 10:19:43 -0500 Subject: Refactor conditional casting, fix upscalers --- modules/devices.py | 8 ++++++++ modules/processing.py | 15 ++++++++------- modules/realesrgan_model.py | 2 +- modules/sd_hijack.py | 2 +- modules/sd_hijack_unet.py | 8 +++++++- 5 files changed, 25 insertions(+), 10 deletions(-) (limited to 'modules/processing.py') diff --git a/modules/devices.py b/modules/devices.py index 6b36622c..0100e4af 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -83,6 +83,14 @@ dtype_unet = torch.float16 unet_needs_upcast = False +def cond_cast_unet(input): + return input.to(dtype_unet) if unet_needs_upcast else input + + +def cond_cast_float(input): + return input.float() if unet_needs_upcast else input + + def randn(seed, shape): torch.manual_seed(seed) if device.type == 'mps': diff --git a/modules/processing.py b/modules/processing.py index 92894d67..a397702b 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -172,8 +172,7 @@ class StableDiffusionProcessing: midas_in = torch.from_numpy(transformed["midas_in"][None, ...]).to(device=shared.device) midas_in = repeat(midas_in, "1 ... -> n ...", n=self.batch_size) - conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(source_image.to(devices.dtype_vae) if devices.unet_needs_upcast else source_image)) - conditioning_image = conditioning_image.float() if devices.unet_needs_upcast else conditioning_image + conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(source_image)) conditioning = torch.nn.functional.interpolate( self.sd_model.depth_model(midas_in), size=conditioning_image.shape[2:], @@ -217,7 +216,7 @@ class StableDiffusionProcessing: ) # Encode the new masked image using first stage of network. - conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(conditioning_image.to(devices.dtype_vae) if devices.unet_needs_upcast else conditioning_image)) + conditioning_image = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(conditioning_image)) # Create the concatenated conditioning tensor to be fed to `c_concat` conditioning_mask = torch.nn.functional.interpolate(conditioning_mask, size=latent_image.shape[-2:]) @@ -228,16 +227,18 @@ class StableDiffusionProcessing: return image_conditioning def img2img_image_conditioning(self, source_image, latent_image, image_mask=None): + source_image = devices.cond_cast_float(source_image) + # HACK: Using introspection as the Depth2Image model doesn't appear to uniquely # identify itself with a field common to all models. The conditioning_key is also hybrid. if isinstance(self.sd_model, LatentDepth2ImageDiffusion): - return self.depth2img_image_conditioning(source_image.float() if devices.unet_needs_upcast else source_image) + return self.depth2img_image_conditioning(source_image) if self.sd_model.cond_stage_key == "edit": return self.edit_image_conditioning(source_image) if self.sampler.conditioning_key in {'hybrid', 'concat'}: - return self.inpainting_image_conditioning(source_image.float() if devices.unet_needs_upcast else source_image, latent_image, image_mask=image_mask) + return self.inpainting_image_conditioning(source_image, latent_image, image_mask=image_mask) # Dummy zero conditioning if we're not using inpainting or depth model. return latent_image.new_zeros(latent_image.shape[0], 5, 1, 1) @@ -417,7 +418,7 @@ def create_random_tensors(shape, seeds, subseeds=None, subseed_strength=0.0, see def decode_first_stage(model, x): with devices.autocast(disable=x.dtype == devices.dtype_vae): - x = model.decode_first_stage(x.to(devices.dtype_vae) if devices.unet_needs_upcast else x) + x = model.decode_first_stage(x) return x @@ -1001,7 +1002,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image = torch.from_numpy(batch_images) image = 2. * image - 1. - image = image.to(device=shared.device, dtype=devices.dtype_vae if devices.unet_needs_upcast else None) + image = image.to(shared.device) self.init_latent = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(image)) diff --git a/modules/realesrgan_model.py b/modules/realesrgan_model.py index 47f70251..aad4a629 100644 --- a/modules/realesrgan_model.py +++ b/modules/realesrgan_model.py @@ -46,7 +46,7 @@ class UpscalerRealESRGAN(Upscaler): scale=info.scale, model_path=info.local_data_path, model=info.model(), - half=not cmd_opts.no_half, + half=not cmd_opts.no_half and not cmd_opts.upcast_sampling, tile=opts.ESRGAN_tile, tile_pad=opts.ESRGAN_tile_overlap, ) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 531790f3..8fc91882 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -171,7 +171,7 @@ class EmbeddingsWithFixes(torch.nn.Module): vecs = [] for fixes, tensor in zip(batch_fixes, inputs_embeds): for offset, embedding in fixes: - emb = embedding.vec.to(devices.dtype_unet) if devices.unet_needs_upcast else embedding.vec + emb = devices.cond_cast_unet(embedding.vec) emb_len = min(tensor.shape[0] - offset - 1, emb.shape[0]) tensor = torch.cat([tensor[0:offset + 1], emb[0:emb_len], tensor[offset + 1 + emb_len:]]) diff --git a/modules/sd_hijack_unet.py b/modules/sd_hijack_unet.py index a6ee577c..45cf2b18 100644 --- a/modules/sd_hijack_unet.py +++ b/modules/sd_hijack_unet.py @@ -55,8 +55,14 @@ class GELUHijack(torch.nn.GELU, torch.nn.Module): unet_needs_upcast = lambda *args, **kwargs: devices.unet_needs_upcast CondFunc('ldm.models.diffusion.ddpm.LatentDiffusion.apply_model', apply_model, unet_needs_upcast) -CondFunc('ldm.modules.diffusionmodules.openaimodel.timestep_embedding', lambda orig_func, *args, **kwargs: orig_func(*args, **kwargs).to(devices.dtype_unet), unet_needs_upcast) +CondFunc('ldm.modules.diffusionmodules.openaimodel.timestep_embedding', lambda orig_func, timesteps, *args, **kwargs: orig_func(timesteps, *args, **kwargs).to(torch.float32 if timesteps.dtype == torch.int64 else devices.dtype_unet), unet_needs_upcast) if version.parse(torch.__version__) <= version.parse("1.13.1"): CondFunc('ldm.modules.diffusionmodules.util.GroupNorm32.forward', lambda orig_func, self, *args, **kwargs: orig_func(self.float(), *args, **kwargs), unet_needs_upcast) CondFunc('ldm.modules.attention.GEGLU.forward', lambda orig_func, self, x: orig_func(self.float(), x.float()).to(devices.dtype_unet), unet_needs_upcast) CondFunc('open_clip.transformer.ResidualAttentionBlock.__init__', lambda orig_func, *args, **kwargs: kwargs.update({'act_layer': GELUHijack}) and False or orig_func(*args, **kwargs), lambda _, *args, **kwargs: kwargs.get('act_layer') is None or kwargs['act_layer'] == torch.nn.GELU) + +first_stage_cond = lambda _, self, *args, **kwargs: devices.unet_needs_upcast and self.model.diffusion_model.dtype == torch.float16 +first_stage_sub = lambda orig_func, self, x, **kwargs: orig_func(self, x.to(devices.dtype_vae), **kwargs) +CondFunc('ldm.models.diffusion.ddpm.LatentDiffusion.decode_first_stage', first_stage_sub, first_stage_cond) +CondFunc('ldm.models.diffusion.ddpm.LatentDiffusion.encode_first_stage', first_stage_sub, first_stage_cond) +CondFunc('ldm.models.diffusion.ddpm.LatentDiffusion.get_first_stage_encoding', lambda orig_func, *args, **kwargs: orig_func(*args, **kwargs).float(), first_stage_cond) -- cgit v1.2.3 From 00dab8f10defbbda579a1bc89c8d4e972c58a20d Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Sun, 29 Jan 2023 11:53:24 +0300 Subject: remove Batch size and Batch pos from textinfo (goodbye) --- modules/processing.py | 2 -- 1 file changed, 2 deletions(-) (limited to 'modules/processing.py') diff --git a/modules/processing.py b/modules/processing.py index afab6790..2d295932 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -450,8 +450,6 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments=None, iter "Size": f"{p.width}x{p.height}", "Model hash": getattr(p, 'sd_model_hash', None if not opts.add_model_hash_to_info or not shared.sd_model.sd_model_hash else shared.sd_model.sd_model_hash), "Model": (None if not opts.add_model_name_to_info or not shared.sd_model.sd_checkpoint_info.model_name else shared.sd_model.sd_checkpoint_info.model_name.replace(',', '').replace(':', '')), - "Batch size": (None if p.batch_size < 2 else p.batch_size), - "Batch pos": (None if p.batch_size < 2 else position_in_batch), "Variation seed": (None if p.subseed_strength == 0 else all_subseeds[index]), "Variation seed strength": (None if p.subseed_strength == 0 else p.subseed_strength), "Seed resize from": (None if p.seed_resize_from_w == 0 or p.seed_resize_from_h == 0 else f"{p.seed_resize_from_w}x{p.seed_resize_from_h}"), -- cgit v1.2.3